Steve H. Murdock, David R. Ellis - Applied Demography_ An Introduction to Basic Concepts, Methods, and Data-Routledge (2020).pdf

Applied Demography
An Introduction to Basic Concepts, Methods, and Data
Steve H. Murdock
David R. Ellis
www.routledge.com an informa business
ISBN 978-0-367-01259-5
Applied
Demography
Steve
H.
Murdock
and
David
R.
Ellis
9780367012595.indd 1 10/21/2018 2:26:07 PM

Applied Demography
An Introduction to
Basic Concepts, Methods, and Data
Steve H. Murdock
and David R. Ellis
~l Routledge
::S~ TaylorFram Croup
AND TORK

Library of Congress Cataloging-in-Publication Data
Murdock, Steven H.
Applied demography : an introduction to basic concepts, methods,
and data / by Steve H. Murdock and David R. Ellis.
p. cm. ·
Includes bibliographical references and index.
ISBN 0-8133-8372-2
1. Demography. I. Ellis, David R. (David Rennie), 1953-
11. Title.
HB849.4.M87 1991
304.6-dc20 91-35208
CIP
ISBN 13: 978-0-367-01259-5 (hbk)
First published 1991 by Westview Press
Published 2018 by Routledge
52 Vanderbilt Avenue, New York, NY 10017
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or
by any electronic, mechanical, or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or retrieval system, without permission
in writing from the publishers.
Notice:
Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.
Copyright © 1991 by Taylor Francis
Routledge is an imprint of the Taylor Francis Group, an informa business

To Joann and Roger
June and Lee
Marijane and Bo

Contents
Ust of Tables and Figures
Preface
Acknowledgments
1
2
Introduction
Rationale and Background, 1
Definition and the Dimensions of
Applied Demography, 3
Organization of the Text, 8
Limitations of the Book, 9
Demographic Concepts and Trends: The Conceptual
Base and Recent Patterns of Demographic Change
Defining Key Concepts and Terms, 11
An Overview of Major Demographic
Trends in the United States, 26
Summary, 66
Conclusions, 67
3 The Materials of Appliecl Demographic Analyses:
4
Data Sources and Principles of Data Use
Indices for Locating Secondary Data, 70
Federal and State Data Compilations, 75
Federal Data Sources, 79
State Data Sources, 100
Nongovernmental Data Sources, 103
Using Secondary Data, 105
Summary and Conclusions, 112
Basic Methods and Measures of
Applied Demography
General Measures, 113
Measures of the Major Demographic
Processes and Variables, 120
Selected Methods for Controlling the Effects
of Demographic Change and Characteristics, 156
Conclusions, 174
ix
xv
xix
1
11
69
113

viii
5
6
Methods for Estimating and Projecting
Populations
Basic Definitions and Concepts, Principles and
Limitations, and General Procedures for Use
in Population Estimation and Projection, 176
Methods of Population Estimation, 181
Methods of Population Projection, 210
Estimates and Projections of Population-Based
Statuses and Characteristics, 234
Evaluation of Population Estimates and
Projections, 241
Conclusions, 248
Summary and Conclusions: The Future of
Population Change and Applied Demography
in the United States
Future Demographic Trends Impacting Products
and Services, 250
The Future of Applied Demography, 265
Conclusions, 273
References
Index
175
249
275
289

Tables and Figures
Tables
2.1 Total Resident Population and Percent Population
Change in the United States, 1790-1990 27
2.2 Components of Population Change for the United
States, 1940-1990 29
2.3 Birth, Death, and Net Migration Measures for
the United States, 1940-1990 30
2.4 Age-Specific Birth, Death, and Migration
Rates in the United States for Selected Years 31
2.5 Population of the United States, Regions,
Divisions, and States, 1900-1990 33
2.6 Population Change in the United States,
Regions and Divisions, 1960-1990 37
2.7 Population and Percentage of Population in the
United States by Urban, Rural, Rural Farm, and
Rural Nonfarm Residence, 1930-1980 39
2.8 Proportion of U.S. Population that Is
Metropolitan and Nonmetropolitan, 1950-1990 40
2.9 Median Age and the Sex Ratio in the United
States, 1900-1990 41
2.10 Population of the United States by Age and Sex,
1940-1989 42
2.11 Percent of the Population by Age Groups in the
United States, 1940-1989 45
2.12 U.S. Population, 1970, 1980, and 1990, Percent
Change in Population 1970 to 1980 and
1980 to 1990, and Proportion of Population
1970, 1980, and 1990 by Race, Hispanic
Origin, and Ethnicity 47
2.13 Percent Distribution of the Resident Population
of the United States by Regions and for the
Ten Largest States by Race and Hispanic
Origin, 1990 48
2.14 Percent Distribution of the Resident Population
of the United States, Regions and States
by Race and Hispanic Origin, 1990 49

x
2.15 Demographic and Socioeconomic Characteristics
of the Population of the United States by
Race/Ethnicity for Selected Years 53
2.16 Marital Status of the Population of the United
States, 1970-1988 56
2.17 Households in the United States by Type,
1970-1990 57
2.18 Estimates of Cohabitation and Marriage Before
the Age of 25 by Age Cohort in 1988 58
2.19 Number and Percent of Households by Persons
in the Household and Average Household
Size for the United States, 1940-1990 59
2.20 Selected Socioeconomic Characteristics of
the Population of the United States,
1940-1988 61
4.1 A Decomposition of the Projected Difference
in the Rate of Participation in Different
Recreational Activities Among Residents of
the United States by Activity, 1990-2000 and
2000-2025 165
4.2 A Decomposition of the Projected Difference
in the Rates of Participation in Different
Recreational Activities Among Residents of
Texas by Activity, 1990-2000 and 2000-2025 166
4.3 Components of a Working Life Table Derived
Using a Standard Life Table 173
6.1 Historical and Projected Population Growth
in the United States by Race and Spanish
Origin, 1950-2050 252
6.2 Percent of Population by Race and Spanish
Origin in the United States, 1950-2050 253
6.3 Projections of the Percent of the U.S.
Population by Age and Race/Ethnicity for
Selected Years, 1990-2050 254
6.4 Three Alternative Projections of the U.S.
Civilian Labor Force by Selected
Characteristics for 2000 256

xi
6.5 Projections of the Number of Persons in the
Labor Force in the United States by
Race/Ethnicity, 1986-2025 257
6.6 Projections of the Number of Residents
Enrolled in Higher Education in the
United States by Race/Ethnicity, 1986-
2025 259
6.7 Median U.S. Household Income in 1989 by
Selected Characteristics 264
Figures
3.1 Short-Form (100% Items) and Long-Form
(Sample Items) Topics in the 1990 Census
of Population and Housing 83
3.2 Publications of the 1990 Census of
Population and Housing 84
3.3 Computerized Products from the 1990 Census 88
4.1 Percentage Change in Population 114
4.2 Crude Rates 116
4.3 General Rates 118
4.4 Specific Rates 119
4.5 Arithmetic Rate of Change 121
4.6 Geometric Rate of Change 122
4.7 Exponential Rate of Change 123
4.8 Child-Woman Ratio (CWR) 125
4.9 Total Fertility Rate (TFR) · 127
4.10 Selected Measures of Infant Mortality 128
4.11 Abridged Life Table for the Male Population
of a Hypothetical Area, 1990 130
4.12 Elements of a Life Table 131
4.13 Life Table Survival Rates 135
4.14 Procedure for Computing Survival Rates for
Multi-Age Age Groups from a Life Table
for Single-Year Age Groups 136
4.15 Procedure for Computing Beginning and
Terminal Age Survival Rates 137
4.16 Migration Rates 139
4.17 Net Migration Rate (NMR) 139
4.18 Residual Migration 140

xii
4.19 Population Density 140
4.20 Population Potential Measure with an Example
of Its Application for a Hypothetical Set
of Areas 143
4.21 Distribution of a Hypothetical Population
by Size of Place Category and the Related
Lorenz Curve 144
4.22 The Cini Coefficient and Index of
Dissimilarity Measures of Population
Distribution 145
4.23 Dependency Ratio (DR) 150
4.24 The Sex Ratio (SR) 150
4.25 Population Pyramid, Texas 151
4.26 Crude, General, and Age-Specific Marriage
Rates 154
4.27 Measures of Educational Progression 155
4.28 Measures of Economic Activity 157
4.29 Direct and Indirect Age Standardization 160
4.30 Unique Components of Nuptiality Tables,
Tables of School Life, and Tables of
Working Life 170
4.31 Example of Using a Table of Working Life
to Determine Income Loss 172
5.1 Example of Controlling to a Total 180
5.2 Projections for a College-Dominated County
by Age for 1980-2020 NOT Adjusting for
Special Populations 180
5.3 Censal-Ratio Method with Symptomatic Data 185
5.4 Censal-Ratio Procedure with Housing Permit
Data: To Estimate the Austin, Texas,
Population for April 1, 1984 188
5.5 Censal-Ratio Method Using Electric Meter
Billing: To Estimate the Austin, Texas,
Population for April 1, 1988 190
5.6 Example of a Simple Ratio Technique 192
5.7 Vital Rates Method 193
5.8 Example of the Use of a Proration Technique 194
5.9 Composite Method 195
5.10 Steps for Completing an Estimate Using the
Ratio-Correlation Method 199

xiii
5.11 Ratio-Correlation Method: To Estimate
Population of Waco, Texas, 1982 200
5.12 Steps in and Example of the Use of a Cohort-
Survival Method of Population Estimation to
Estimate the Population of McLennan County,
Texas, April 1, 1988 206
5.13 Example of a Ratio-Based Technique 215
5.14 Example of a Land-Use Technique 216
5.15 Hypothetical Example of a Simple Economic-
Based Population Projection Method 221
5.16 Steps in and Example of the Use of the Cohort-
Component Method to Project the Population of
Harris County, Texas, by Five-Year Cohorts
from 1990 to 2000, Assuming 1980
Age-Sex Specific Fertility Rates and
Age-Sex Specific Survival Rates and
1970-1980 Age-Specific Net Migration Rates 235
5.17 Example of the Use of Three Commonly Used
Error Measures 246

Preface
For more than 15 years I have worked with local and state
planners and analysts and private-sector marketing and planning
specialists, attempting to share with them knowledge of demograph-
ic. concepts, data bases, and methods for addressing pragmatic
issues. At the same time, I have been involved with numerous
professional demographers in gaining recognition of the needs of
decisionmakers and the role of demographic data in the decision-
making process. I have also taught both social demography and
basic demographic methods courses to a diverse set of students from
such disciplines as sociology, psychology, political science, urban
and regional planning, history, anthropology, real estate develop-
ment, recreation and parks, and numerous other disciplines. All of
these activities have convinced me that demographic knowledge is
not only required in many different forms of analysis, but that much
of the existing demographic literature is too specialized for the
applied analyst who must examine a diverse range of phenomena,
only some of which are demographic.
The second author has likewise worked with private- and public-
sector decisionmakers for more than a decade. This experience,
coupled with his return to graduate school and his enrollment in
several classes in demography, convinced him that demography had
much to offer the policy analyst. At the same time, most works on
demography were either too specialized to meet the needs of ap-
plied analysts or attempted to provide broad overviews of interna-
tional population patterns that, although informative, were likely to
be of little direct utility to policy analysts.
This work reflects our belief that a single-source document is
necessary that can both introduce someone with only a basic social
science background to the concepts, data, and methods of applied
demography and can offer insight to professional demographers
regarding the specific methods and issues likely to be required of
them in pursuing applied demographic problems.
This work also represents our attempt to, at least partially, ad-
dress the need of the emerging area of applied demography for texts
that attempt to define its subject matter, its data, and its methods.
It represents an attempt to contribute to the development of what
we believe will be an increasingly important area of analysis in the
coming decades.

xvi
Finally, this work represents an effort aimed at drawing together
in a single source works that we have developed over 15 years in
the course of attempting to meet the needs of those who do demo-
graphic analyses. We have compiled numerous sets of workshop
materials and related workbooks and manuals on such topics as
small-area population estimates and projections, basic demographic
methods, and sources of information for business and government.
These materials, although clearly not sufficient to form the total
basis for this worlc, made it evident that a single, synthesized work
that was concisely focused on the concepts, methods, and materials
of greatest utility in completing applied analyses was needed and
likely to be of utility to applied analysts.
To address these concerns, we have developed a work that we
hope provides a basic introduction to the subject matter and meth-
ods of demography as applied to pragmatic issues and that is useful
to professional demographers who need more detailed information
on the areas of analyses likely to be of most importance in applied
uses of demography. Thus, the first two chapters introduce the
reader to demography and applied demography and provide a base
of knowledge about basic demographic concepts and current demo-
graphic trends. Chapter 3 introduces the data sources most often
used in demography. Although many of the data sources discussed
are widely known, we believe they are sufficiently detailed that
even professional demographers will benefit from it. Chapter 4
presents an introduction to the methods of applied demography that
provides essential background knowledge for those new to the
subject and examples of the applied uses of demographic data and
methods that introduce the professional demographer to substantive
issues addressed by applied analysts. Chapter 5 provides a detailed
discussion of methods of population estimation and projection and
of the evaluation of estimates and projections. These are among the
tasks most frequently required of applied demographers, and their
applications to small areas is seldom sufficiently covered in standard
demography curricula. Finally, Chapter 6 examines the problems
and opportunities likely to emerge from future changes in the
population and in applied demography in the United States in the
coming decades.
The work is intended to be useful to those with a basic educa-
tion in a social science or related discipline and requires no mathe-
matical skills beyond basic algebra. It will serve as a useful text for
multidisciplinary upper-level undergraduate and beginning-level
graduate courses in applied demography. It should also be a useful

xvii
reference source for the libraries of those who do applied demo-
graphic analysis in business, government, and academia.
Anyone attempting such a work is painfully aware that space
and other limitations prevent its being as comprehensive as one
would like. Likewise, it is not possible for this work to provide
sufficiently thorough discussions of several complex procedures to
allow its readers to employ such methods without the use of addi-
tional references. We have described such methods and demon-
strated them sufficiently to allow users to both know where to
obtain information necessary to apply these methods and the types
of uses to which these procedures may be appropriately applied.
Although the work has limitations, we hope that it proves bene-
ficial to its intended audiences in gaining basic knowledge of the
applied uses of demographic concepts, data, and methods. We trust
that it will soon be followed by other works providing additional,
and increasingly sophisticated, assistance to those who use demog-
raphy to address pragmatic issues. Even more important, we hope
that the work assists readers to more effectively use applied demo-
graphic concepts, data, and methods to arrive at solutions to real-
world problems.
Steve H. Murdock

Acknowledgments
In the completion of this work, the support, assistance, and
encouragement of numerous persons and agencies must be acknowl-
edged. The Department of Rural Sociology and the Texas Agricul-
tural Experiment Station in the Texas A:M University System
provided financial support for this effort and receive our sincere
appreciation. We wish also to thank the Real Estate Center at Texas
A:M University, especially its director, Dr. Richard Floyd. The
support of the center for the authors has been essential to the
completion of the work and to our gaining sensitivity to the needs
of a major segment of data users. We also extend our appreciation
to the Texas State Data Center and Texas Population Estimates and
Projections Programs and to the coordinating agency for these
programs, the Texas Department of Commerce, for allowing us to be
involved in these programs and to thus gain insight into the needs
of some of those persons most likely to use this work.
In the preparation of the book, numerous people have provided
assistance in preparing examples, in manuscript preparation, and in
providing critical reviews of the volume. Those who have assisted
in the development of initial examples for the works on which this
volume is partially based include Sean-Shong Hwang, Banoo Parpia,
John DeMontel, Pam Hopkins, Ken Backman, and Martha Nelson.
We thank them, even if belatedly. Recent students who have given
of their time and deserve our appreciation include Gavin Smith,
Rickie Fletcher, Jaime Vinas, Alvin Luedke, Marie Ballejos, Erik
Koehlert, and Paul Johnston. We also thank several staff members
including Beverly Pecotte, Darrell Fannin, Md. Nazrul Hoque,
George Galdiano, and Stephanie Rogers for their tireless efforts in
preparing data, proofreading, and copying the work for various
purposes. We owe special appreciation to Delma Jones and Teresa
Ray who tirelessly typed repeated drafts of the work and to Edwin
Gene and Elizabeth Porter whose expertise was essential to finishing
the work. We owe our most sincere thanks to Patricia Bramwell,
who was instrumental in the completion of every phase of the work
and who cheerfully tolerated the cranky authors during the final
, phases of the work. The work clearly would not have been com-
pleted without her extraordinary efforts in organizing and directly
participating in nearly all aspects of the work.
Special appreciation is also due to a former colleague who made
major contributions to all of the earlier works from which parts of
this work are drawn. This is Rita R. Leistritz. Her encouragement

xx
to undertake the works from which this is drawn and her tireless
efforts in developing countless examples cannot be adequately
acknowledged. Thank you, Rita, for your decade of effort.
We also wish to thank Donna Nunez who tirelessly edited the
work, repairing the authors' damaged grammar and punctuation
and providing consistency for two people who seem to thrive on
inconsistency. Thank you, Donna, for your efforts~
We owe particular appreciation to our reviewers who reviewed
the entire document and gave us useful and constructive criticisms.
These include Ken Backman, Stan Drezek, Tom Hirschi, Dan Lich-
ter, Rogelio Saenz, and Paul Voss. To each of them, we extend our
sincere appreciation for assisting us in making this a better work.
Finally, we extend our thanks to our colleagues, staff, friends,
and families who endured our impatience and our neglect of other
activities during the completion of the work.
S.H.M.

1
Introduction
Rationale and Background
Demography has been popularized as it has become evident that
demographic characteristics and trends impact many aspects of our
society. Population change and the characteristics of the population
have effects on a wide range of factors, including markets for private
goods and services (Pol, 1987), forms of urban and regional growth
(Berry and Kasarda, 1977), the potential for economic development
(Backman, 1989), the likely incidence of disease and mortality
(Murdock et al., 1989a), and political redistricting and voting pat-
terns (Hill and Kent, 1988). Population patterns affect levels of
economic resources and poverty (Macunovich and Easterlin, 1990),
incidences of crime (Cohen and Felson, 1979; Stahura and Sloan,
1988), characteristics of the labor force (U.S. Bureau of Labor Statis-
tics, 1989), changes in enrollments in elementary and secondary
schools and in higher education (National Center for Educational
Statistics, 1989), changes in housing and real estate patterns (Stern-
lieb and Hughes, 1986; Murdock and Hamm, 1988a), and numerous
other factors (Russell, 1984; Merrick and Tordella, 1988). Demogra-
phy is important to those involved in product and service market-
ing, strategic and corporate planning, urban and regional analyses,
real estate development, economic development, medical and health
care, political analysis, financial analysis, crime prevention, person-
nel and human resource development, education, and many other
fields.
It is not the population patterns and trends themselves that are
the focus of attention for such persons, however, but the implica-
tions of these trends for nondemographic factors and events.
Applied demography thus focuses on pragmatic concerns of interest
to professionals whose training and experience lie largely outside
the small community of professional demographers.

2
In fact, recognition of the importance of demographics is so
pervasive that nearly all professionals involved in private- or public-
sector marketing and planning use demographic data and perform
demographic analysis. Many have been forced to gain knowledge of
demographic processes and concepts, learn how to obtain and
manipulate demographic data, and master demographic analysis
techniques. These professionals often find themselves needing to
locate information to profile the current characteristics of the popula-
tion of alternative market or service areas; estimate the current and
project future populations likely to effect the demand for goods and
services; and to identify and quantify the effects of age, race/ethnici-
ty, household composition, and other factors on the use of goods
and services. Even when they are not directly responsible for the
development of demographic data and analyses (because the data
are purchased from private data provision firms), these analysts are
usually responsible for ensuring that the data and analyses are
appropriate. Such analysts must obtain knowledge of the demo-
graphic concepts, data sources, and the techniques underlying the
data and analyses that have been purchased.
Unfortunately, these professionals often find it difficult to obtain
the knowledge required to complete such tasks, because it is scat-
tered among a number of courses offered in formal demographic
training programs in academic settings or is available in a growing
but widely scattered set of materials in applied demography (Rives
and Serow, 1984; Pol, 1987; Saunders, 1988; Merrick and Tordella,
1988). Information on data sources are even more difficult to locate
because it is part of many different academic and applied fields of
study but unique to no single discipline (Murdock and Hamm,
1988b). In sum, practitioners have found that no single source exists
to address their needs.
Many professional demographers who were formally trained in
academic settings are becoming increasingly involved in the applied
uses of demography and are finding their formal training has not
properly prepared them to complete the tasks required of them in an
applied setting. For example, although they may have had several
courses that have provided them with indepth information on alter-
native techniques for completing regression analyses, they may have
had as little as a single class period in a demographic methods
course on techniques of population estimation. In this class period
they may have only examined such techniques as they are applied to
nations or states rather than small areas such as counties, places, or

3
census tracts. They are likely to find, however, that the formulation
of population estimates for such small areas is among those tasks
most often required of them.
They may also find that they are required to extend their demo-
graphic knowledge far beyond the areas pursued in their graduate
training. This training may have required them to complete analy-
ses of the effects of demographic factors on social stratification and
inequality, segregation, suburbanization, and levels of socioeconom-
ic development. They are likely to have reviewed numerous studies
of the interrelationships between fertility control and economic
development, the determinants of mortality differentials, and the
factors affecting the adoption of contraception or abortion practices.
They are much less likely to have examined analyses of the effects of
migration on the market for multi-family housing, the effects of
changing racial/ethnic composition on retail markets, or the implica-
tions of differential rates of population growth on the need to relo-
cate a public health clinic. Professional demographers new to the
world of applied research may find themselves searching unsuccess-
fully for a source that brings together the information they are likely
to require on a frequent basis.
This book attempts to meet the needs of both those who are not
trained in demography, but who are increasingly required to either
do demographic analyses or evaluate the results of such analyses,
and of those who have been trained in demography but require
more information on its applied dimensions. It does this by provid-
ing an introduction to: (1) demographic concepts and processes as
used in demography and applied demography; (2) sources and
typical applied uses of the most widely used demographic data; and
(3) techniques for analyzing demographic patterns and the effects of
demographic factors on socioeconomic conditions and characteristics.
Its intent is to provide one of the first relatively comprehensive
single-source introductions to the concepts, methods, and data of
applied demography. We begin this task by defining and delineat-
ing the subject matter of applied demography.
Definition and the Dimensions
of Applied Demography
An important starting point for any work is the definition of its
subject matter, in this case, applied demography. Applied demog-
raphy must be seen as a part of the broader field of demography.

4
However, within neither demography nor applied demography is
there universal agreement concerning the definition of what is, and
what is not, a proper area for demographic analysis. Therefore, the
reader should be aware that the definitions provided here do not
necessarily represent a consensus among demographers about the
definition of demography or applied demography.
The overall field of demography can be simply defined as the
study of human populations. Hauser and Duncan (1959), however,
note that demography has maintained two parallel traditions. One
is the domain of formal demography which has focused on the precise
mathematical measurement of the three demographic processes of fertility,
mortality, and migration. The sources of change in these processes,
the trends in these processes, the differentials in these processes,
and the interrelationships among these processes form the major
emphases in formal demography. The study of formal demographic
processes is often closely associated with mathematical demography.
Formal demography is an important but rather specialized subfield
within demography.
The second tradition in demography is broader and has a larger
number of adherents. It examines the determinants and conse-
quences of the demographic processes and of the size, distribution,
and composition of the populations that result from them. Thus
social demography can be defined as the study of the determinants and
consequences of population size, distribution, and composition and of the
demographic processes of fertility, mortality, and migration that determine
them. The emphases within this area of study has been on examin-
ing the interrelationships between demographic variables and other
social and economic variables. This concept of demography is
dominant in most academic departments teaching demography in
the United States. By comparison to formal demography, social
demography represents a substantial broadening of the subject
matter of demography.
In many regards, applied demography represents a further
extension of demography from the broader issues and dimensions
examined in social demography. As Rives and Serow note,
In our view, applied demography is that branch of the disci-
pline (of demography) that is directed toward the production,
dissemination, and analysis of demographic and closely relat-
ed socioeconomic information for quite specific purposes of
planning and reporting. To distinguish 11 applied11 pursuits
from other lines of demographic inquiry, we would further
suggest that applied demography is more concerned with the

measurement and interpretation of current and prospective
population change than with the behavioral determinants of
this change. . . .
Applied demography almost always deals with information
on population size, growth and composition for specific geo-
graphic areas. Thus there is an identifiable difference in the
unit of analysis: Applied demographers tend to focus on
geographic units and their population characteristics, while
others are more concerned with individuals and their demo-
graphic behavior (1984: 9-10).
5
Applied demography is thus different than the broader field of
demography in its relative emphases within the content areas of
demography. Rives and Serow (1984), suggest several emphases
that they believe separate the applied from the more basic aspects of
the discipline. We add to the areas delineated by Rives and Serow
(items 2, 3, and 5 below) and suggest that the differences between
basic and applied demography can be seen in terms of different
emphases within the following dimensions:
1. Scientific goal - Science can be seen as having three pri-
mary goals: description, explanation, and prediction.
Demography as a basic science tends to emphasize expla-
nation with secondary emphases on description and
prediction. Applied demography tends to emphasize
prediction, followed by description and explanation. In
addition, many applied uses of demography attempt to
establish concomitant demographic factors (e.g., for profil-
ing market segments). Such coincidental• occurrences
are seldom the focus of basic demographic analyses.
2. Time referent - Basic demography may examine demo-
graphic phenomena for historical, current, or future time
periods, but most frequently tends to involve attempts to
explain past events. Applied demography tends to place
emphasis on current and future patterns.
3. Geographic focus - Basic demography often attempts to
explain either international- or national-level patterns.
Applied demography tends to examine patterns for
subnational areas such as county and/or subcounty areas
(e.g., blocks, tracts). In addition, although general

6
demographic analyses are nearly equally likely to employ
aggregate areal data and data on individuals, applied
demographic analyses place very heavy reliance on aggre-
gate areal data for small areas.
4. Purpose of the analyses - The science of demography in
its basic form tends to emphasize analyses intended to
generate basic knowledge about the causes of demograph-
ic change which can be generalized as widely as possible
across as many different types of areas as possible. In
applied demographic analyses, the emphasis is on the
application of knowledge to discern the consequences or
concomitants of demographic change rather than on basic
knowledge generation. Applied demographic analyses
often use data to discern the extent to which the findings
from general studies of other areas apply to a specific
study area.
5. Intended use of analytical results - Basic demography is
intended primarily to enhance the base of knowledge in
the discipline, knowledge which is shared among scholars
within the discipline. The results of applied demographic
analyses are intended to inform decisionmaking among
non-demographers relative to the planning, development,
and/or distribution of public- or private-sector goods or
services.
Taken together, these emphases suggest that applied demography
can be defined as
the study of population size, distribution, and composition
and of the processes of fertility, mortality, and migration
in a specified study area or areas with emphases on gaining
knowledge of the consequences and concomitants of demo-
graphic change to guide decisionmaking related to the
planning, development, and/or distribution of public- or
private-sector goods or services for current and future use in
the study area or areas.
As such a definition suggests, applied demography requires
knowledge of both the basic science of demography and the means
by which it can be applied to address pragmatic and policy-related
questions.

7
The content of applied demography may also be examined by
describing the demographic variables on which its analyses tend to
concentrate. These variables include both demographic and those
found to have such dose relationships to demographic variables that
it is common practice to include them in almost any demographic
profiling of an area. These variables are
-population size
-population change
-mortality
-fertility
-migration (both national and international)
-population distribution (relative to metropolitan and
non-metropolitan areas, central cities and suburbs, rural
and urban areas, by the population size, density
of settlement, and among blocks, tracts, etc. of an area)
-compositional characteristics
·age
·sex/gender
·race
·ethnicity
·marital status (including never married, married,
separated, divorced, and widowed)
·household and family types
(including family and nonfamily households
and family and nonfamily households by sex
and marital status of householder [head] and
presence and/or number of children)
·educational status (both years and degrees
completed)
·employment by
-status (employed, unemployed or underemployed)
-occupation
-industry
·income, wealth, and poverty
·socioeconomic status (summative measures using
income, education, and occupational variables).
Of these variables, the education, employment, income, and soci-
oeconomic status variables might be considered as social and
economic rather than demographic variables. However, common
practice has so often included them in demographic analyses that it
is essential for those wishing to do applied demographic analyses to

8
be familiar with the data sources and measures of these variables.
Oearly other analysts might include additional variables or delete
some of the variables noted here, but we believe that such variables
are sufficiently encompassing that, if one has gathered data and
completed analyses of these variables for an area, one can be said to
have completed a relatively complete demographic analysis of an
area. Consideration of these variables relative to the applied dimen-
sions noted above can thus be seen as delineating the content of
applied demography. The description of the content and trends in
these variables, the sources of data on them, and the measures and
techniques for analyzing them is the focus of this book.
Organization of the Text
In the remainder of Chapter 1, we describe the organization of
the text and delineate the limitations of the work. In so doing, we
attempt to introduce readers to key dimensions examined in the
work and alert them to topics for which additional references should
be consulted. At the end of the work, references to additional de-
tailed sources are provided.
Chapter 2 defines and delineates the major trends in each
demographic concept covered in the work. As noted above, these
include the basic demographic variables of population change, age,
sex, race/ethnicity, household, family, and marital status, population
size, geographic patterns of population distribution, and the three
demographic processes of fertility, mortality, and migration. Also
examined are variables closely related to the basic demographic
variables, including employment status, occupation, industry,
income, education, and socioeconomic status. These variables are
defined and the trends in such variables likely to impact factors of
interest to applied demographers are described. As a result of
examining this chapter, the reader should obtain a basic understand-
ing of demographic variables and of the role of such variables in
altering socioeconomic factors of relevance to applied private- and
public-sector interests.
Chapter 3 examines the sources of data on the variables de-
scribed in Chapter 2. National and international, state and local,
and private data sources are described. The discussion includes an
examination of the forms of data available and of the limitations in
obtaining and using such data. A detailed examination of the data
products from the 1990 Census is presented and an analysis of the
implications of these products for data use is provided.
The next section describes measures and techniques for analyz-

9
ing the variables discussed in Chapters 2 and 3. Chapter 4 examines
basic measures of each of the variables and provides an introduction
to more comprehensive techniques utilizing multiple variables and
concepts such as life-table techniques (including a basic introduction
to multiple-decrement techniques), methods of standardization, and
rate decomposition. Because applied analyses tend to emphasize
current and future patterns, an entire chapter, Chapter 5, is devoted
to this topic. Thus, techniques to estimate and project population
and to evaluate population estimates and projections are examined
in Chapter 5. For each of these topics, examples of the use of the
techniques to address applied questions are presented.
The concluding chapter, Chapter 6, examines future trends that
are likely to become the focus of applied demographic analyses in
the future. Topical and substantive areas expected to provide the
basis for the expansion of applied demographic analyses in the
coming decades are then discussed. Finally, we examine the current
status of applied demography and suggest opportunities and poten-
tial problems affecting its future development.
Limitations of the Book
As with any such effort, space considerations, as well as the
experience and knowledge base of the authors, have limited this
book. The variables and techniques described and demonstrated are
limited to those we believe are most likely to be of use in applied
demography and are clearly only some of those which might be
examined.
In addition, the use of these factors are demonstrated for areas
in the United States so that the increasingly important international
uses of demography are not directly addressed. Similarly, emphasis
is placed on data sources used for applications in the United States.
It is also important to note that since this book was written as 1990
Census materials were just beginning to be released, much of the
discussion of 1990 Census products is based on the publication plans
of the U.S. Bureau of the Census. If the 1990 Census is similar to
past censuses, the final products are likely to be somewhat different
in form and more limited than those initially planned.
Greater emphasis is also placed on somewhat simpler techniques
rather than more sophisticated methods. For example, sophisticated
multiple-decrement life table techniques and multi-state regional
projection models are examined in only a very general manner.
This reflects our attempt to cover those topics we believe are likely
to be most frequently used by those who are entering the field of

10
applied demography and which are used in applied demography as
presently practiced. As the field of applied demography develops,
increasingly sophisticated techniques should come into more
common usage, and efforts such as this will require updating and
expansion.
Finally, it is likely that this effort is limited somewhat by the au-
thors' bases of experience which have largely been in the public
sector. Although a concerted effort was made to overcome this
limitation, it is likely that the authors' backgrounds and experiences
affected and perhaps limited the work in regard to some private-
sector uses of demographic techniques.
Despite. these limitations, we hope the work will be a useful
addition to the applied demographic literature. We also hope that
this attempt to introduce the concepts, methods, and data of applied
demography will encourage other scholars and practitioners to
develop additional works of utility for those who, not only study,
but also apply the body of knowledge in demography to address
pragmatic issues. It is to further explicate such issues and concepts,
as well as the data and techniques used to address them, that we
now tum our attention.

2
Demographic Concepts and Trends:
The Conceptual Base and Recent Patterns
of Demographic Change
The discussion in this chapter is intended to define the major
concepts and variables used in applied demography and to provide
information that will allow the reader to obtain an initial base of
demographic knowledge regarding current patterns for the measures
of these concepts and variables. It must be recognized, however,
that no single chapter, or any single work, can replace the need for
continuous study to obtain and maintain knowledge of demographic
change.
Defining Key Concepts and Terms
In this section, we examine some of the key concepts and terms
used in demography and demographic analyses. It is essential for
those using demographic data to be aware of the underlying defini-
tions and dimensions of demography's key concepts. We delineate
these concepts briefly below indicating both how they are defined
and the major differentials or variations in them among different
demographic groups and relative to other demographic, social, and
economic factors.
Population
Perhaps the most basic of all terms in demography is that of
population. A population consists of the persons living in a specific geo-
graphical area at a specific point in time (see Ryder, 1964 for a useful
description of the concept of population). Two aspects of the con-
cept of population as used in demography are important to empha-
size.

12
First, the term population tends to be used to refer to aggregate
characteristics of a population living in an area; that is, to character-
istics that are descriptive of the population but not necessarily of any
given individual within the population. For example, a population's
death rate is not reducible to the individuals within the population.
That is, any given person in an area is either alive or dead at a
given point in time; he or she has no death rate. On the other
hand, a population's death rate is the aggregate effect of all deaths
in the population. A death rate is thus uniquely an aggregate rather
than an individualistic measure.
A second aspect of the concept of population as used in demog-
raphy (and in statistics) is that it is used to refer to all of the persons
rather than to simply some (a sample) of the persons in an area.
Demographers often refer to a subgroup of a total population as the
population of persons with certain characteristics (e.g., the popula-
tion of females, the population of black residents), but when the
term population is used, the emphasis is generally on the total, the
sum total of, persons within an area.
Subpopulations and Cohorts
Persons using demographic data often also refer to subpopula-
tions such as the old, the young, blacks, whites, Hispanics, the baby
boomers, and similar groups. Any population group in a specified
area composed of persons with one or more common characteristics
can be referred to as a subpopulation. The concept of a cohort is
more specific and refers to agroup of persons with the common character-
istic of being born during the same period of time. Members of a cohort
may have other common characteristics (e.g., they may be males or
females, black, Hispanic, white), but they will always be persons of
similar ages. In addition, it should be recognized that the cohort is
a concept used in a very unique way in the social sciences (Glenn,
1977). It tends to refer not only to the possession of a common
biological age, but also to the fact that persons in any given cohort
are passing through the life cycle exposed to certain similar effects.
Cohort connotes not only birth during aspecified period, but commonality
resulting from the fact that its members have been socializ.ed during a period
of time with specific socioeconomic and historical events that are likely to
cause them to exhibit similar behaviors and have similar perspectives. For
example, those who reached adulthood during the Great Depression
of the 1930s are commonly referred to as the depression cohort,
those socialized during the 1960s as the sixties generation, those

13
born from 1946 to 1964 as the baby-boom generation, and those born
after 1964 as the baby-bust generation. Such groups are seen as
having unique characteristics that are a function not only of age, but
also of sharing a commonality of experiences during their childhood
and young-adult formation years (Ryder, 1965).
In demography and the social sciences generally, the concept of
cohort is also used to connote a specific form of analysis in which
groups of persons (i.e., given cohorts) are followed through time in
an attempt to discern whether certain characteristics displayed by
them, such as changes in rates of births, income levels, etc., are a
function of cohort effects or of other factors. Often, cohort effects
are differentiated relative to the effects of a specific period of time
(referred to as period effects) and effects that are a function of age
(that is, age effects). By comparing the patterns for a cohort across
time relative to the patterns for persons at the cohort's age at several
different points in time (relative to period effects) and relative to
patterns for different age groups at different points in time (age
effects), the unique effects of being a member of a given cohort can
be, at least partially, identified (Mason et al., 1973; Glenn, 1977;
Palmore, 1978; Rodgers, 1982).
Population Change
Population change is a function of three processes referred to as
the demographic processes or components. These are births, deaths
and migration. The relationship between these variables is perhaps
best seen in the simple population equation (sometimes also called the
lJookkeeping equation of population). This equation is as follows:
P P B D M
t2 = tl + tl - t2 - tl - t2 + tl - t2
Where: Pt2 = population for a second date (t2)
Pt1 = population at the base date (t1)
Bt1 - t2 = number of births that occur during the time
interval from the base date (t1) to the second
date (t2)
0 t1 - t2 = number of deaths that occur during the time
interval from the base date (t1) to the
second date (t2)

14
Mt1 - 12 = amount of net migration that occurs during
the time interval from the base date (t1) to
the second date (12)
Therefore, to understand population change, it is necessary to
understand patterns of births, deaths, and migration.
Understanding the sources of population change, whether it is a
result of patterns of births and deaths (processes whose combined
effects are referred to as natural increase or natural change) or of migra-
tion, is of vital importance because the determinants and conse-
quences of the processes of natural increase and migration are quite
different.
Death is a result of physiological processes and the attempt to
lengthen life is a major goal of nearly every society. Fertility in-
volves a biological process which results from sexual behavior that
may or may not hav~ been intended to produce a conception and
birth. Migration is a behavior involving moving from one area to
another. Although migration often involves reactions to physical
factors (e.g., shortages of food and other basic necessities for surviv-
al), migration is clearly the demographic process that is most often a
result of non-physiological processes, such as employment, income,
and other socioeconomic changes (Long, 1988).
As a result, although deaths and births impact a population by
decreasing or increasing its size, their effects on other nondemo-
graphic and socioeconomic factors are usually long-term. Migration
by contrast has a more immediate impact on an area because it is
more likely to involve young adults in their family-formation ages.
In terms of commercial activities, births and deaths are likely to
have immediate impacts on only a few markets (such as markets for
baby goods) and may lead to long-term growth or decline in markets
for housing and other goods and services. However, migration
tends to have immediate impacts, reducing markets for products and
services in areas with net outmigration and creating immediate
demands for all those goods and services necessary to establish a
residence in areas with patterns of net inrnigration.
The Demographic Processes
(Components of Population Change)
As noted above, the three processes that change populations are fertil-
ity, mortality, and migration. These involve births into a population,
deaths from a population, and migration either into or out of a

15
population. Although these processes are sufficiently well known as
to not require the presentation of extensive definitions, selected
aspects of each, and related terms often associated with each, re-
quire some description.
Fertility. Fertility refers to reproductive behavior in populations.
Fertility rates indicate the relative incidence of births in a popula-
tion. Fertility is commonly distinguished from fecundity which refers
to the biological capacity to conceive and bear children. Fertility tends to
be highest among women in their twenties and lower among
women of younger or older ages with the child-bearing ages being
variously defined as starting at age 10 or 15 years of age and extend-
ing to ages 44 or 49. Recently, women in their thirties have shown
increases in fertility. Although the rates for women in their thirties
remain lower than those for women in their twenties, the pattern of
high fertility for women in their late thirties is largely unprecedent-
ed. At present, it is unclear whether this new pattern is a tempo-
rary result of delayed child-bearing among baby-boom-era women or
a new longer term pattern of increased fertility (however, see Ryder,
1990). Fertility has also tended to be higher among populations
with fewer socioeconomic resources. This applies both to societies
taken as a whole (e.g., fertility is generally higher in developing
than in developed nations} and also to specific groups within any
given society (i.e., persons with fewer socioeconomic resources tend
to have more children than those with more socio-economic re-
sources}.
Mortality. Mortality refers to the incidence of deaths in a popula-
tion. It is commonly distinguished from morbidity which refers to the
incidence of disease in a population. It is often discussed in terms of the
contravailing process of survival-that is, the probability of surviving
over a given period of time. Mortality in the United States and
other developed nations has tended to demonstrate the presence of
what some refer to as an epidemiological transition (Preston, 1976).
This is a shift in an area from conditions in which a majority of
deaths occur from infectious diseases (pneumonia, diarrhea, dysen-
tery, etc.} to ones in which chronic diseases (coronary disease,
cancer, etc.) are the major causes of death. Mortality tends to also
be differentiated by socioeconomic factors such that mortality is
substantially higher among those with more limited socioeconomic
resources. As discussed in detail below, the analysis of mortality is
often completed using a set of procedures referred to as life-table

16
techniques, techniques in which the distribution and impacts of
deaths over time are simulated in a hypothetical population. Be-
cause it is one indicator of an area's likely level of economic devel-
opment, infant mortality (deaths to persons during their first year of
life) is often used as a measure of socioeconomic development in
analyses of socioeconomic conditions.
Migration. Migration refers to the movement of persons in a popula-
tion from one area to another. Unlike the demographic processes of
fertility and mortality, migration is not discretely fixed in time and
space, that is, to define migration requires that one define when and
how far someone has moved. Migration is usually distinguished
from both daily patterns of movement and short-distance permanent
moves. That is, commuting and similar, frequent patterns of re-
peated travel that do not involve a change in residence and move-
ments within the same general residence area (e.g., a move from
one housing unit to another in the same neighborhood) are not
commonly referred to as migration. The U.S. Bureau of the Census
defines migration in terms of a change in residence in which the origin
residence and the destination residence are in different counties.
Migration researchers have variously defined migration (Ritchey,
1976) but Mangalam and Schwarzweller (1968) have usefully defined
migration as involving movement of a person from one social system
to another in which the migrant is required to change friendship and
social and economic interrelationships. Whatever the definition,
migration tends to result from a complex set of economic, demo-
graphic, and social factors (Long, 1988) and has, as a result, received
extensive attention from other social scientists as well as demogra-
phers (Ritchey, 1976; Greenwood, 1985; Lichter and DeJong, 1990).
Migration is distinguished also by its direction and by whether
or not it involves crossing a national boundary. Migration involving
two nations is referred to by the terms immigration and emigration.
When referenced in regard to the receiving nation, persons moving
into that nation have immigrated to it while persons leaving it are
emigrating from it. Migration within a nation is referred to using
the terms inmigration and outmigration for movement (in the
United States defined as movement involving a change in residence
from one county to another) from the perspective of the receiving
and sending areas respectively. All areas tend to have both in- and
outmigration (and/or if it also involves international movement, im-
and emigration). As a result, two terms are frequently used to

17
identify the joint effects of in and out migration (or im- and emigra-
tion). These terms are gross migration, to refer to the sum of in and out
movements, and net migration, to refer to the difference between in and out
movement. Net migration is perhaps the most widely used term with
a plus sign being used before a net migration value to indicate net
inmigration and a negative sign used to indicate net outmigration
relative to a reference area.
As a process, migration tends to occur most frequently among
young adults and to decrease in frequency with age, to be more
prevalent among members of populations with higher levels of
education, higher incomes, and higher status occupations (that is,
among persons with greater socioeconomic resources) and among
those in developed nations. The level of migration also tends to
increase during periods of economic expansion and to be reduced by
periods of recession and depression (Greenwood, 1985).
Population Distribution
Population distribution refers to how the population of an area is dis-
tributed relative to its physical land area and according to key sites or types
of sites (e.g., rural and urban areas, small and large cities) in the area..
Populations are distributed within an area as a result of a variety of
physical and socioeconomic factors such as environmental features
or employment patterns. Populations redistribute themselves by the
three demographic processes of fertility, mortality, and migration,
with migration providing the most common form of rapid redistribu-
tion. An area's population distribution is commonly described
according to such categories as rural and urban, metropolitan and
nonmetropolitan, and by the size of the population of settlement
sites, by the density of settlement, etc.
In general, developed nations such as the United States have
shown patterns of increasing concentration of their populations in
large urban centers. As a result of such patterns, by 1990, 77.5
percent of the population of the United States lived in metropolitan
centers compared to 22.S percent who lived in nonmetropolitan
areas. Also prevalent in the United States in recent decades has
been an increasing concentration of residents in suburban areas
within larger metropolitan areas (Frey and Speare, 1988) and the
more extensive growth of the southern and western regions of the
United States relative to the northeastern and midwestem regions of
the United States (Long, 1988).

18
Population Composition
Population composition refers to the characteristics of a population.
Such characteristics include whether the population is young,
middle aged, or elderly; predominantly male or female; composed
primarily of single or married adults; and of persons living primarily
in single-person or multi-person households or in families. It in-
volves knowing how many occupied housing units are rented and
how many are owned; how many persons are white or black;
Hispanic or non-Hispanic; wealthy or poor; well-educated or poorly
educated; employed in professional and white-collar occupations or
blue-collar and laborer occupations; and employed primarily in ex-
tractive industries (such as agriculture or mining), or in manufactur-
ing or service industries. Knowledge of such characteristics is
among the most important factors in understanding how to use
demographic information to address such pragmatic issues as how a
population will react to a given set of events or a new product or
service. We briefly examine key compositional characteristics of
populations by describing several of the major demographic charac-
teristics and the major differentials associated with them within the
U.S. population.
Age. Age is commonly measured as the age of a person as of
their last birthday. Age is a biological and chronological factor with
demographic, social, and economic importance. Certain rights, (e.g., the
right to vote and to marry) and obligations (for military duty or legal
culpability) are related to age. As noted above, the concept of
cohort, referring to a group of persons born during a specific period
of time, is a commonly used age-related concept in demography.
Similarly, certain age-determined groups related to specific stages in
the life cycle and/or specific dates are also commonly referred to in
applied demographic analyses. School-age persons are commonly
those 3-to-17 or 18 years of age, college-aged are generally those 18-
to-24 years of age, women of child-bearing age are those who are 10
or 15-to-44 or 49 years of age, middle-aged those 40 or 45-to-60 or 64
years of age, and the elderly those 65 years of age or older. The
baby-boom generation refers to those born in the years inclusive of
1946 through 1964 and the baby-bust generation to those born after
1964. Age is generally reported in either single years or five-year
age groups starting with the five-year age group of 0 through 4
years of age and ending with an age group that includes persons in
a specific age and all older ages (e.g., 65 or 75 and older). Median

19
age is perhaps the most commonly used measure of age. The most
often noted trend is that the age of the population (at least in de-
veloped nations such as the United States), has increased substan-
tially such that the median age of residents of the United States was
roughly 23 in 1900, 33 in 1990, and is expected to be about 36 in the
year 2000 (Spencer, 1989).
Sex or Gender. Sex is a variable with biological, demographic, social.,
and economic significance. Gender is now the commonly used term to
connote the nonbiological differences associated with differences in sex. In
this text, we use the term sex because emphasis is placed on biologi-
cal differences. This is not intended, however, to diminish the
importance of the critical socioeconomic dimensions entailed in
gender differences.
Although approximately 105 males are born per 100 females,
due to the greater life-expectancy of females, the number of females
becomes roughly equal to the number of males between the ages of
20 and 30, and females outnumber males by nearly 2-to-1 at ages
over 80. Females have historically been the focus of discrimination
and received substantially lower returns to their labors, earning 60
to 65 percent of that earned by males in the same jobs. In addition,
females are heavily concentrated in clerical and other occupations
with low returns to labor. The distribution between the sexes is
generally described simply in terms of the percent of the population
0£ each sex or by the sex ratio which indicates the number of males per
100 females.
Race/Ethnicity. Race and ethnicity are commonly used to refer
to differences among population groups related to differences in
cultural, historical, or national-origin characteristics. Although the
concept of race was once assumed by some segments 0£ some socie-
ties to describe a base of biological differences, race has come to indi-
.cate differences that are largely socioeconomic and cultural. Ethnicity
generally refers to the national, cultural, or ancestral. origins of a people.
In the two most recent censuses, both concepts were measured by
respondents self-identifying themselves using two separate ques-
tions. For example, one question on the 1990 Census form asked re-
spondents to identify themselves using the racial categories of
white; black; American Indian, Eskimo or Aleut; or Asian and Pacif-
ic Islander with the last category having nine alternative Asian and
Pacific Islander categories (Chinese, Filipino, Hawaiian, Korean,
Vietnamese, Japanese, Asian Indian, Samoan, Guamanian) plus an
other (Asian and Pacific Islander) category with space provided to

20
write in a response. Finally, this question provided an other
category with a space for the respondent to write in a response.
A second question asked census respondents to indicate whether
they were of Spanish/Hispanic Origin, for which they were given
the response categories of no and yes with the yes response having
the alternative categories of response of Mexican or Mexican Ameri-
can or Chicano; Puerto Rican; Cuban; and other Hispanic with a
blank being provided to write in a specific response to the other
Hispanic category.
These two questions are intended to determine both the race
and Spanish/Hispanic Origin for each respondent but many
respondents are apparently confused by these questions. For
example, nearly 90 percent of Hispanics have historically reported
themselves to be white but in the 1980 and 1990 Censuses many
reported themselves as being in the other race category. Thus, of
the 9.8 million persons who indicated that their race was other in
1990, more than 97 percent were Hispanics. Many Hispanics appar-
ently used the Other category as a residual category because they
were uncertain how to respond to the race question. Terms such as
Anglo, which is commonly used to refer to white non-Hispanics,
cannot be determined directly from the census items but must be
derived by cross-classifying the results from the race and ethnicity
questions. It is obvious that race and ethnicity are complex concepts
both for those who would measure them and for persons who
respond to questions about them.
In addition to questions on race and ethnicity, other data on
heritage are also available from the census and elsewhere. These
indicators of heritage include country of birth and ancestry (such as
whether a respondent is English, German, etc.). These latter data
are important for identifying such factors as preferences in food and
other products that have distinct cultural origins.
In analyses for the United States, the minority groups most
often examined are blacks, Hispanics, and Asians. The most impor-
tant demographic differentials among such groups in the United
States are the substantially faster rates of growth among minority
populations relative to majority groups and the increasing share of
the population that is minority. In 1980, for example, blacks were
11.7 percent of the population, persons of Asian extraction made up
1.5 percent of the population and persons of Hispanic origin ac-
counted for 6.4 percent of the U.S. resident population of
226,545,805. From 1980 to 1990, the total population increased by
9.8 percent, but the black population increased by 13.2 percent, the
Asian population by 107.8 percent and the Hispanic population by

21
53.1 percent. By 1990, blacks made up 12.1 percent, Asians 2.9
percent and Hispanics 9.0 percent of the 248,709,873 persons in the
United States, together accounting for nearly 60 million persons. In
addition, by 2025, U.S. Bureau of the Census projections (Spencer
1986; 1989) suggest that blacks could account for 14.6 percent of the
population, persons in other races (including Asians) for 6.5 percent
and Hispanics for 13.1 percent. Clearly, patterns associated with
these groups will increasingly shape public- and private-sector
events in U. S. Society.
For purposes of applied product- and service-related analyses,
the importance of race and ethnicity lies primarily in the fact that
racial and ethnic minorities, such as blacks and Hispanic Americans,
tend to have more limited socioeconomic resources. Poverty rates
are two to three times those for whites, incomes approximately 60 to
70 percent of those for whites, and levels of education are substan-
tially less than those for whites (for example, in 1980, 40% of
Hispanics in the United States and 27% of blacks had 8 or fewer
years of education compared to just 17% of whites, while roughly
8% of Hispanics and blacks had a college education compared to
17% of whites). This affects the purchasing powers of such minori-
ties and increases their levels of need for many types of public serv-
ices. This unfortunate relationship between minority status and
reduced socioeconomic resources is pervasive across nearly all re-
gions of the United States and is evident among certain minority
groups in other nations as well. By contrast, Asians tend to have
lower levels of poverty, to be more highly educated, and to have
higher incomes than whites. Because of such differences, race and
ethnic differences are a major topic of demographic analyses.
Marital Status. Marital status is closely related to the likely
economic circumstances of the household members within married-
couple versus unmarried-person households, the probability that a
woman will bear off-spring, and numerous other factors. Distinc-
tions are usually made between those persons who have never been
in an officially recognized union, referred to as the never married;
those in such a union, the married; and those who have previously
been in such a union but are either separated, divorced, or wid-
owed. Increasingly, however, it is evident that a substantial number
of persons are in unions that lead them to make joint decisions, but
whose unions lack the formal status of marriage, such as persons
who are cohabitating (Bumpass and Sweet, 1989). The trends in
marital status over time show that an increasing proportion of

22
persons will either not ever be married or will find themselves in a
broken union of some form.
Marital status and its trends are important for those who do
applied analyses because those in marital unions tend to have more
resources than those in other forms of unions or those who are not
in unions. Analyses show that persons in households that have
been disrupted by marital dissolution are likely to experience
substantial disadvantages compared to those in intact households
relative to income and socioeconomic opportunities (Bianchi and
McArthur, 1991). They are likely to have lower purchasing power
and more imminent needs for public services than those in married
unions. The delineation of the variable of marital status thus
continues to be of importance.
Household and Family Characteristics. Household and family
characteristics are important because they indicate ways that group-
ings of intimate persons are united in response to demographic,
social, and economic conditions. They are purchasing and consum-
ing units, and their numbers and characteristics have significant
implications for the demand for goods and services. As generally
defined, a household refers to the persons living in a single housing unit.
A housing unit is any type of residence (house, apartment, mobile
home, townhouse, condominium, etc.) that is occupied as a separate
living quarters (quarters in which occupants live and eat separately
from persons in other households and which have access to their
living area from the outside of a building). Households are of one of
two types, family or nonfamily. Family households c.onsist of two or more
persons who are related by marriage, birth, or adoption, while nonfamily
households c.onsist of one person or two or more unrelated persons living in
a single housing unit. Within family households, distinctions are
commonly made between families with married couples (both with
and without children) and those involving a male or female house-
holder with one or more children or other relative. The term
householder was established in the late 1970s to avoid the use of the
term of head of household which persons tended to assume referred
to a male. A householder is the person in whose name a unit is owned or
rented or anyone so designated as the major supporter of the household by
other household mem11ers. As with the term head, it is largely used as
a term indicative of the person who provides a majority of the
support for a household.
Trends in households and families have been among the most
important demographic changes affecting the public and private

23
sectors. In general, these trends show that the size of households
has decreased (from an average of 3.67 persons in 1940 to 2.63 per-
sons per household in 1990), the number of households involving
married-couple families has declined (from 70.6% in 1970 to 55.1%
in 1990), and nonfamily households are growing more rapidly than
family households (e.g., family households increased by 11% from
1980 to 1990, while nonfamily households increased by 29.0%).
These changes are important because they have affected both the
number of households and the socioeconomic resources of house-
holds. For example, the number of households in the United. States
increased from 63.4 million in 1970 to 91.9 million in 1990, an in-
crease of 28.5 million. However, if the average size of households
in 1970 of 3.17 persons had prevailed in 1990 (instead of the average
household size of 2.63 persons), there would have been only 76.3
million households in 1990 rather than 91.9 million. Thus, it can be
argued that 15.6 million of the 28.5 million increase in households
from 1970 to 1990 was a result of changes in household size, rather
than population change and other factors. The wealth of house-
holds is also markedly affected by their composition. For example,
although median household income in all households in the United
States was $28,906 in 1989, it was $38,664 for married-couple fami-
lies but only $17,383 for families with a female householder and no
spouse present. Household and family characteristics clearly require
careful analysis because they have quantitative and qualitative
impacts on a population's standard of living.
The only persons who do not live in households are those who
live in various types of institutions, such as those in college dormito-
ries, long-term care facilities, ·military bases, prisons, and other insti-
tutional settings. These persons are referred to as the group-quarters
population. Although they are a small proportion of the total U.S.
population (about 6.7 million of 248.7 million in 1990), they must be
removed from the total population in examining and computing
household size and are a significant part of the populations of some
areas. Their significance for applied public- and private-sector
analyses lies in the fact that they tend to have distinctly different
patterns of expenditures and service usage. Failure to recognize that
an area has a large group-quarters' population is likely to lead to a
faulty analysis of the socioeconomic limitations and opportunities of
the population in a .market or service-delivery area. In addition, as
noted below, failure to adjust for group quarters populations in
making population estimates and projections can lead to inaccuracies
in estimates and projections.

24
Educational Status. The level of education and training in a
population is an increasingly important indication of that
population's ability to compete in the global market place. Education
is commonly measured in either years of school completed or in terms of the
attainment of certain levels of education such as grade school, high
school, technical school, college, graduate school, or professional
school. Although educational involvement can occur at any age, it
is most commonly examined relative to such involvement in the
ages from about 3 or 5 years of age to 35 years of age. Trends in
education have generally been ones of increased general levels of
education in the United States since 1940 with the proportion of
persons completing high school nearly tripling since 1940 but with
marked differentials in education remaining between those with
larger socioeconomic resource bases and those with smaller resource
bases.
Employment Status, Occupation, and Industry of Employment.
Employment refers to the characteristic of being involved in an activi.ty that
results in the attainment of resources for the person or per$ons involved.
In the United States, the characteristic of employment in a popula-
tion is most often assessed relative to a population's involvement in
gainful activity as measured by the proportion of eligible persons
(usually defined as persons 14 or 16 through 64 years of age) who
are either employed or unemployed. It is also measured in terms of
the type of job held by those employed, referred to as the occupa-
tion of employment (e.g., employment in professional or technical
occupations, crafts or service occupations), or the type of business,
referred to as the industry of employment (e.g., agriculture, mining,
manufacturing, services). Those in the labor force but not employed at a
given point in time are the unemployed. Attempts are also sometimes
made to assess the extent to which a population is underemployed as
indicated by fewer hours of work than is considered normal for a
person employed full-time (full-time employment is variously de-
fined as involving employment of 30, 35, or 40 or more hours per
week) and/or employment of persons in jobs with skill and educa-
tional requirements that are less than the levels of education and
skill they possess (Lichter and Constanzo, 1987).
The major trends in patterns of employment are those toward
increased proportions of persons being employed in service occupa-
tions and industries and a decreasing proportion employed in labor
and other low-skill occupations and in extractive (such as agriculture

25
and mining) or manufacturing industries. Of significance as well is
the increase in the proportion of women in the labor force, even
among those with young children. Finally, there remains a substan-
tial difference in levels of unemployment and underemployment
among those with larger and fewer socioeconomic resources, those
with fewer resources having substantially higher rates of unem-
ployment and underemployment, lower economic returns to their
labor, and longer periods of unemployment between jobs.
Income, Wealth, and Poverty. These characteristics indicate the
relative resources of a population for obtaining goods and services.
Income generally refers to money income received. on a recurrent basis as a
return for labor. It may include wages, pension funds, various forms of
public assistance, interest income, and even in-kind resources (e.g., the
value of a rent-free residence). The three measures most commonly
used to measure it are per capita income, mean income, and median
income. Per capita income is the arithmetic mean income per person
in an area. Mean income is often computed per household or family.
Median income is the income level that equally divides a ranked
income distribution of persons, households or families. Wealth refers
to the possession of goods, property, and other items that have a
market value; that is, that could be sold for a given amount.
Poverty is the absence of wealth and is an officially designated
amount of money which varies over time (depending on assess-
ments of the cost of living, household type and size, and the
number of children in a household). Income is commonly discussed
either in terms of current (nominal) dollar values or in terms of
constant dollars; that is, expressed in terms of the dollars for a
specific year for which adjustments have been made for rates of
inflation.
Data on income have shown relatively little change in incomes
for households since the late 1970s, when constant dollar income
values are examined. The elderly have shown the largest increases
in income of any age group, while relative incomes of racial and
ethnic minorities and of women and children have shown few
gains in the last decade relative to those for majority populations
and males. Wealth tends to be concentrated in majority populations
and among those in middle and elderly ages and much of the asset
wealth of Americans has been found to lie in the value of homes
(U.S. Bureau of the Census, 1990c). Poverty has remained relatively
stable in the total population but has increased among children and
decreased among the elderly.

26
Socioeconomic Status. Sodoeconomic status is a variable which
attempts to measure the combined effects of income, occupation, and educa-
tion. As commonly defined, the socioeconomic status of persons in
a population is a function of employment in certain occupations and
the possession of higher income and educational levels. In the
United States, employment in professional fields (such as medicine
and law), high incomes, and advanced levels of education common-
ly connote higher socioeconomic status. This status involves both
the possession of monetary resources and of prestige that allows one
to have a greater influence on decisions. Although socioeconomic
status is largely a social variable, the influence of socioeconomic
characteristics on such demographic factors as infant mortality, fertil-
ity rates, rates of migration, the density of settlement, household
size, as well as numerous other factors, suggest its relevance in
demographic analyses.
Socioeconomic status can be formally measured through the
use of several widely used indices which combine income, educa-
tion, and occupational factors into a single score. Among the most
widely used of such scales are those by Duncan et al. (1972) and
Nam and Powers (1983). However measured, socioeconomic status
is an important variable in the determination of purchasing patterns
and preferences for private-sector goods and services and the need
for many types of public services.
An Overview of ¥ajor Demographic Trends in the United States
The above concepts are ones that are central to demographic
analyses. Having provided a basic overview of their content, it is
important to briefly describe changes in the patterns related to these
factors. Such basic knowledge is essential because it allows the
applied analyst to anticipate the demographic conditions and trends
likely to be evident in an analysis for any given area and to evaluate
the likely accuracy of an analysis by comparing patterns identified in
it to general patterns and trends. Below, a basic overview is provid-
ed of recent and projected future trends in the demographic factors
described above for the United States.
Population Change
Table 2.1 provides data showing the historical growth of the
population of the United States from the first census in 1790 to the
most recent 1990 Census. The data in this table show that the
United States has had a history of rapid growth, exceeding 30

Year
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
Table 2.1: Total Resident Population and Percent
Population Change In the United States,
1790-1990
Tota I Percent
Population Change
3,929,214
5,308,483 35.1
7,239,881 36.4
9,638,453 33.1
12,866,020 33.5
17,069,453 32.7
23,191,876 35.9
31,443,321 35.6
39,818,449 26.6
50,155,783 26.0
62,947,714 25.5
75,994,575 20.7
91,972,266 21.0
105,710,620 14.9
122,775,046 16.1
131,669,275 7.2
151,325,790 14.9
179,323,175 18.5
203,302,031 13.4
226,545,805 11.4
248,709,873 9.8
Source: Values for 1790-1970 from United States Department
of Commerce, Bureau of the Census. Historlcal Statistics of
the United States: Colonial Times to 1970, Part 1 and Part 2,
Washington DC: U.S. Government Printing Office, 1975.
Values for 1980 and 1990 from the PL94-171 Census Tapes
for the appropriate censuses.
27

28
percent per decade for all decades from 1790 through
1860 and 20 percent for those from 1860 through 1910. Most of the
decades of the twentieth century have produced patterns of reduced
growth relative to those of the eighteenth and nineteenth century.
Rates of growth after 1910 have been at levels of less than two
percent per year, and the most recent census shows the 1980 to 1990
period to have produced the slowest growth of any decade in the
twentieth century, except for the decade of the Great Depression.
Slow growth is the prevailing pattern and one that is likely to con-
tinue.
Components of Population Change
U.S. population growth has been largely dependent on
natural increase, despite extensive immigration. In fact, analyses of
data since the early 1800s suggests that even during the period of
most extensive immigration to the United States, 1880 to 1920,
migration never accounted for more than 40 percent of population
growth in any decade (Nam and Philliber, 1984). Table 2.2 shows
the components of growth for the period from 1940 to 1990. An
analysis of this table shows that migration has become a renewed
source of growth in recent decades. Migration, which was 3.3 to 3.5
million in the 1950s and 1960s, exceeded 14 million between 1970
and 1990, while natural increase peaked during the height of the
baby boom in the 1950s and then declined. Thus, the estimates of
intercensal change in Table 2.2, indicate that natural increase was
16.9 million during the 1980s compared to 24.6 million in the 1950s,
a decline of 31 percent. Such trends suggest that population growth
in the United States will be increasingly dependent on immigration
from other nations. In addition, the origin of immigrants to the
United States have shifted from Europe and other developed west-
ern nations of the world during the last few decades of the last
century and the first decades of this century to Mexico, South and
Central America, and Asia during the most recent decades (Bouvier
and Gardner, 1986).
The data in Tables 2.3 and 2.4 show patterns for the three
demographic components both over time (Table 2.3) and by age
(Table 2.4). The patterns by age are critical to understanding the
impacts of these processes, because the wide variability in the rates
for these processes by age can lead to substantial changes in the
number of vital events and in the number of migrants, even if the
rates by age have shown relatively little change.

Table
2.2:
Components
of
Population
Olange
for
the
United
States,
1940-1990
(numbers
In
thousands)
Population
Natura
I
Percent
of
Net
Percent
of
Change
Increase
Increase
Immigration
Increase
Total
Previous
Previous
from
Natural
Previous
from
Net
Year
Population
Decade
Decade
Increase
Decade
Immigration
1940
131,669
1950
151,326
19,657
13,791
70.2
5,866
29.8
1960
179,323
27,997
24,635
88.0
3,362
12.0
1970
203,302
23,979
20,448
85.3
3,531
14.7
1980
226,546
23,244
13,999
60.2
9,245
39.8
1990
248,710
22,164
16,893
76.2
5,271
23.8
Souru:
Population
values
for
1940-1980
from
the
Census
of
Population
for
selected
years.
Population
values
for
1990
from
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Population
Trends
and
Congressional
Apportion-
ment,•
1990
Census
Profile
No.1,
Washington,
DC:
U.S.
Government
Printing
Office,
1991.
Estimates
of
components
of
population
change
for
1940-50,
1950-60,
and
1960-70
from
Bogue,
D.J.
The
Population
of
the
United
States.
New
York:
Free
Press,
1985.
Estimates
of
components
of
population
change
for
1970-80,
•
United
States
Department
of
Com-
merce,
Bureau
of
the
Census.
Cumnt
Population
Reports,
P-25,
No.
1023,
Washington,
DC:
U.S.
Government
Printing
Office,
1989.
Components
of
change
for
1980
to
1990
computed
using
data
from
United
States
Department
of
Com-
merce,
Bureau
of
the
Census.
Current
Population
Reports,
P-25,
No.
1044,
Washington,
DC:
U.S.
Government
Printing
Office,
1989
and
from
Monthly
Vital
Statistic
Report,
Vol.
39,
No.12,
Washington
DC:
National
Center
for
Health
Statis-
tics,
April,
1991.
~

30
Table 2.3: Birth, Death, and Net Migration Measures• for the
United States, 1940-1990
Year
1940
1950
1960
1970
1980
1990
Year
1940
1950
1960
1970
1980
1990
Year
1940
1950
1960
1970
1980
1988
Crude
Birth
Rate
19.4
24.1
23.7
18.4
15.9
16.7
Crude
Death
Rate
10.8
9.6
9.5
9.5
8.1
8.6
Annual
Number of
Immigrants
70,756
249,187
265,798
438,000
530,639
643,025
Fertility Measures
General
Fertility
Rate
79.9
106.2
118 .8
87.9
68.4
71.1
Total
Fertility
Rate
2.3
3.1
3.7
2.5
1. 8
1.9
Mortality Measures
Infant
Morta 1i ty
Rate
54.9
33.0
27.0
21. 4
12.9
9.1
Life
Expectancy
at Birth (yrs.)
62.9
68.2
69.7
70.8
13.1
75.0
Migration Measures
Year
50-51
60-61
70-71
80-81
85-86
Total Percent
Involved in
Internal
Migration
5.6
6.3
6.5
6.2
6.7
*For definitions of these rates, See Chapter 4
Source: Birth and death data from the National Center for
Health Statistics for the respective years. Migration
data for 1940-1980 from Bogue, D.J. The Population of
the United States, New York: Free Press, 195. Data for
1988 for migration from United States Department of
Commerce, Bureau of the Census. Current Population
Reports, P-25, No. 1057, Washington, DC: U.S. Gov-
ernment Printing Office, 1990. Values for 1990 com-
puted using data from Current Population Reports, P-25,
No. 1018.

Table 2.4: Age-S~ Birth, Death, and Migration
Rates In the United States for Selected
Years
1980 1990
Age Birth Rate Birth Rate
10-14 1. 1 0.8
15-19 53.0 49.3
20-24 115.1 105.5
25-29 112.9 110.9
30-34 61.9 72.3
35-39 19.8 26.0
40-44 3.9 5.0
45-49 0.2 0.2
1980 1990
Age Death Rate Death Rate
1 year 12.9 9.4
1-.4 0.6 0.5
5-14 0.3 0.2
15-24 1. 2 1.0
25-34 1.4 1.4
35-44 2.3 2.2
45-54 5.8 4.7
55-64 13.5 11.8
65-74 29.9 26.2
75-84 66.9 61.4
85+ 159.8 149.7
1985-86
Age Migration Rate
1-4 9. 1
5-9 6.7
10-14 5. 1
15-19 6.5
20-24 13. 1
25-29 12.5
30-34 8.1
35-44 6.0
45-54 4.0
55-64 3.5
65-74 2.0
aAll rates are per 1,000 persons. Rates for 1990 as projected In
Current Population Reports, Series P-25, No. 1018.
Source: Birth and death rates from the National Center for
Health Statistics for the respective years. Migration rates
from United States Department of Commerce, Bureau of
the Census. Current Population Reports P-20, No. 425.
Washington, DC: U.S. Government Printing Office, 1988.
31

32
The fertility rates in Table 2.3 clearly show that fertility
peaked during the baby-boom decades of the 1950s and 1960s and
declined substantially by 1980. Although the rates for 1990 suggest
that fertility rates increased during the 1980s, the rates for 1990 are
still substantially lower than those in 1960 or 1970.
Mortality measures indicate that mortality has declined and
life expectancy increased during the last several decades. The crude
death rate has declined by 20 percent, the infant mortality rate has
declined by more than 80 percent, and life expectancy has increased
by 12 years since 1940.
Finally, the data on migration in this table point to an in-
creasing level of international immigration and to a continuing, rela-
tively high incidence of internal migration within the United States.
The age-specific rates in Table 2.4 show how sensitive each
of the three demographic processes is to age differences. Fertility
rates reach their peak between the ages of 20 and 30. Although
rates for those over 30 have increased substantially in recent years,
the birth rate is still highest in the age groups under 30 years of age
and declines thereafter. Death rates show a pattern sometimes
referred to as the age-curve of mortality, with relatively high death
rates occurring among persons under one-year of age, followed by
relatively low rates through ages 35-44. Mortality then begins to in-
crease so that between the ages of 55 and 64 mortality is again as
high as during infancy and then increases sharply in older age
groups. Finally, the data in Table 2.4 show that migration is, like
fertility, concentrated in the young adult years.
These age-specific patterns suggest that populations with
large proportions of their populations in their young adult years will
tend to have high levels of migration and fertility and relatively low
levels of mortality, while aging populations will show increased
levels of mortality and reduced fertility and migration. The high
levels of population growth and mobility during the 1960s and
1970s, and to some extent, the 1980s were promoted by the relative-
ly young age structure of the population resulting from the large
size of the baby-boom cohort born during 1946 to 1964. Given the
much smaller size of succeeding cohorts, the future seems likely to
bring patterns of reduced fertility, lower mobility, and increased
mortality.
Population Distribution
Knowledge of how a population is distributed is of critical
importance for understanding the distribution of population-related

Table
2.5:
Population
of
the
United
States,
Regions,
Divisions,
and
States,
1900-1990
United
States/
Population
(in
thousands)a
Regions
and
Divisions/
States
1990
1980
1970
1960
1950
1900
United
States
248,710
226,546
203,302
179,323
151,326
76,212
Regions
and
Divisions
Northeast
50,109
.9,135
.9,061
H,671
39,71
21,0.7
New
England
13,207
12,348
11,847
10,509
9,314
5,592
Middle
Atlantic
37,602
36,787
37,213
34,168
30,164
15,455
llidwest
59,669
51,166
56,590
51,619
H,61
26,333
East
North
Central
42,009
41,
682
40,263
36,225
30,399
15,986
West
North
Central
17,660
17,183
16,328
15,394
14,061
10,347
South
15,..6
75,372
62,113
5.,973
.7,197
2.,52.
South
Atlantic
43,567
36,959
30,679
25,972
21,182
10,443
East
South
Central
15,176
14,666
12,808
12,050
11,477
7,548
West
South
Central
26,703
23,747
19,326
16,951
14,538
6,532
West
52,716
.3,172
3.,131
21,053
20,
190
.,309
Mountain
13,659
ll,
373
8,290
6,855
5,075
1,675
Pacific
39,127
31,800
26,548
21,
198
15,
115
2,634
States
by
Division
New
England
Maine
1,228
1,125
994
969
914
694
New
Hampshire
1,
109
921
738
607
533
412
Vermont
563
511
445
390
378
344
Massachusetts
6,016
5,737
5,689
5,149
4,691
2,805
Rhode
Island
1,003
947
950
859
792
429
Connecticut
3,287
3,
108
3,032
2,535
2,007
908
(continues)
CJ.
CJ.

~
Table
2.5
(rontinued)
United
States/
Population
(in
thousands)a
Regions
and
Divisions/
States
1990
1980
1970
1960
1950
1900
lliddle
Atlantic
New
York
17,990
17,558
18,
241
16,782
14,830
7,269
New
Jersey
7,730
7,365
7,171
6,067
4,835
1,884
Pennsylvania
11,
882
11,864
11,801
11,319
10,498
6,302
Hast
North
Central
Ohio
10,847
10,798
10,657
9,706
7,947
4,
158
Indiana
5,544
5,490
5,195
4,662
3,934
2,516
Illinois
11,
431
11,427
11,
110
10,081
8,712
4,822
Michigan
9,295
9,262
8,882
7,823
6,372
2,421
Wisconsin
4,892
4,706
4,418
3,952
3,435
2,069
West
North
Central
Minnesota
4,375
4,076
3,806
3,414
2,982
1,751
Iowa
2,777
2,914
2,825
2,758
2,621
2,232
Missouri
5,
117
4,917
4,678
4,320
3,955
3,
107
North
Dakota
639
653
618
632
620
319
South
Dakota
696
691
666
681
653
402
Nebraska
1,578
1,570
1,485
1,411
1,326
1,066
Kansas
2,478
2,364
2,249
2,179
1,905
1,470
(amtinues)

Table
2.5
(amtinuetl)
United
States/
Population
(in
thousands)a
Regions
and
Divisions/
States
1990
1980
1970
1960
1950
1900
South
Atlantic
Delaware
666
594
548
446
318
185
Maryland
4,781
4,217
3,924
3,101
2,343
1,188
District
of
Columbia
607
638
757
764
802
279
Virginia
6,187
5,347
4,651
3,967
3,319
1,854
West
Virginia
1,793
1,950
1,744
1,860
2,006
959
North
Carolina
6,629
5,882
5,084
4,556
4,062
1,894
South
Carolina
3,487
3,122
2,591
2,383
2,
117
1,340
Georgia
6,478
5,463
4,588
3,943
3,445
2,216
Florida
12,938
9,746
6,791
4,952
2,771
529
Baal
South
Central
Kentucky
3,685
3,661
3,221
3,038
2,945
2,147
Tennessee
4,877
4,591
3,926
3,567
3,292
2,021
Alabama
4,041
3,894
3,444
3,267
3,062
1,829
Mississippi
2,573
2,521
2,217
2,178
2,179
1,551
Weal
South
Central
Arkansas
2,351
2,286
1,923
1,786
1,910
1,312
Louisiana
4,220
4,206
3,645
3,257
2,684
1,382
Oklahoma
3,
146
3,025
2,559
2,328
2,233
790
Texas
16,987
14,229
11,
199
9,580
7
,711
3,049
(continues)
~

Table
2.5
(amtinued)
United
States/
Population
(in
thousands)a
Regions
and
Divisions/
States
1990
1980
1970
1960
1950
1900
llo-tain
Montana
799
787
694
675
591
243
Idaho
1,007
944
713
667
589
162
Wyoming
454
470
332
330
291
93
Colorado
3,294
2,890
2,210
1,754
1,325
540
New
Mexico
1,515
1,303
1,017
951
681
195
Arizona
3,665
2,718
1,775
1,302
750
123
Utah
1,723
1,461
1,059
891
689
277
Nevada
1,202
800
489
285
160
42
Pacific
Washington
4,867
4,132
3,413
2,853
2,379
518
Oregon
2,842
2,633
2,092
1,769
1,521
414
Ca
Ii
forni
a
29,760
23,668
19,971
15,717
10,586
1,485
Alaska
550
402
303
226
129
64
Hawaii
1,108
965
770
633
500
154
~otals
shown
are
derived
from
unrounded
values.
Source:
United
States
Department
of
Commerce,
Bureau
of
the
Census.
PopuJatton
Trends
and
Congressional
Appor-
tionment,•
1990
Census
Profile
No.
1,
Washington,
DC:
U.S.
Government
Printing
Office,
March,
1991.
~

Table
2.6:
Population
Change
in
the
United
States,
Regions
and
Divisions,
1960-1990
Change
in
Population
Number
(in
thousands)
Percent
1980
1970
1960
1980
1970
1960
United
States/
to
to
to
to
to
to
Regions
and
Divisions
1990
1980
1970
1990
1980
1970
-
United
States
22,16'
23,2·H
23,979
9.1
11.f
13.t
Kegions
and
Divisions
Northeast
1,674'
75
f,313
3.f
0.2
9.1
New
England
858
501
1,338
7.0
4.2
12.7
Middle
Atlantic
815
-426
3,045
2.2
-1.
1
8.9
Midwest
803
2,275
t,971
1.f
4.0
9.6
East
North
Central
327
1,419
4,038
0.8
3.5
11.
1
West
North
Central
476
856
933
2.8
5.2
6.1
South
10,074'
12,559
7,140
13.f
20.0
14.3
South
Atlantic
6,608
6,280
4,707
17.9
20.5
18.1
East
South
Central
510
1,858
758
3.5
14.5
6.3
West
South
Central
2,956
4,421
2,375
12.4
22.9
14.0
West
9,614'
1,334
6,715
22.3
23.9
24.2
Moun
ta
in
2,286
3,083
1,435
20.1
37.2
20.9
Pacific
7,328
5,251
5,350
23.0
19.8
25.2
Source:
Population
Trends
and
Congressional
Apportionment,•
1990
Census
Profile
No.
1,
U.S.
Bureau
of
the
Census,
Washington,
DC:
U.S.
Government
Printing
Office,
March,
1991.
()
'I

38
effects. Within the United States, population redistribution has been
nearly a continuous process since the founding of the Nation. As is
evident in Tables 2.5 and 2.6, recent decades have brought patterns
of more rapid growth and inmigration to the western and southern
parts of the United States and reduced growth and outmigration
from the northeastern and midwestern parts of the United States.
During the 1970s, the population of the Northeast increased by only
0.2 percent, the population of the Midwest by 4.0 percent, the
South's by 20.0 percent, and the West's by 23.9 percent. Similarly
in the 1980s, the population of the Northeast increased by 3.4
percent, that in the Midwest by 1.4 percent, that in the South by
13.4 percent, and that in the West by 22.3 percent. The growth of
the West has been particularly dramatic, with its population increas-
ing from 4.3 million persons in 1900 to 52.8 million in 1990--an
increase of more than 1,100 percent.
A few states have played a major role in recent patterns of
population growth. California, Texas, and Florida together account-
ed for 42 percent of all population growth in the United States from
1970 to 1980 and for 54 percent of all growth from 1980 to 1990. By
1990 nearly 12 percent of all Americans lived in California, and
California, New York, Texas, and Florida together were the homes
of nearly 1 out of every 3 persons in the United States.
Tables 2.7 and 2.8 present data on the distribution of the
population according to two other widely used geographical catego-
ries. The data in Table 2.7 show how the population of the Nation
has increasingly shifted from rural to urban residences, from 44
percent of persons living in rural areas and nearly 25 percent living
on farms in 1940 to 26 percent living in rural areas and only 2.5
percent living on farms in 1980. Similarly, the proportion of non-
metropolitan residents has declined from 44 percent of the popula-
tion in 1950 to less than 23 percent in 1990 (Table 2.8). Oearly, the
distribution of the population of the United States has changed
substantially during the past half century.
Age and Sex Characteristics
The age and sex composition of the population affects the
demand for goods and services by affecting the level and types of
demands of the population. Tables 2.9 through 2.11 provide data on
these characteristics for the population of the United States.

Table
2.7:
Population
and
Percentage
of
Population
in
the
United
States
by
Urban,
Rural,
Rural
Farm,
and
Rural
Nonfarm
Residence,
1930-1980
Population
Percentage
of
Population
Total
Rural
Rural
Rural
Rural
Year
Population
Urban
Rural
Farm
Non
farm
Urban
Rural
Farm
Non
farm
1930
122,775,046
68,954,823
53,820,223
30,157,513
23,662,710
56.2
43.8
24.5
19.3
1940
131,669,275
74,423,702
57,245,573
30,216,188
27,029,385
56.5
43.5
22.9
20.6
1950a
150,697,361
96,467,686
54,229,675
23,048,350
31,181,325
64.0
36.0
15.3
20.7
1960b
178,466,732
124,714,055
53,752,677
13,431,791
40,320,886
69.9
30.1
7.5
22.6
1970
203,212,877
149,334,020
53,878,857
10,588,534
43,290,323
73.5
26.5
5.2
21.
3
1980c
226,545,805
167,054,638
59,491,167
5,617,903
53,873,264
73.7
26.3
2.5
23.8
a1950
census
definitions
of
urban-rural
and
rural
farm
and
nonfarm.
The
total
population
as
reported
here
for
1950
ls
different
than
in
previous
tables
because
data
in
previous
tables
reflect
post-1950
corrections
while
data
on
rural
and
urban
populations
are
available
only
for
the
count
values
shown
here.
b1960
census
definitions
of
urban-rural
and
rural
farm
and
nonfarm.
c1980
census
definitions
of
urban-rural
and
rural
farm
and
nonfarm.
Source:
Data
were
obtained
from
the
U.S.
Census
of
Population
and
Housing
(United
States
Department
of
Commerce,
Bureau
of
the
Census,
1930-1980).
~

Table
2.8:
Proportion
of
U.S.
Population
that
is
Metropolitan
and
Nonmetropolitan,
1950-1990
1950
1960
1970
1980
1990
Metropolitan
56.1
63.0
68.6
74.8
77.5
Nonmetropolitan
43.9
37.0
31.4
25.2
22.5
Source:
Metropolitan
and
Nonmetropolitan
Values
for
1950-1980
from
Bogue,
D.J.
The
Populatkm
of
tire
United
States,
New
York:
Free
Press,
1985.
Values
for
1990
from
U.S.
Bureau
of
the
Census
Press
Release
CB-91-66,
Washington,
OC,
April
1991.
~

Table 2.9: Median Age and the Sex Ratio In the
United States, 1900-1990
Year Median Age Sex Ratio
1900 22.9 104.4
1910 24.1 106.0
1920 25.3 104.0
1930 26.5 102.5
1940 29.0 100.7
1950 30.1 98.6
1960 29.5 97.1
1970 28.1 94.8
1980 30.0 94.5
1990 32.9 95.1
Source: From United States Department of
Commerce, Bureau of the Census. Charac-
teristics of the Population,• U.S. Census of
Population 1980, Chapter 3, General Popula-
tion Characteristics PCB0-1-81 (United States)
Washington, DC: U.S. Government Printing
Office, 1983. Values for 1990 from STFlA
for the United States.
41

~
Table
2.10:
Population
of
the
United
States
by
Age
and
Sex,
1940-1989
Under
5-9
10-14
15-19
Year
Total
5
years
years
years
years
Total
1940a
131,669,275
10,541,524
10,684,622
11,745,935
12,333,523
1950b
150,697,361
16,163,571
13,199,685
11,119,268
10,616,598
1960c
179,323,175
20,320,901
18,691,780
16,773,492
13,219,243
1970
203,
211,
926
17,154,337
19,956,247
20,789,468
19,070,348
1980
226,545,805
16,348,254
16,699,956
18,242,129
21,168,124
1989
248,239,000
18,752,000
18,212,000
16,950,000
17,812,000
Male
1940a
66,061,592
5,354,808
5,418,823
5,952,329
6,180,153
1950b
74,833,239
8,236,164
6,714,555
5,660,399
5,311,342
1960c
88,331,494
10,329,729
9,504,368
8,524,289
6,633,661
1970
98,912,192
8,745,499
10,168,496
10,590,737
9,633,847
1980
110,053,
161
8,362,009
8,539,080
9,316,221
10,755,409
1989
120,982,000
9,598,000
9,321,000
8,689,000
9,091,000
Female
1940a
65,607,683
5,186,716
5,265,799
5,793,606
6,153,370
1950b
75,864,122
7,927,407
6,485,130
5,458,869
5,305,256
1960c
90,991,681
9,991,172
9,187,412
8,249,203
6,585,582
1970
104,299,734
8,408,838
9,787,751
10,198,731
9,436,501
1980
116,492,644
7,986,245
8,160,876
8,925,908
10,412,715
1989
127,258,000
9,155,000
8,891,000
8,260,000
8,721,000
(amtinues)

Table
2.10
(amtinued)
20-24
25-29
30-34
35-39
40-44
Year
years
years
years
years
years
Total
1940a
11,587
,835
11,096,638
10,242,388
9,545,377
8,787,843
1950b
11,481,828
12,242,260
11,517,007
11,246,386
10,203,973
1960c
10,800,761
10,869,124
11,949,186
12,481,109
11,600,
243
1970
16,371,021
13,476,993
11,430,436
11,
106,851
11,980,954
1980
21,318,704
19,520,919
17,560,920
13,965,302
11,669,408
1989
18,702,000
21,699,000
22,135,000
19,6:al,OOO
16,882,000
Male
1940a
5,692,392
5,450,662
5,070,312
4,745,659
4,419,
135
1950b
5,606,293
5,972,078
5,624,723
5,517,544
5,070,269
1960c
5,272,340
5,333,075
5,846,224
6,079,512
5,675,881
1970
7,917,269
6,621,567
5,595,790
5,412,423
5,818,813
1980
10,663,231
9,705,107
8,676,796
6,861,509
5,708,210
1989
9,368,000
10,865,000
11,
078,000
9,731,000
8,294,000
Female
1940a
5,895,443
5,
645,
916
5,172,076
4,799,718
4,368,708
1950b
5,875,535
6,270,182
5,892,284
5,728,842
5,133,704
1960c
5,528,421
5,536,049
6,102,962
6,401,597
5,924,362
1970
8,453,752
6,855,426
5,834,646
5,694,428
6,162,141
1980
10,655,473
9,815,812
8,884,124
7,103,793
5,961,198
1989
9,334,000
10,834,000
11,
058,
000
9,890,000
8,588,000
(amtinues)
....
(JJ

Table
2.10
(amtinued)
45-49
50-54
55-59
60-64
65
years
Year
years
years
years
years
and
over
Total
1940a
8,255,225
7,256,846
5,843,865
4,728,340
9,019,314
1950b
9,070,465
8,272,188
7,235,120
6,059,475
12,269,537
1960c
10,879,485
9,605,954
8,429,865
7,142,452
16,559,580
1970
12,
115,939
11,
104,018
9,973,028
8,616,784
20,065,502
1980
11,089,755
11,710,032
11,615,254
10,087,621
25,549,427
1989
13,521,000
11,375,
000
10,726,000
10,867,000
30,984,000
Male
1940a
4,209,269
3,752,750
3,011,364
2,397,816
4,406,120
1950b
4,526,366
4,
128,648
3,630,046
3,037,838
5,796,974
1960c
5,357,925
4,734,829
4,
127,
245
3,409,319
7,503,097
1970
5,851,334
5,347,916
4,765,821
4,026,972
8,415,708
1980
5,388,249
5,620,670
5,481,863
4,669,892
10,304,915
1989
6,601,000
5,509,000
5,121,000
5,079,000
12,636,000
Female
1940a
4,045,956
3,504,096
2,832,501
2,330,524
4,613,194
1950b
4,544,099
4,143,540
3,605,074
3,021,637
6,472,563
1960c
5,521,560
4,871,125
4,302,620
3,733,
133
9,056,483
1970
6,264,605
5,756,102
5,207,207
4,589,812
11,
649,
794
1980
5,701,506
6,089,362
6,133,391
5,417,729
15,244,512
1989
6,920,000
5,866,000
5,605,000
5,788,000
18,348,000
a
The
total
Jipulation
for
1950
shown
here
ls
different
than
that
shown
in
previous
tables
because
the
data
in
previous
tables
reflect
post-1950
corrections
while
data
on
population
by
age
and
sex
are
available
only
for
the
count
values
shown
here.
b
Denotes
first
year
for
which
figures
include
Alaska
and
Hawaii.
c
Excludes
23,372
persons
for
whom
age
ls
not
available.
Source:
Data
for
1940
throug!!
1980
from
the
decennial
censuses
for
each
period.
Data
for
1989
from
United
States
De-
partment
of
Commerce,
Bureau
of
the
Census.
Current
Population
Reports
P-25,
No.
1057,
Washington
DC:
U.S.
Government
Printing
Office,
1990.
,,..
,,..

Table
2.11:
Pen:ent
of
the
Population
by
Age
Groups
In
the
United
States,
1940-1989
Year
5yrs
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65+
1940
8.0
8.1
8.9
9.4
8.8
8.4
7.8
7.3
6.7
6.3
5.5
4.3
3.6
6.9
1950
10.7
8.8
7.4
7.0
7.7
8.1
7.6
7.5
6.8
6.0
5.5
4.8
4.0
8.1
1960
11.3
10.4
9.4
7.3
6.0
6.1
6.7
7.0
6.5
6.
1
5.4
4.6
4.0
9.2
1970
8.4
9.8
10.2
9.4
8.
1
6.6
5.6
5.5
5.9
6.0
5.5
4.9
4.2
9.9
1980
7.2
7.4
8.1
9.2
9.4
8.6
7.7
6.2
5.2
4.9
5.2
5.
1
4.5
11.3
1989
7.6
7.3
6.8
7.2
7.5
8.7
8.9
7.9
6.8
5.4
4.6
4.4
4.4
12.5
Source:
Data
for
1940
through
1980
from
the
decennial
censuses
for
each
period.
Data
for
1989
from
United
States
De-
partment
of
Commerce,
Bureau
of
the
Census.
Current
Population
Reports
P-25,
No.
1057,
Washington
DC:
U.S.
Government
Printing
Office,
1990.
~

46
The data in Table 2.9 point to several key dimensions of age and
sex differences. During the period from 1900 to 1990, the median
age increased rapidly such that by 1990 the median age was 10
years older than in 1900. Similarly, changes in the sex ratio (the
number of males per 100 females) reflect an aging population and
changes in immigration patterns. Males predominate at younger
ages with about 105 males being born per 100 females, but because
survival rates for females are higher than those for males, as a
population ages the ratio of males to females declines. Thus,
the decline in the sex ratio from 104 in 1900 to 95 in 1990 reflects
an aging population base in which the lower mortality among
females leads to the number of females decreasing less rapidly than
the number of males. It may also reflect the disproportionate
number of males among immigrants to the United States during the
last decades of the nineteenth and first decades of the twentieth
centuries.
Tables 2.10 and 2.11 provide further information on the age
and sex composition of the population. An analysis of the data in
these tables show several patterns of significance. First, it is evident
that for each time period the number of males exceeds the number
of females at the younger ages; but, between the ages of 20 and 30,
the number of females comes to exceed the number of males, and by
age 65, the number of females exceeds the number of males by 40 to
50 percent. The data in these tables also demonstrate the continued
aging of the U.S. population, particularly since the 1960s. Whereas
over 38 percent of the population was less than 20 years of age and
only 9 percent was 65 years of age or older in 1960, by 1989, the
percentage under 20 years of age was 28 percent and the percent 65
years of age or older was 12.5 percent.
Racial and Ethnic Characteristics
Race and ethnicity are characteristics that have historically
been closely related to access to socioeconomic resources, with those
from minority groups having substantially lower levels of access and
smaller resource bases than majority group members. Tables 2.12
through 2.14 present information on the racial and ethnic character-
istics of the population of the United States for recent periods. In
examining these data, it is critical to recognize that race and ethnici-
ty as measured by the census questionnaire are two separate items
(see the discussion above). The data in the top panel of Table 2.12

Table
2.12:
U.S.
Population,
1970,
1980,
and
1990,
Percent
Change
In
Population
1970
to
1980
and
1980to1990,
and
Proportion
of
Population
1970,
1980,
and
1990
by
Race,
Hispanic
origin,
and
Ethnidty
Percent
Proportion
of
Racia
1
I
Number
Change
Population
Ethnic
Category
1970
1980
1990
1970-80
1980-90
1970
1980
1990
Race
and
Hispanic
Origin
White
178,107,190
189,035,012
199,686,070
6.
1
5.6
87.6
83.4
80.3
Black
22,549,815
26,482,349
29,986,060
17.4
13.2
11.
1
11.
7
12.1
Other
2,555,872
11,028,444
19,037,743
331.5
'
72.6
1.
3
4.9
7.6
Hispanic•
9,294,509
14,603,683
22,354,059
57.1
53.1
4.6
6.5
9.0
Total
203,212,877
226,545,805
248,709,873
11.5
9.8
Ethnicity
Ang
lob
168,812,682
180,602,838
188,128,296
7.0
4.2
83.1
79.7
75.7
Black
22,549,815
26,091,857
29,216,293
15
..
7
12.0
11.
1
11.5
11.
7
Hispa~ic
9,294,509
14,603,683
22,354,059
57.1
53.1
4.6
6.5
9.0
Other
2,555,872
5,247,427
9,011,225
105.3
71.
7
1.
2
2.3
3.6
Total
203,212,877
226,545,805
248,709,873
11.5
9.8
100.0
100.0
100.0
aHispanlcs
may
be
of
any
race.
As
a
result,
white,
black
and
other
sum
to
the
total
population
and
Hispanics
are
Included
among
those
In
these
three
radal
categories
as
well
as
being
shown
as
a
separate
ethnic
group.
~or
1970,
Spanish-surnamed
persons
were
assumed
to
be
Hispanic
and
all
Hispanics
were
subtracted
from
the
white
total
to
obtain
Anglos.
The
values
shown
for
the
black
and
other
populations
for
1980
and
1990
In
this
table
do
not
Include
blacks
or
persons
of
other
races
who
are
of
Hispanic
origin.
Source:
U.S.
Bureau
of
the
Census,
U.S.
Census
of
Population,
fourth
count
census
tapes
for
1970
and
the
PL94-171
tape
files
for
1980
and
1990.
,,,.
'
I

Table
2.13:
Percent
Distribution
of
the
Resident
Population
of
the
United
States
by
Regions
and
for
the
Ten
Largest
~
States
by
Race
and
Hispanic
Origin,
1990
American
Asian
United
States/
Indian,
or
Regions/
Eskimo
or
Pacific
Other
Hispanic
States
Total
White
Black
Aleut
Islander
Race
Origin•
United
States
100.0
100.0
100.0
100.0
100.0
100.0
100.0
R.egions
Northeast
20.4
21.1
18.7
6.4
18.4
17.0
16.8
Midwest
24.0
26.1
19.1
17.3
10.6
8.4
7.7
South
34.4
32.8
52.8
28.7
15.4
24.0
30.3
West
21.
2
20.0
9.4
47.6
55.6
50.6
45.2
States
California
12.0
10.3
7.4
12.4
39.1
40.2
34.4
New
York
7.2
6.7
9.5
3.2
9.5
JO.I
9.9
Texas
6.8
6.4
6.7
3.4
4.4
18.4
19.4
Florida
5.2
5.4
5.9
1.9
2.1
2.4
7.0
Pennsylvania
4.8
5.3
3.6
0.8
1.9
I.
2
1.0
111
ino
is
4.6
4.5
5.7
I.
I
3.9
4.9
4.0
Ohio
4.4
4.8
3.9
I.
0
1.3
0.6
0.6
Michigan
3.7
3.9
4.3
2.8
1.4
0.9
0.9
New
Jersey
3.
1
3.
I
3.5
0.8
3.7
2.8
3.3
North
Carolina
2.7
2.5
4.9
4.1
0.7
0.3
0.3
aPersons
of
Hispanic
origin
can
be
of
any
race.
Souru:
Census
Bureau
Completes
Distribution
of
1990
Redistricting
Tabulations
to
States,•
U.S.
Bureau
of
the
Census
Press
Release
CB91-100,
March
11,
1991.

Table
2.14:
Percent
Distribution
of
the
Resident
Population
of
the
United
States,
Regions
and
States,
by
Race
and
Hispanic
Origin,
1990
American
Asian
United
States/
Indian,
or
Regions/
Eskimo
or
Pacific
Other
Hispanic
States
Total
White
Black
Aleut
Islander
Race
Origin
United
States
100.0
10.3
12.1
0.1
2.9
3.9
9.0
llortheast
100.0
12.1
11.0
0.2
2.6
3.3
7
.4:
Connecticut
100
.
.
0
87.0
8.3
0.2
1.
5
2.9
6.5
Maine
100.0
98.4
0.4
0.5
0.5
0.1
0.6
Massachusetts
100.0
89.8
5.0
0.2
2.4
2.6
4.8
New
Hampshire
100.0
98.0
0.6
0.2
0.8
0.3
1.0
New
Jersey
100.0
79.3
13.4
0.2
3.5
3.6
9.6
New
York
100.0
74.4
15.9
0.3
3.9
5.5
12.3
Pennsylvania
100.0
88.5
9.2
0
.,1
1.
2
1.0
2.0
Rhode
Island
100.0
91.4
3.9
0.4
1.8
2.5
4.6
Vermont
100.0
98.6
0.3
0.3
0.6
0.1
0.7
llidwest
100.0
17.2
9.6
0.6
1.3
1.4:
2.9
I
11
i
no
is
100.0
78.3
14.8
0.2
2.5
4.2
7.9
Indiana
100.0
90.6
7.8
0.2
0.7
0.7
1.8
Iowa
100.0
96.6
1.
7
0.3
0.9
0.5
1.
2
Kansas
100.0
90.1
5.8
0.9
1.
3
2.0
3.8
Michigan
100.0
83.4
13.9
0.6
1.
1
0.9
2.2
Minnesota
100.0
94.4
2.2
1.
1
1.
8
0.5
1.2
Missouri
100.0
87.7
10.7
0.4
0.8
0.4
1.2
Nebraska
100.
0
93.8
3.6
0.8
0.8
1.0
2.3
North
Dakota
100.0
94.6
0.6
4.1
0.5
0.3
0.7
Ohio
100.0
87.8
10.6
0.2
0.8
0.5
1.3
South
Dakota
100.0
91.6
0.5
7.3
0.4
0.2
0.8
Wisconsin
100.0
92.2
5.0
0.8
1.1
0.9
1.9
(continues)
tt

01
0
Table
2.14
(amtinued)
American
Asian
United
States/
Indian,
or
Regions/
Eskimo
or
Pacific
Other
Hispanic
States
Total
White
Black
Aleut
Islander
Race
Origina
South
100.0
76.1
11.5
0.7
1.3
2.1
7.9
Alabama
100.0
73.6
25.3
0.4
0.5
0.1
0.6
Arkansas
100.0
82.7
15.9
0.5
0.5
0.3
0.8
Delaware
100.0
80.3
16.9
0.3
1.4
1.
1
2.4
Florida
100.0
83.1
13.6
0.3
1.
2
1.
8
12.2
Georgia
100.0
71.0
27.0
0.2
1.
2
0.7
1.7
Kentucky
100.0
92.0
7.1
0.2
0.5
0.2
0.6
Louisiana
100.0
67.3
30.8
0.4
1.0
0.5
2.2
Maryland
100.0
71.0
24.
9
0.3
2.9
0.9
2.6
Mi
s
5
i
SS
i
pp
i
100.0
63.5
35.6
0.3
0.5
0.1
0.6
North
Carolina
100.0
75.6
22.0
1.
2
0.8
0.5
1.
2
Oklahoma
100.0
82.1
7.4
8.0
1.
1
1.
3
2.7
South
Carolina
100.0
69.0
29.8
0.2
0.6
0.3
0.9
Tennessee
100.0
83.0
16.0
0.2
0.7
0.2
0.7
Texas
100.0
75.2
11.9
0.4
1.
9
10.6
25.5
Virginia
100.0
77.4
18.8
0.2
2.6
0.9
2.6
West
Virginia
100.0
96.2
3.
1
0.1
0.4
0.1
0.5
(mntinues)

Table
2.14
(continued)
American
Asian
United
States/
Indian,
or
Regions/
Eskimo
or
Pacific
Other
Hispanic
States
Total
White
Black
Aleut
Islander
Race
Origin
We•t
100.0
75.1
5.-
1.1
7.7
9.-
19.J
A
I
a
ska
100.0
75.5
4.
1
15.6
3.6
1.
2
3.2
Arizona
100.0
80.8
3.0
5.6
1.
5
9.1
18.8
California
100.0
69.0
7.4
0.8
9.6
13.2
25.8
Colorado
100.0
88.2
4.0
0.8
1.
8
5.1
12.9
Hawaii
100.0
33.4
2.5
0.5
61.
8
1.
9
7.3
Idaho
100.0
94.4
0.3
1.4
0.9
3.0
5.3
Montana
100.0
92.7
0.3
6.0
0.5
0.5
1.
5
Nevada
100.0
84.3
6.6
1.6
3.2
4.4
10.4
New
Mexico
100.0
75.6
2.0
8.9
0.9
12.6
38.2
Oregon
100.0
92.8
1.
6
1.4
2.4
1.8
4.0
Utah
100.0
93.8
0.7
1.4
1.
9
2.2
4.9
Washington
100.0
88.5
3.1
1.
7
4.3
2.4
4.4
Wyoming
100.0
94.2
0.8
2.
1
0.6
2.3
5.7
aPersons
of
Hispanic
origin
can
be
of
any
race.
Source:
·eensus
Bureau
Completes
Distribution
of
1990
Redistricting
Tabulations
to
States,•
U.S.
Bureau
of
the
Census
Press
Re-
lease
CB91-100,
March
11,
1991.
01
.....

52
show the number of persons by race (white, black, and other) and
by Spanish-origin (who can be of any race and are thus also includ-
ed in the data by race) so that the values do not sum to 100 percent
of the population. In the bottom panel, four mutually exclusive
groups have been derived by subtracting the Spanish-origin popula-
tion by race from the total population by race.
The data in Table 2.12 show that minority population growth
has been substantially faster than that of the white and Anglo major-
ity groups and that the proportion of the population consisting of
minority group members has increased substantially in recent
decades. The proportion composed of black Americans has shown a
moderate increase, but the proportion composed of other groups has
increased by roughly six times and the proportion of the population
that is Spanish-origin (Hispanic) has doubled between 1970 and
1990. By 1990, approximately one-fourth of the U.S. population was
of minority status. In addition, a majority of the net growth in the
U.S. population in the period from 1970 to 1990 (58% of all growth)
was a result of increases in non-Anglo population groups.
The data in Tables 2.13 and 2.14 show those states and
regions in which the largest relative proportions and the largest
number of persons in different racial and ethnic groups are located.
These data indicate that the four largest states of California, New
York, Texas, and Florida accounted for 31 percent of the total popu-
lation, 30 percent of the black population, nearly 71 percent of the
Hispanic population, and 55 percent of the Asian population in
1990. California alone was the area of residence for 34 percent of all
Hispanics, 39 percent of Asians, 7 percent of blacks, and was the
residence of nearly l-in-8 of all Americans in 1990. The South was
the area of residence for nearly 53 percent of all black Americans,
the West the home of nearly 56 percent of all Asians and of more
than 45 percent of all Hispanics. Together, the South and West
were the areas of residence for 62 percent of blacks, 76 percent of
American Indians, 71 percent of Asians, and 75 percent of all His-
panics in 1990. Clearly, then, the western and southern regions of
the United States are the major regions of residence of America's
minority populations.
As noted above, minorities have distinct characteristics that
affect their socioeconomic and other resources. Some of the charac-
teristics of minority groups which affect their life chances are shown
in Table 2.15. The data in this table clearly show minorities to be
younger than the white population and to have higher fertility and
lower life expectancies than nonminority populations. Their family
sizes are generally larger and the proportion of their households that

Table
2.15:
Demographic
and
Socioeconomic
Characteristics
of
the
Popu1atlon
of
the
United
States
by
Race/Ethnicity
for
Selected
Years
Racia
1
Group
Ethnicity
Total
Characteristic
Population
White
Black
Other
Hispanic
Median
Age
(in
years)
(1988)
32.3
33.2
27.5
28.8
25,8
Sex
Ratio
(1988)
95.0
95.7
90.2
95.7
101.
2
Fertility
Crude
Fertility
Rate
(1987)
15.7
14.5
21.
2
(a)
(a)
Total
Fertility
Rate
(1987)
1.87
1.77
2.35
(a)
(a)
Life
Expectancy
at
Birth
(in
yrs.)
(1988)
74.
9
75.5
71.5
(a)
(a)
Household
Characteristics
(1990)
Average
Household
Size
2.63
2.58
2.88
(a)
3.47
Percent
of
Family
Households
70.8
70.5
71.
2
75.2
81.
6
Percent
with
.Fem
a
1
e
Householder
28.4
26.1
47.5
23.6
27.3
(continues)
01
(JJ

Table
2.15
(continued)
Characteristic
Income
and
Poverty
(1989)
Median
Household
Income
(1989)
Poverty
Status
(1989)
Percent
of
Persons
in
Poverty
Percent
of
Families
in
Poverty
Median
Years
of
Education
(a):
Data
not
available.
(1988)
Total
Population
$28,906
12.8
10.3
12.7
Racial
Group
Ethnicity
White
Black
Other
Hispanic
$30,406
$18,083
(a)
$21,921
10.0
30.7
(a)
26.2
7.8
27.8
(a)
23.4
12.7
12.4
(a)
12.0
Scmu:
United
States
Department
of
Commerce,
Bureau
of
the
Census.
U.S.
PopuJation
Estimates
by
Age,
Sex,
Race
and
Hispanic
Origin:
1989,
•
Cumnt
Population
Reports
P-25,
No.
1057,
Washington,
DC:
U.
S.
Government
Printing
Office,
1990.
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Money,
Income
and
Poverty
Status
in
the
United
States,
1989,
•
Cumnt
Population
Reports
P-60,
No.
168,
Washington,
DC:
U.S.
Government
Printing
Office,
1990.
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Household
and
Family
Characteristics:
March,
1990
and
1989,
•
Cumnt
Population
Reports
P-20,
No.
447,
Washington,
DC:
U.S.
Government
Printing
Office,
1990.
United
States
Department
of
Commerce,
Bureau
of
the
Census.
The
Hispanic
PopuJation
in
the
United
States:
March,
1990,
•
Current
Population
Report
P-20,
No.
449,
Washington,
DC:
U.S.
Government
Printing
Office,
1991.
~

55
tend to be of nontraditional forms is higher. Minority incomes tend
to be about 60 percent of those of nonminority groups, while their
levels of poverty are two to three times those of nonminorities.
They also tend to have substantially lower levels of education. The
data in Table 2.15 suggest that minorities have demographic and
socioeconomic characteristics that are likely to give them unique
client and consumer characteristics.
Marital and Household Characteristics
The characteristics of the population relative to marital status
and household characteristics play a major role in determining the
demand characteristics of the primary consuming units in the United
States--households and families. Tables 2.16 through 2.19 provide
data showing changes in marital status, family and nonfamily
household status, rates of cohabitation, households by number of
persons, and average household size for recent periods.
_ The data in Tables 2.16 through 2.19 show an increasing
proportion of persons living in nontraditional marital and household
arrangements and decreases in the size of households. The data in
Table 2.16 show an increase in the proportion divorced from 3.3
percent in 1970 to 7.2 percent in 1988 and decreases in the propor-
tion of persons married. Similarly the data in Table 2.17 show an 11
percent decline in the proportion of family households (from about
81% to 70%) and a corresponding increase in nonfamily households
(from 19% to 30%) during the period from 1970 to 1990. Equally
apparent is the fact that within family households, growth has been
slowest among the married-couple family type. For example, from
1970 to 1980 the number of married-couple households increased by
less than 10 percent while households with male and female house-
holders increased by 41and58 percent, respectively. Similarly, in
the 1980s, married-couple households increased by only 3.2 percent,
while households with male and female householders increased by
84 and 22 percent, respectively. The aging of the large baby-boom
generation out of the initial household formation ages into the more
stable middle-ages led to substantially smaller overall rates of in-
crease in the number of households in the 1980s than in the 1970s.
However, traditional household types continued to show dispropor-
tionately low rates of growth during the 1980s such that married-
couple households accounted for only 55 percent of all households
by 1990.
Table 2.18 provides data which suggest that nontraditional
arrangements have been increasing proportionately. These data

Table
2.16:
Marital
Status
of
the
PopuJation
of
the
United
States,
1970-1988
1970
1980
1988
Marital
Status
Number
~
Number
~
Number
~
Never
Married
38,051,042
25.5
43,236,000
25.1
49,496,000
25.7
(Single)
Divorced
4,930,875
3.3
9,711,000
5.7
13,968,000
7.2
Separated
5,588,426
3.7
3,920,000
2.3
4,458,000
2.3
Widowed
11,746,212
7.9
12,451,000
7.2
13,532,000
7.0
Married
89,079,417
59.6
102,800,000
59.7
111,456,000
57.8
Total
149
t
395
I
972
100.0
172,
118,000
100.0
192,910,000
100.0
Source:
Data
on
married,
widowed,
divorced,
separated,
and
sJngle
(never
married)
for
1970
from
the
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Historical
Statistics
of
the
United
States,
Colonial
Times
to
1970,
Part
1
and
Part
2,
Washington,
DC:
U.S.
Government
Printing
Office,
1975
and
data
for
1980
and
1988
from
the
United
States
Department
of
Commerce,
Bureau
of
the
Census,
Marital
Status
and
Living
Arrangements:
March,
1988,
Current
Population
Reports
P-20,
No.
433,
Washington
DC:
U.S.
Government
Printing
Office,
1989.
01
°'

Table
2.17:
Households
in
the
United
States
by
Type,
1970-1990
Year
Type
of
Percent
Change
Household
1970
1980
1990
1970-80
1980-90
Tota
1
(thousands)
63,401
80,776
91,947
27.4
13.8
Percent
Family
81.
2
73.7
70.2
15.7
8.4
Married
couple
70.6
60.8
55.1
9.8
3.2
Male
only
1.
9
2.1
3.4
U.1
84.3
Female
only
8.7
10.8
11.6
58.3
22.3
Percent
Nonfamily
18.8
26.3
29.8
77.0
29.0
Source:
Data
for
1970
and
1980
from
the
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Household
and
Family
Olaracteristlcs:
March
1990
and
1989,
Current
Population
Reports
P-20,
No.
447,
Washington,
DC:
U.S.
Government
Printing
Office,
1990.
Data
for
1990
from
the
STF1A
data
for
the
United
States.
(JI
~

Table
2.18:
Estimates
of
Cohabitation
and
Marriage
Before
the
Age
of
25
by
Age
Cohort
In
1988
Males
(J)
Females
(J)
-
Age
Cohabit
Marry
Union•
Cohabit
Marry
Union•
25-29
33
38
67
37
61
76
30-34
29
51
66
26
67
76
35-39
24
55
68
16
72
78
40-U
11
66
70
7
79
82
45-49
8
68
70
3
82
83
~e
proportion
In
unions
will
not
equal
the
sum
of
cohabiting
and
married
persons
because
persons
may
have
both
cohabitated
and
been
married
by
age
25.
Source:
Bumpass,
L.L.
and
Sweet,
J.A.
National
Estimates
of
Cohabitation,•
Demography
26:615-625,
1989.
g:

Table
2.19:
Nwnber
and
Percent
of
Households
by
Persons
Jn
the
Household
and
Average
Household
Size
for
the
United
States,
1940-1990
(numbers
Jn
thousands)
Households
1940
1950
1960
1970
1980
1990
by
Persons
Per
Household
Number
~
Number
~
Number
~
Number
~
Number
~
Number
~
One
person
2,481
7
.1
4,737
10.9
6,871
13.0
10,692
17.1
18,300
22.6
22,999
24.7
Two
persons
8,667
24.8
12,529
28.8
14,616
27.8
18,129
28.8
25,300
31.3
30,
114
32.3
Three
persons
7,829
22.4
9,808
22.6
9,941
19.0
10,903
17.3
14,100
17.5
16,128
17.3
Four
persons
6,326
18.
1
7,729
17.8
9,277
17.6
9,935
15.8
12,700
15.7
14,456
15.4
Five
or
more
9,646
27.6
8,666
19.9
11,905
22.6
13,215
21.
0
10,400
12.9
9,651
10.3
All
Households
34,949
100.0
43,469
100.0
52,610
100.0
62,874
100.0
80,800
100.0
93,348
100.0
Average
Persons
Per
Household
3.67
-
3.37
-
3.35
-
3
.17
-
2.75
-
2.63
100.0
Sourrt:
Bogue,
D.J.
The
Population
of
the
United
States:
Historical
Trends
and
Future
Projections
New
York:
The
Free
Press,
1985.
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Historical
Statistics
of
the
United
States
Colonial
Times
to
1970,
Washington,
DC:
U.S.
Government
Printing
Office,
1975.
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Household
and
Family
Characteris-
tics:
March
1990
and
1989,
•Current
Population
Reports
P-20,
No.
447,
Washington,
DC:
U.S.
Government
Printing
Office,
1990.

60
show that for males who were 25 to 29 years of age at the time of
the survey in 1988, nearly as many had cohabitated as had married
by the age of 25. Among those males who were 25 years old in
1940 to 1944 (who were 45 to 49 in 1988), the proportion married at
the age of 25 was more than 8 times the proportion cohabitating.
Also of interest in this table are the data showing that the propor-
tion of males in unions has changed relatively little over the years
represented by these cohorts (1965 through 1988). These data
suggest that it is the form of unions, not the tendency to be in
unions, that has changed over the past few decades.
Table 2.19 presents data on the number and proportion of
households by size from 1940 through 1990. It is evident that the
average size of households has declined from 1940 through 1990 as a
result of both a decline in the number of larger households and an
increase in the number of smaller households. From 1940 to 1990,
the proportion of one- and two-person households increased from 32
percent to 57 percent of all households, while the number in four-
and five-person households decreased from 46 to 26 percent.
An analysis of the data in Tables 2.16 through 2.19 suggests
that there have been dramatic changes in American households and
families in the past several decades (Sweet and Bumpass, 1987).
These are changes that have produced a very different consuming
unit than existed previously and one which is likely to demand a
more diverse range of products and services.
Socioeconomic Characteristics
Table 2.20 provides selected data on changes in the socioec-
onomic characteristics of the population during recent decades. The
data in this table verify several well-known patterns. They show a
substantial increase in the U.S. population during the past 50 years
with the proportion of high school graduates increasing from 24
percent in 1940 to 76 percent in 1988 and the proportion of persons
with 4 or more years of college increasing from less than 5 to more
than 20 percent. Also apparent is the increase in employment in
service industries and white-collar professions and the decline in
employment in extractive industries (such as agriculture and mining)
and in blue-collar occupations. Finally, the data point to little in-
crease in wealth during the 1970s and 1980s. The data in Table 2.20
suggest that the socioeconomic characteristics of the U.S. popula-
tion, like its demographic characteristics, have changed substantially
over the past several decades.

Table
2.20:
Selected
Sod.oeconomlc
Owacteristlcs
of
the
Population
of
the
United
States,
1940-1988
Year
1940
1950
1960
1970
1980
1988
Population
25
Years
of
Age
or
Older
by
Level
of
Educational
Attainment
College
()
4
years
and
over
4.6
6.2
7.7
10.7
16.2
20.3
1-3
years
5.5
7.4
8.8
10.6
15.7
17.0
High
School
()
4
years
14.3
20.8
24.6
31.1
34.6
38.9
1-3
years
15.2
17.4
19.2
19.4
15.3
11.
7
Elementary
School
()
8
years
28.2
20.8
17.5
12.8
8.0
5.2
5-7
years
18.5
16.4
13.8
10.0
6.7
4.4
0-4
years
13.7
11.
1
8.3
5.5
3.6
2.4
Median
Years
of
Education
8.6
9.3
10.6
12.1
12.5
12.7
(continues)
°'
i-

Table
2.20
(continued)
Year
Total
1940
52,705
1950
62,208
1960
69,628
1970
82,771
1980
106,940
1988
123,378
Employment
Status
of
the
Noninstitutional
Population
by
Sex:
1940-1988
(civilian
labor
force,
numbers
In
thousands)
Percent
of
Labor
Labor
Force:
Force
Participation
Rate
Unemployment
that
is
Employed
Unemployed
Rate
Female
Total
Male
Female
45,070
7,635
14.5
24.5
52.8
78.9
25.4
58,918
3,288
5.3
29.6
59.2
86.4
33.9
65,778
3,852
5.5
33.4
59.4
83.3
37.7
78,678
4,093
4.9
38.1
60.4
79.7
43.3
99,303
7,637
7.1
42.5
63.8
77.4
51.
5
116,677
6,701
5.4
51.5
65.9
81.
9
53.2
(continues)

Table
2.20
(continued)
Labor
Force
by
Occupation
and
Industry
for
the
Population
of
the
United
States,
1950-1988
3
Labor
Force
1950
1980
1988
Number
Percent
Number
Percent
Number
Percent
Occupation:
White
Collar
20,750,383
37.3
51,745,154
56.1
64,722,000
56.3
Blue
Collar
34,756,930
62.7
40,352,237
43.9
50,247,000
43.7
Industry:
Extractive
Industries
11,374,479
20.5
9,681,365
10.0
11,525,000
10.1
Manufacturing
14,575,692
26.3
21,914,754
22.4
21,320,000
18.5
Transportation,
Communication,
and
Utilities
4,368,302
7.9
7,087,455
7.2
8,064,000
7.0
Wholesale
and
Re
ta
i
I
Trade
10,547,569
19.
1
19,933,926
20.4
23,663,000
20.6
Services
12,044,705
21.
7
33,874,389
34.7
44,964,000
39.1
Government
2,488,778
4.5
5,147,466
5.3
5,432,000
4.7
(continues)
°'
(jJ

°'
.,..
Table
2.20
(r.ontinued)
Income
and
Poverty
Data
Income
(in
1989
dollars)
Median
Income
Mean
Income
Value
Standard
Value
Standard
Error
Error
Year
(do
I
la
rs)
(do
11
a
rs)
(do
11
a
rs)
(dollars)
1970
27,913
102
31,962
102
1975
27,197
111
31,758
99
1980
26,651
137
31,697
116
1981
26,020
136
31,085
113
1982
25,919
117
31,236
116
1983
26,167
118
31,883
118
1984
26,751
122
32,777
121
1985
27,218
148
33,496
133
1986
28,168
146
34,800
141
1987
28,
447
134
35,377
144
1988
28,537
138
35,656
158
1989
28,906
161
36,520
161
(r.ontinues)

Table
2.20
(c.ontinued)
Poverty
Percent
of
P·ercent
of
Year
Families
Persons
1960
20.7
22.2
1970
10.9
12.6
1980
11.
5
13.0
1989
10.3
12.8
aOccupation
and
Industry
data
for
different
years
are
not
directly
comparable
because
of
changes
in
categories
over
time.
Source:
School
Enrollment:
Social
and
Economic
Characteristics
of
Students,•
Currtnt
Population
Reports
P-20,
No.
433,
Washington,
DC:
U.S.
Government
Printing
Office,
1990.
U.S.
Bureau
of
Labor
Statistics.
Employmmt
and
Earnings
(Table
A-33).
Also
data
for
1950
to
1980
from
Employment
and
Earnings
(Tables
1
and
2),
and
Handl1oolc
of
lAbor
Statistics
(Table
15).
Data
for
1940
from
U.S.
Bureau
of
the
Census.
Historical
Statistics
of
the
United
States
(Series
Dtt-25).
Washington,
DC:
U.S.
Government
Printing
Office,
1975.
Data
for
1988
from
U.S.
Bureau
of
Labor
Statistics,
Employment
and
Earnings,
Washington,
DC:
U.S.
Government
Printing
Office,
January,
1989.
United
States
Department
of
Commerce,
Bureau
of
the
Census.
Money,
Income
and
Poverty
Status
in
the
United
States,
1989,
•
Cumnt
Population
Reports
P-60,
No.
168,
Washington,
DC:
U.S.
Government
Printing
Office,
1990.
$

66
Summary
The data in this section have shown several patterns which
can be summarized as follows:
1. The rate of population growth in the United States has
shown a nearly continuous decline since the Nation's formation. It
is expected to continue to decline in the future (see Chapter 6).
2. The Nation's growth has resulted primarily from natural
increase but immigration has played an increasingly important role
in recent decades with the origins of immigrants having shifted from
Europe to Mexico, South and Central America, and Asia in recent
decades.
3. The most rapid population growth in the United States in
recent decades has been in the West and South with slower growth
occurring in the Northeast and Midwest. Three western and south-
ern states--California, Florida, and Texas-were the major centers of
such growth in the 1970s and the 1980s.
4. The population of the United States is aging. The median
age in the United States was nearly 33 in 1990 compared to about 23
in 1900 and is projected to be more than 40 by 2050.
5. The sex ratio has declined from 104 males per 100 females at
the turn of the century to about 95 in 1990. More male than female
babies are born, but because of lower mortality rates among females,
they come to outnumber males between the ages of 20 to 30 and by
the elderly ages the number of females is nearly double that of the
number of males.
6. The U.S. population is becoming increasingly racially and
ethnically diverse. Growth among Hispanic, Asian, black, and other
racial/ethnic minorities is substantially greater than among whites or
Anglos. As a result, ethnic minorities made up about 25 percent of
the total population in 1990. These patterns are consequential
because minority populations have characteristics that are quite
different than those of majority populations, most notably they have
a much lower level of access to socioeconomic resources and smaller
resource bases than majority groups.

67
7. The proportion of married-couple households is declining
relative to nontraditional household forms. Because of the decrease
in married-couple households, the increasing rate of dissolution of
marriages, and higher ages at first.marriage, the size of households
has declined substantially with one- and two-person households
now accounting for a majority of all households. Smaller sized
consuming units seem likely to prevail for some time into the future.
8. The socioeconomic characteristics of the population of the
United States point to a nation that has an increasing proportion of
its people employed in service industries and managerial occupa-
tions. Income levels are relatively high but poverty rates also
remain relatively high for some persons (particularly minorities).
Gains in real income have been limited in the past decade, making
the future socioeconomic characteristics of the population difficult to
predict.
Conclusions
In this chapter, we have examined basic demographic con-
cepts and trends in some of the variables used to measure them.
The intent has been to provide an introduction to both the key
concepts and variables in demography and to establish a base of
information on the major trends in these factors in the population of
the United States. Although this is but an introduction to the
concepts and to the trends in their respective measures, the discus-
sion has hopefully demonstrated that the study of demographic
factors is likely to be of relevance to nearly all those involved in
planning, marketing, and other analyses of products and services.

3
The Materials of Applied Demographic Analyses:
Data Sources and Prindples of Data Use
In this chapter we examine the data sources commonly used in
applied demographic analyses. Because single sources usually con-
tain data on more than one of the demographic concepts and varia-
bles discussed in Chapter 2, the discussion here focuses on general
sources rather than on sources of data for each variable. This format
is necessary to avoid redundancy and also provides information on
data on non-demographic variables likely to be of utility to applied
analysts.
Applied demographic analyses may involve either (or both)
primary data collected by the analysts through survey or other
techniques and secondary data collected by other, usually govern-
mental, entities. However, because of the large number of areas
included in demographic analyses, the wide range of data items
required for such analyses, and the time and monetary costs associ-
ated with primary data collection, secondary data are the major
sources of information used in demographic analyses. The emphasis
here is thus largely limited to an examination of secondary sources
of demographic and related information.
In addition, because emphasis in this work is placed on applied
analyses involving demographic and other variables, we examine
not only data sources for information that are specifically demo-
graphic, but also sources of information on related topics that are
commonly used in conjunction with demographic information for
addressing pragmatic issues. These include information on econom-
ic, social, governmental, as well as demographic factors. Finally,
because it is difficult to anticipate the range of data needs of applied
analysts, we begin the discussion with an examination of indices
that can be used to locate data on a wide array of topics. After the
discussion of such indices, we present sources of information pro-
vided by federal, state, and private data sources. In the final section

70
of the chapter, we examine principles and procedures that should be
considered in the use of secondary data.
No claim is made that the sources examined are exhaustive of all
those likely to be of utility to the applied demographer, nor that the
uses and limitations described are applicable to such factors for any
given data set. In addition, the reader should recognize that the
sources of secondary data are so extensive that no single discussion
can adequately describe all available sources of information. The
fact that this provides only an introduction to secondary data
sources for applied demographic analysis must be recognized.
Indices for Locating Secondary Data
Frequently, applied demographic analysts are asked to examine
the relationships between demographic variables and other factors
with which they may be unfamiliar or on which they may have only
limited information. For example, they may have been asked to
obtain data on a topic which they have never before analyzed, or
they may have become aware of data on a topic but cannot recall the
organization that published it; perhaps they know the author of the
publication or its date of publication, but not the full citation neces-
sary to readily obtain the information. Under such circumstances, it
is useful to have knowledge of indices which provide references to
data sources. These indices are primarily available for public data
sources published within the United States, but a limited number
are also available for international and for private data sources.
These indices usually present citations to data sources arranged by
author, subject, the sponsoring or publishing agency, year, publica-
tion series, and similar categories. They are generally published
several times a year and cumulated into quarterly, annual, and
multi-year volumes. These indices are discussed below in terms of
general indices and agency and state indices.
General Indices
Among the general indices to federal and other data sources,
several are particularly useful. These include:
· The Monthly Catalog of U.S. Government Publications
· The American Statistical Index (ASD
· The Congressional Information Service Index (CIS)
·The Index to U.S. Government Periodicals

71
· The Index to International Statistics
· The Statistical Reference Index
The Monthly Catalog of U.S. Government Publications is the
oldest (published since 1895), most comprehensive, and inclusive
single source for locating items published by the U.S. Government.
Nearly all items published by any agency, department, office or
other part of the U.S. Government are cited in the Monthly Cata-
log. As the name implies, it is published monthly, cumulated
annually, and periodically (every 5 to 10 years). All items in the
Monthly Catalog are cross referenced by department or agency,
title, author, subject, and by report series for periodically appearing
reports. If one knows very little about the source for the exact data
item needed (i.e., one knows only the publication date for a data
item, only the subject matter, only the author, only the title, etc.),
the Monthly Catalog is usually the best starting point. It is avail-
able in any library with a government documents section.
The American Statistical Index (ASI) is intended to be a master
guide and index to the statistical publications of the U.S. Govern-
ment. The term statistical is loosely interpreted, however, and
nearly every item published by the U.S. Government that has
numerical data will be listed in the ASL The ASI is a relatively new
index compared to the Monthly Catalog and consists of a Retrospec-
tive Edition published in 1974 and annual and monthly supple-
ments. The Retrospective Edition contains some statistical items
going back to the early 1960s, but its comprehensive coverage is
essentially limited to periods since the early 1970s. Its monthly and
cumulated annual supplements are indexed by subject, title, author,
and geographical area.
A particularly useful aspect of the ASI is its two-part organiza-
tion. Each issue has a Part I that contains a cross-listed index of
items, and a Part II that contains a brief abstract of the item. This·
abstract can often be useful in eliminating items whose titles make
them appear to be appropriate but which on closer examination do
not provide the data desired. If one knows that quantitative, numer-
ical data are required, the use of the American Statistical Index is
nearly always desirable. As with the Monthly Catalog, nearly all
libraries will have the ASI.
The Congressional Information Service (CIS) index contains
citations for all items published by the U.S. Congress, including
committee reports and hearings. Although it is less likely to be of
interest to the statistical data user than the ASI, the CIS is a very

72
useful index and can be a time saving reference if one knows that
the data item required is a product of Congressional activities. The
as is published monthly, cumulated annually, and like the ASI, is
cross referenced by numerous categories.
The Index to U.S. Government Periodicals provides an index to
major U.S. Government periodicals (journals or magazines). Al-
though the U.S. Government publishes over 1,000 periodicals, the
Index attempts to provide references only to those which have
substantive articles of lasting research and reference value. Thus, its
coverage is limited to about 200 such periodicals, but a majority of
those likely to be of interest to applied analysts are indexed. The
Index to U.S. Government Periodicals is published quarterly,
cumulated annually, and cross referenced by several title and subject
categories.
The Index to International Statistics was first published in the
early 1980s. Prior to that time there were few indices that
referenced publications from international organizations such as the
United Nations. This index covers publications of more than 80
International Intergovernmental Organizations (IGOs), such as the
United Nations, the Organization of American States, the Organiza-
tion for E.conomic Development, and similar entities. It is published
monthly and cumulated quarterly and annually. It includes a refer-
ence and an abstract for each citation. Items are indexed by subject,
title, geographical area, issuing source, publication number, and
author. This index is published by a private company and thus,
unlike all of the other indices noted above, it will not be found in
the government documents section of libraries, but in the general
references section.
The Statistical Reference Index was first published in 1983. It is
one of the first indices to provide references to items published by
private organizations and by state governments. Among the types
of organizations included are trade, professional, and nonprofit
institutions and associations, business organizations, commercial
publishers, independent research organizations, state government
agencies, and university and affiliated research organizations. It is
published monthly and cumulated quarterly and annually. Items
are indexed by subject, author, title, issuing source, and subject
categories. As with the international index, this index is published
by a private firm and is in the general reference rather than the
government publications section of most libraries.
Although these are only six of a large number of general indices,
they are extremely useful ones. A few minutes with each of these

73
indices will assist one in identifying their particular strengths and
weaknesses and their likely utility for different types of analysis.
Agency and State-Spedfic lndia!s
In addition to the general indices to publications and data de-
scribed above, nearly every federal agency publishes a separate
index to its own publications, and most states produce one or more
similar indices. The Bureau of the Census, the U.S. Department of
Agriculture, the National Center for Health Statistics, the U.S.
Department of Labor, the Department of Health and Human Serv-
ices, the Department of Energy, the Department of Defense, the
Department of Education, and numerous other departments and
agencies publish such indices. In this section, two such indices are
examined as examples of such indices. The reader is reminded,
however, that for whatever agency one is interested in, an index to
publications and data is usually available and can expedite the loca-
tions of specific data items.
The Bureau of the Census Catalog is one of the most useful
agency-based catalogs for economic, business, demographic, and
social data users. It is published quarterly and cumulated annually
and for multi-year periods. One particularly useful version of the
Bureau's catalog is the Bureau of the Census Catalog of Publica-
tions: 1790-1972. This single source contains citations for every-
thing published by the Bureau from the first census in 1790 through
1972. With this single source and the yearly supplements since
1972, one can readily obtain a set of references to an extensive set of
historical data bases.
The catalog is arranged by subject field, geographic area, and
subject, and contains two major parts. The first part lists Census
Bureau publications, and the second part lists data files (computer-
ized data) and special tabulations. If one knows that the data re-
quired have been published by the Bureau, the Bureau of the
Census Catalog is the reference to use. The most recent issues of
the census catalog for individual years now include extensive histor-
ical references on the subjects of recently published reports. It also
contains the names, addresses, and telephone numbers for other
federal and state sources of assistance for obtaining economic and
demographic data.
The Bibliography of Agriculture plays a similar role to the
Census Bureau's catalog. The bibliography is published monthly
from sources compiled by the U.S. National Agricultural Library, the

74
Food and Nutrition Information Center, the American Agricultural
F.conomics Documentation Center, and Agriculture Canada. The
Bibliography of Agriailture is divided into ten sections: a main
entry section, five main entry subsections, a geographic index, a
corporate author index, a personal author index, and a subject
index. For those engaged in agricultural research, this index is an
extremely valuable reference source.
As noted above, nearly all major federal agencies and depart-
ments have indices to their publications and data files. Analysts
with specialized interests should become familiar with the indices
for those agencies whose data they use frequently. In addition to the
agency and general indices noted above, there are often additional
sources of assistance in locating documents published by state agen-
cies. Some states have an official entity such as the state library
which indexes state documents and publishes the index. These
indices sometimes include a periodicals section and are usually
published monthly and cumulated quarterly and annually. They
generally contain references to items indexed by agency, subject,
author, and in several other ways. Even in states where such in-
dices are not published, most state libraries maintain a checklist of
state agency publications received by the library. Consultation with
a librarian in a library's government documents' section will usually
provide the necessary information on the existence and best means
of accessing data items for states.
Aids in Using Indices
In using the above indices it is useful to be aware of both poten-
tial sources that list federal and state agencies that publish data and
any unique referencing systems that are used to catalog such refer-
ences. In the final part of this section, we examine two such aids.
One of the keys to locating a specific type of information is
knowing of the existence of an agency charged with the responsibili-
ty of collecting data on the item of interest. For federal agencies,
one of the sources that is very useful in locating the applicable
agency is a publication entitled The United States Government
Manual. This volume is published annually as a special addition of
the Federal Register. It is a comprehensive guide to the agencies,
departments, bureaus, and other divisions of the Federal Govern-
ment. For each government entity, it provides a description of its
principal officials, purpose and role, history, source of authority,

75
major programs, and most importantly, an address and telephone
number where information on available data can be obtained. This
publication can be obtained from the Office of the Federal Register
and is available in most public libraries.
The indices described above are readily available in most
public libraries, and the information requirements for their use are
limited. However, one factor common to all of the bibliographies
referencing federal government publications requires brief discussion
here. This is the classification (or coding) scheme used to classify
U.S. Government publications. This system, which is similar to the
Library of Congress or Dewey Decimal Systems used for general
library materials, is the Superintendent of Documents' Classifica-
tion System.
The system consists of a referencing code that uses a combina-
tion of letters and numbers such as:
C3.186/11:988
The letter refers to the government agency (in the example above,
the U.S. Bureau of the Census) and the numbers refer to various
subordinate offices, publication series, and other items specifying
the exact nature of the publication. When such an entry ends in a
number, that number usually refers to the year of publication of the
reference. A more complete description of the Superintendent of
Documents' Classification System and the system identifiers for
selected key agencies' central offices is available in most of the
Federal Government indices delineated above.
The Superintendent of Documents' Oassification System code is
as essential in locating a federal government publication in a library
as the Library of Congress classification system is for locating other
library books. Most libraries arrange their government documents
sections in terms of such numbers, and given this number, docu-
ment librarians can readily ascertain whether they have a given
publication. This system identifier is referenced in all of the Federal
Government bibliographies cited above, and the importance of
obtaining sufficient familiarity with this system so that one can
readily identify such numbers cannot be overemphasized.
Federal and State Data Compilations
The compiling of data items from multiple individual sources is
a tedious task. Although it is often necessary to examine different

76
sources to compile different data items, there are an increasing
number of publications that contain data compilations for specified
areas and time periods. These compilations are extremely useful in
preparing descriptions of areas for service or marketing profiles.
Before proceeding to the discussion of individual sources, we briefly
examine several commonly used national and state data compila-
tions. Federal data compilations are presented first, followed by
state compilations.
Federal Data Compilations
The number of data compilations available for examining the
economic, demographic, and other characteristics of the Nation, and
its component units, is extensive. The discussion here must be
limited to only a few widely used sources. The discussion focuses
on the following compilations:
Statistical Abstract of the United States
Historical Statistics of the United States:
Colonial Times to 1970
County and City Data Book
State and Metropolitan Area Data Book
Congressional District Data Book
The Statistical Abstract of the United States is a data source
published annually by the Bureau of the Census that is familiar to
nearly any analyst who has written a high school or college term
paper. Its familiarity, however, often leads one to ignore the utility
of this publication. Although the Abstract is oriented to the provi-
sion of national and state level data, and contains few items for
substate areas (other than major cities), it covers a large number of
data items for several time periods. Demographic, health, educa-
tion, law enforcement, environment, federal, state and local gov-
ernment, social welfare, national defense, employment, income,
consumer prices, national elections, banking, finance, insurance,
business, communications, energy, science, transportation, basic
industry statistics, and even selected international statistics are
presented in the Abstract. It is a very useful data source for both
obtaining national and state level data and for locating sources likely
to contain substate data as well.

77
One of the most useful parts of the abstract is one of which most
analysts are not aware. This is the exten.Sive Guide to Sources in-
cluded in the appendices of every edition of the Abstract. Arranged
alphabetically by subject, this guide contains references to the
primary sources of statistical information for numerous items.
Because of the general and familiar categories used in the guide, it
serves as an excellent source for identifying the agencies likely to
publish data on a given item and can serve as a useful first step in
identifying sources that can be further identified through the use of
the indices described above. Condensed versions of statistical ab-
stract data are published periodically in the Pocket Data Book and
in the U.S.A. Statistics in Brief.
The Historical Statistics of the United States: Colonial Times to
1970 is a 1,200-page, two-volume data compilation prepared by the
Bureau of the Census (1975) in celebration of the Nation's bicenten-
nial. It is the third in a series of volumes (the two others appeared
in 1949 and 1960) intended to provide a convenient reference source
on key U.S. data items. It presents data for time periods from 1790
through 1970. A large majority of the data entries in these volumes
are presented only for the Nation as a whole, though a few items
contain state data as well. A few of the items for which data are
presented include population, national income and wealth, land and
water use, climate, forests, minerals, major industry indicators,
energy, government, and communication. These volumes are useful
primarily for the histories they present for each data item. For
example, if one wishes to know the changes that have been made in
the definition of the consumer price index, the poverty level index,
in the definition of a farm, the year that a given data item was first
collected, or the major sources of given data items, the descriptions
of data items at the beginning of each major section of this source
(e.g., population, energy) can be valuable. This source is extremely
useful for those doing historical or time-series analyses.
The County and City Data Book has been published since 1939
by the Bureau of the Census at roughly five-year intervals. The data
book presents data items derived from each of the major U.S.
Bureau of the Census' most recent population, housing, business,
agriculture, and government censuses, surveys, and other programs.
Although this volume provides data for only the most recent data
collection point (the last census or other reecent data source), its
strength lies in its geographical coverage. Hundreds of data items
are provided not only for the Nation, federal administrative regions,

78
census divisions, and all states, but also for all counties, metropoli-
tan statistical areas (MSAs), and cities having 25,000 inhabitants or
more (and selected items for places of 2,500 or more) in the United
States. The data book also provides extensive appendices and de-
scriptive materials that are very useful in gaining familiarity with
census terms and definitions. For profiling activities, the County
and City Data Book is an extremely valuable resource.
The State and Metropolitan Area Book is a recently (since 1980)
developed data summary patterned after the County and City Data
Book. It presents a variety of statistical information for states and
for metropolitan areas in the United States. A recent volume, for
example, contained state rankings for more than 128 items, nearly
1,900 statistical items for each state and 300 items for each metropol-
itan area. The publication is published more frequently than the
five-year interval used for the County and City Data Book.
The Congressional District Data Book is also similar to the
County and City Data Book but is published for the congressional
districts as constituted after the most recent census. In addition to
providing demographic, economic, and other data for each congres-
sional district, this source has an accompanying Congressional Dis-
trict Atlas that presents maps showing the boundaries and a list of
the counties and municipalities in each district. It is an essential
source for those involved in political analyses.
The national data compilations described above are useful for a
large number of descriptive data collection and analysis tasks.
Although they cannot replace the need for frequent consultation
with basic data sources, they can be used to expedite basic data
collection and to locate data sources.
State Data Compilations
Data compilations at the state level have been produced much
less systematically. Individual, agency, or university groups often
prepare such summaries for given places or periods of time, but the
content and frequency of publication of such items makes them
generally less useful than the national compilations.
Many states publish state almanacs or data books on a recurrent
basis (e.g., the Texas Almanac published by The Dallas Morning
News has been published for over fifty years). These almanacs
often contain data as well as promotional material for counties and
major cities and contain brief writeups on each of several major
topics by state specialists on each topic. Although oriented to more

79
general audiences than the Census Bureau's compilations, state
almanacs are often valuable references for those seeking to gain
basic familiarity with the characteristics of a state.
In summary, then, national and state-level data compilations
provide summaries of a large number of data items produced by
numerous sources. Although the geographical and time referents
for such summaries vary widely, they are extremely useful for area
profiling, for gaining a general familiarity with a large number of
areas, and for locating potential data sources for specific data items
and geographical areas.
Federal Data Sources
Any attempt to describe federal data sources must be a very
limited one because literally hundreds of agencies produce data
likely to be valuable to data users. The discussion here concentrates
on a description of those agencies likely to be of greatest interest to
applied demographic analysts.
Although there are numerous agencies in the Federal Govern-
ment that collect data, a few collect a large proportion of all data
collected. The Economic Statistics Service.in the Department of
Agriculture, the Bureau of the Census, and the Bureau of Labor
Statistics have historically had roughly 50 percent of the total budget
for statistical analysis for current (ongoing) programs and nearly all
of the statistical budget for periodic programs. Many agencies
simply provide funding for data collection by these three and still
other agencies simply analyz.e such data or make other compilations
from collected data. An analysis of just a few programs, then, can
provide a useful overview of the major federal data sources (see also
Wallman, 1988).
The discussion in this section focuses on a description of the
major programs of a selected number of agencies. The agencies
included were selected on the basis of their overall importance in
the Federal statistical system and on the basis of their likely utility
to applied demographic analysts. The agencies to be described
include:
U.S. Bureau of the Census
U.S. Bureau of Economic Analysis
U.S. Bureau of Labor Statistics
National Center for Health Statistics
National Center for Education Statistics

80
The major programs and types of data provided by each of these
five agencies are briefly described below.
The U.S. Bureau of the Census
The Bureau of the Census is clearly the dominant data collection
agency in the Federal Government with the largest budget and
number of personnel devoted to data collection. Its best known
activities are those associated with decennial censuses, but it, in fact,
is involved in a wide range of ongoing censuses and survey activi-
ties in addition to the U.S. Census of Population and Housing.
The Census Bureau publishes data for governmental jurisdictions
and for statistically defined areas. Data are available for all units
from the Nation as a whole down to the individual city block. Data
are provided in published, tape, microfiche, floppy disk, and CD-
ROM (compact) disk forms. Although most data published by the
Bureau are for geographical units, data are also provided on micro-
data• sets which contain information on individual persons and
households (e.g., the Public Use Microdata Sample). In these
microdata sets, precautions have been taken to prevent identification
of individual persons and households to protect their confidentiality,
but the full data set for each individual and household (except for
those which would identify a specific person or household) are pro-
vided. The Bureau's major data collection programs include various
censuses and survey programs.
Censuses of the U.S. Bureau of the Census. The major censuses
conducted by the U. S. Bureau of the· Census are:
Census of Population and Housing
Census of Retail Trade
Census of Wholesale Trade
Census of Service Industries
Census of Manufacturing
Census of Mineral Industries
Census of Construction Industries
Census of Transportation
Census of Governments
Census of Agriculture
Of these censuses, the Population and Housing Census is
conducted every ten years in years ending in •o•. Each of the other
censuses is conducted every five years in years ending in ·2· and

81
•7• (e.g. 1987 and 1992). As censuses, they attempt to obtain data
on the universe of units of interest. The Population and Housing
Census attempts to obtain selected data on every individual and
household in the United States, while the other censuses attempt to
·obtain data on every unit or establishment (business, farm or unit of
government) in the Nation. Other specialized data items are ob-
tained from large sample surveys conducted in conjunction with
each census. Data are provided for a large number of geographical
units down to the county level for all of the censuses with subcoun-
ty data being available from some censuses (e.g., Census of Popula-
tion and Housing).
The 1990 Census. The Census of Population and Housing is the
census of most interest to applied demographers because it is the
source of basic information for each decade. The 1990 Census
provides data for the United States as a whole, for regions and
divisions within the United States, for states, counties, minor civil
divisions (or census county divisions), places (cities, towns, and
villages), census tracts (or block numbering areas in areas without
tracts [usually rural areas]), block groups, and blocks. The geo-
graphical coverage of the 1990 Census is extensive.
The 1990 Census was the largest and most expensive in history
with more than 249 million persons being counted (including per-
sons overseas) and the costs exceeding 2.6 billion dollars. The 1990
Census was not only the largest in history, it also introduced
numerous innovations in procedures and products which are suffi-
ciently different than those from past censuses to merit discussion
here (see also Robey, 1989; U.S. Bureau of the Census, 1989).
Two aspects of the procedures used in the census are particular-
ly noteworthy. First, to complete the 1990 Census, the U.S. Bureau
of the Census formulated census blocks for the entire Nation. This
created the largest number of units for analysis ever used by the
census and promises to improve data availability for many forms of
analysis. Although such blocks are often composed of quite large
geographical areas in low population density areas (such that some
rural areas actually have fewer geographical tabulation areas than in
previous censuses), the formation of blocks offers significant new
opportunities for data users. .
Of central importance to the delineation of blocks for the entire
nation and the increased geographical coverage of the 1990 Census
was the development of the Topologically Integrated Geographic
Encoding and Referencing (TIGER) system. This system was de-
veloped in cooperation with the U.S. Geological Survey and for the

82
first time produced a census for which there are computerized maps
for the entire United States. This system contains geocodes (longi-
tudes and latitudes for the boundaries) for all census geographic
areas and additional information showing the topographic and other
physical and man-made features of geographic areas. Although the
U.S. Bureau of the Census does not provide software for the use of
the TIGER system, when combined with a comprehensive Geo-
graphical Information System (GIS) available from numerous sources
in the private sector, it offers unprecedented abilities to use geo-
graphically referenced census data with other geographically refer-
enced information.
The 1990 Census also resulted in new products, a rearrangement
and referencing of products on a wider range of media than ever
before, and new challenges in its completion. Its content and major
products include publications and computer tapes such as those
noted in Figures 3.1 through 3.3. As noted in Figure 3.1, as with
past censuses, the census was conducted with a short form, which
went to about 5-of-every-6 households and contained basic informa-
tion, and a long form, which went to about 1-in-every-6 households
and contained all of the items on the short form plus many more de-
tailed items.
The publications from the 1990 Census use a different referenc-
ing system than in previous censuses. Prior to 1990, census reports
were referred to by the initials PC for population census, HC for
housing census and PHC for reports that combined items from the
population and housing censuses. In 1990, as shown in Table 3.2,
the reports are referred to by the initials CP for census of population
reports, by CH for census of housing, and CPH for census of popu-
lation and housing reports. In addition to this change in referencing
systems, the 1990 reports also contain somewhat different informa-
tion in different volumes than in previous censuses. Most notable
among these changes is the fact that data for American Indian
Reservations and Alaskan Native Areas, for Metropolitan Statistical
Areas, and for Urbanized Areas are presented in separate reports.
This was done in order to improve the timeliness of publication
which has often been delayed by the inclusion of such areas in
comprehensive, multi-area publications in previous censuses. Most
notable among those items that were eliminated in the 1990 Census
was the Detailed Characteristics of the Population,• which had
been chapter D of the census volumes for several previous censuses.
The computerized products of the 1990 Census shown in Figure
3.3 are similar in many regards to those for 1980. As with past
censuses, the amount of information available on computer-related

Figure 3.1: Short-Form (l!JO'lb Items) and Long-Form (Sample Items)
Topics in the 1990 Census of Population and Housing
Population
Household relationship
Sex
Race
Short-Form (100%) Items
Housing
Number of units in structure
Number of rooms in unit
Tenure (owned or rented)
83
Age
Marital status
Hispanic origin
Value of home or monthly rent paid
Congregate housing
Vacancy characteristics
Population
Long-Form (Sample) Items
Housing
Social characteristics:
Education (enrollment and attainment)
Place of birth, citizenship, and
year of entry to the United States
Ancestry
Language spoken at home
Migration (residence in 1985 vs. 1990)
Disability
Fertility
Veteran status
Economic characteristics:
Labor force
Occupation, industry,
and class of worker
Place of work and journey to work
Work experience in 1989
Income in 1989
Year last worked
Condominium status
Plumbing and kitchen fadlitles
Telephone in unit
House heating fuel
Source of water and method of
sewage disposal
Vehicles available
Year structure built
Year moved into residence
Number of bedrooms
Farm residence
Shelter costs, including utilities
Source: The 1990 Census of Population and Housing: Tabulation and Publication Program.
U.S. Department of Commerce. U.S. Bureau of the Census. Washington, DC:
U.S. Bureau of the Census, 1989.

84
Figure 3.2: Publications of the 1990 Census of Population and Housing
1990 Census of Population and Housing (CPH)
100-Percent Data
CPH-1-Sumrnary Population and Housing Oiaracteristics. This report provides total
population and housing unit COWlts as well as summary statistics. on age, sex,
race, Hispanic origin, household relationship, wli.ts in structure, value or rent,
number of rooms, tenure, and vacancy characteristics for local governments.
CPH-2-Population and Housing Unit Counts. This report provides total population
and housing unit counts for 1990 and previous censuses. Data are shown for
states, counties, minor civil divisions (MCDs)/census county divisions (CCDs),
places, state component parts for MSA's and UA's, and summary geographic
areas (for example, urban and rural, and metropolitan and nonmetropolltan
residence).
100-Percent Data and Sample Data
CPH-3-Population and Housing Characteristics for Census Tracts and Block
Block Numbering Areas. One report published for each MSA/PMSA and one
for the non-MSA/PMSA balance of each state showing data for most of the
population and housing subjects included in the 1990 census. Statistics
presented for a MSA/PMSA state-county-place of 10,000 or more-census
tract/block numbering area geographic hierarchy.
CPH-4-Population and Housing Characteristics for Congressional Districts of
the 103rd Congress. One report ls available for each state and the District of
Columbia showing population and housing data for congressional districts as
well as counties, places of 10,000 o~ more inhabitants, and MCDs of 10,000 or
more within each congressional district.
Sample Data
CPH-5-Summary Social, Economic, and Housing Characteristics. This report
provides sample population and housing data for local governments, including
American Indian and Alaskan Native Areas.
1990 Census of Population (CP)
100-Percent Data
CP-1-General Population Oiaracteristics. Detailed statistics on age, sex, race,
Hispanic origin, marital status, and household relationship characteristics are
presented for states, counties, places of 1,000 or more inhabitants, MCDs of
1,000 or more inhabitants in selected states, and summary geographic areas.
(rontinues)

85
Figure 3.2 (rontinued)
CP-1-lA-General Population Characteristics for American Indian and
Alaskan Native Areas (Al/ANA's). Data are shown for American Indian
and AJaskan Native Areas-American Indian reservations, trust lands, tribal
jurisdiction statistical areas in Oklahoma, tribal designated statistical areas,
Alaskan Native village statistical areas, and AJaskan Native Regional Corpo-
rations.
CP-1-18-General Population Characteristics for Metropolitan Statistical Areas
(MSA's). This report indudes data for individual MSA's and their compo-
nent areas.
CP-1-lC-General Population Characteristics for Urbanized Areas (UA's). This report
indudes data for individual UA's and their component areas.
Sample Data
CP-2-Social and Economic Characteristics. This report focuses on the
population subjects collected on a sample basis in 1990. Data are shown for
states (induding summaries such as urban and rural), counties, places of 2,500
or more inhabitants, Minor Ovll Divisions (MCDs) of 2,500 or more inhabitants.
CP-2-lA-Social and Economic Characteristics for American Indian and
Alaskan Native Areas. Data are shown for American Indian and AJaskan
Native Areas.
CP-2-18-Sodal and Economic Characteristics for Metropolitan Statistical Areas. Data
are shown for MSA's.
CP-2-1C-Social and Economic Characteristics for Urbanized Areas. Data are shown
for UA's.
CP-3-Population Subject Reports. Thirty reports covering population subjects
and subgroups. Geographic areas generally will include the United States,
regions, and divisions; some reports may indude data for other large geograph-
ic areas. Tentative topics and titles indude:
-Characteristics of the Rural and Farm Population
-Geographical Mobility for States and the Nation
-Geographical Mobility for Metropolitan Areas
-Recent and Lifetime Migration
-Journey to Work: Metropolitan Commuting Flows
-Journey to Work: Characteristics of the Workers ln
Metropolitan Areas
-Place of Work
-Current Language of the American People
-Education
-The Older Population of the United States
(rontinues)

86
Figure 3.2 (amtinued)
-Persons In Institutions and Other Group Quarters
-Detailed Social and Economic Characteristics of the
Population
-Households, Famllies, Marital Status, and Living
Arrangements
-Fertility
-American IndJans, Eskimos, and Aleuts In the United
States
-Characteristics of American Indians by Tribe and
Language for Selected Areas
-Characteristics of the Asian and Pacific Islander
Population In the United States
-Characteristics of the Black Population In the United
States
-Persons of Hispanic Origin In the United States
-Ancestry of the Population In the United States
-The Foreign-Dom Population In the United States
-Employment Status, Work Experience, and Veteran
Status
-Occupational Characteristics
-Industrial Characteristics
-Occupation by Industry
-Earnings by Occupation and Education
-Sources and Structure of Household and Family
Income
-Characteristics of Persons In Poverty
-Poverty Areas In the United States
-Characteristics of Adults with Work Disabilities,
Mobility Umltations, or Self-Care Umltations
1990 Census of Housing (CH)
100-Percent Data
CH-1-General Housing Characteristics. Detailed statistics on units in structure, value
and rent, number of rooms, tenure, and vacancy characteristics are presented
for states, counties, places of 1,000 or more Inhabitants, MCDs of 1,000 or more
Inhabitants, and summary geographic areas.
CH-1-lA-General Housing Characteristics for American Indian and Alaskan Native
Areas. Data are shown for American Indian and Alaskan Native
areas--American Indian reservations, trust lands, tribal jurisdiction statistical
areas In Oklahoma, tribal designated statistical areas, Alaskan Native village
statistical areas, and Alaskan Native Regional Corporations.
CH-1-18-General Housing Characteristics for Metropolitan Statistical Areas.
This report Includes data for the individual MSA's and their component
areas.
(amtinues)

87
CH-1-lC-General Housing Characteristics for Urbanized Areas. Data are shown for
Individual UA's and their component areas.
Sample Data
CH-2-Detailed Housing Characteristics. This report focuses on housing data
collected on a sample basis In 1990. Data are shown for states (Including
summaries such as urban and rural), counties, places of 2,500 or more Inhabi-
tants and MCDs of 2,500 or more Inhabitants.
CH-2-lA-Detalled Housing Characteristics for American Indian and Alaskan Native
Areas. Data are shown for American Indian and Alaskan Native areas.
CH-2-lB-Detalled Housing Characteristics for Metropolitan Statistical Areas. Data are
shown for MSA's.
CH-2-lC-Detailed Housing Characteristics for Urbanized Areas. Data are shown
for UA's.
CH-3-Houslng Subject Reports. Ten housing subject reports are to be available
for 1990. Geographic areas shown In the reports generally will Include the
United States, regions, and divisions; some reports may Include data for other
large geographic areas. Tentative topics and titles Include:
-Metropolitan Housing Characteristics
-Mobile Homes
-Recent Mover Households
-Housing of the Elderly
-Condominium Housing
-Structural Characteristics
-Utilization of the Housing Stock
-Housing Quality Indicators
-Second Mortgage Households
-Characteristics of New Housing Units
Source: The 1990 Census ofPopulation and Housing: Tabulation and PubUcation Program.
U.S. Department of Commerce. U.S. Bureau of the Census. Washington, DC:
U.S. Bureau of the Census, 1989.

88
Figure 3.3: Computerized Products from the 1990 Census
Public Law 94-171 File for 1990
Public Law 94-171 Census Tape for 1990 was developed for redistricting.
Contains Information for:
- states
- counties
- minor dvil divisions/census county divisions
- places
- census tracts/block numbering areas
- block groups
- blocks
Contains data for the following characteristics for the total populations
and for persons 18 years old and older:
- total population
- counts of the population by race for white; black; Asian and
Padflc Islander; American Indian, F.sldmo and Aleut; and other
- total Hispanic origin
- cross tabulations of data for persons not of Hispanic origin by
race
STF1 - Summary Tape File 1
SlFl includes 100-percent population and housing counts and character-
istics similar in subject content to the 1980 STFl but with expanded
detail:
~ Ao provides data for states and their subareas in hierarchical
sequence down to the block-group level.
flk ~provides data for the full geographical hierarchy for
states to the block level.
~ !:,;. for the United States has the following geographic struc-
ture: United States, regions, divisions, states (including
summaries such as urban and rural) counties, places of
10,000 or more, MSA's, and UA's.
(amtinues)

Bk Qi. provides data by state for congressional districts of the
103rd Congress as well as for counties and places of
10,000 or more inhabitants within each congressional
district.
STF2 contains 100-percent population and housing characteristics similar
to the 1980 S1F2. It lndudes records for the total population and itera-
tions for race and Hispanic origin:
~ Ao provides data for census tracts/BNAs, places of 10,000 or
more inhabitants (to the tract/BNA level), and whole
tract/BNA summaries.
~ Jt is an inventory-type ftle rather than hierarchical in struc-
ture. Data are presented for the state (including
summaries such as urban and rural), counties, places of
1,000 or more inhabitants.
~ ~ shows data for the United States, regions, divisions,
states, counties, and places of 10,000 or more inhabi-
tants, MSA's, and UA's.
S'IF3 lndudes sample population and housing characteristics similar tn
subject content to the 1980 S1F3 but with expanded detail:
~ Ao provides data for states and their subareas in hierarchical
sequence down to the block group level. Summaries for
whole places, whole census tracts/block numbering
areas, and whole block groups.
Bk B.;. provides data summarized for 5-digit ZIP Codes within
each state, including county portions of the areas.
Ek ~ provides data for the United States, regions, divisions,
states, counUes, places of 10,000 or more inhabitants,
MSA's, and UA's.
(rontinues)
89

90
Figure 3.3 (continued)
File D: provides data content by state for congressional districts,
as well as for counties, places of 10,000 or more Inhabi-
tants within each congressional district.
STF4 • Summary Tape File 4
SlF4 contains sample population and housing characteristics but shows
more subject detail than SlF3. Each file of SlF4 Includes records for the
total population and iterations for race, Hispanic origin, and possibly
selected ancestry groups:
~ A: provides data for census tracts/BNAs In MSAs and In the
remainder of each state In a geographical hierarchy of
county, places to the census tract/BNA level. Whole
census tract/BNA summaries also are Included.
File B: is an Inventory-type file rather than In a hierarchical
structure. Data are presented for the state (Including
summaries such as urban and rural), counties, places of
2,500 or more inhabitants, MCD's of 2,500 or more
Inhabitants.
File .C,: presents data for the United States, regions, divisions,
states (Including urban and rural and metropolitan and
nonmetropolitan components), counties, places of 10,000
or more Inhabitants, MSA's, and UA's.
Public Use Microdata Sample (PUMS)
PUMS are computerized files containing a sample of Individual long-
form census records showing most population and housing characteris-
tics:
S·Percent County Groups PUMS: presents most population and housing
characteristics on the sample questionnaire for a 5-percent sample of
housing units. It shows data for county groups or smaller areas with
100,000 or more Inhabitants. This file is similar to the 1980 PUMS-A
Sample.
(continues)

1-Percent Metropolitan Statistical Areas PlJMS: presents most popula-
tion and housing characteristics on the sample questionnaire for a 1-
percent sample of housing units. It shows data for MSAs that are
used in the 1990 Census. This me is similar to the 1980 PUMS-B
Sample.
Special Computer Tape Files
Census/Equal Employment Opportunity (EBO): provides sample census
data to support affirmative action planning for equal employment
opportunity for all counties, MSA's, and places of 50,000 or more
inhabitants.
County-to-County Migration File: providing summary records by state
for all intrastate county-to-county migration streams and significant
interstate county-to-county migration streams.
1990 Census Data Available on CD-ROM:
PL94-171 Redistricting Data
STFlA
STF lB (Block Statlstics)
STFlC
STF3A
STF 3B (Zip Code)
STF3C
EEO
County-to-County Migmtlon F1le
Source: Tht 1990 Ctnsus of Population and Housing: Tabulation and Publica-
tion Program. U.S. Department of Commerce. U.S. Bureau of the
Census. Washington, DC: U.S. Bureau of the Census, 1989.
91

92
media is substantially greater than that which is published, with less
than 10 percent of all information being available in printed form.
What was perhaps most innovative in the 1990 Census was the use
of new media for the dissemination of the 1990 results. Although
used by the U.S. Census Bureau during the late 1980s, the 1990
Census was the first census to make extensive use of the release
of data on CD-ROMs (Compact Disk Read Only Memory). These
disks are similar to those used for recordings of popular music.
They contain extensive information on a single disk (information
equivalent to about four 6250 BPI computer tapes or about 1,500 low
density floppy disks), are intended for use in a personal computer
environment, and allow the user to access information at any loca-
tion on the disk. This overcomes many of the data access problems
encountered in the use of hierarchically arranged computer tapes.
The 1990 Census is also controversial. Prior to the time it was
conducted, legal suits and legislation were initiated to prevent the
Bureau from counting illegal aliens, to include and to not include
persons living overseas in the apportionment count, and to count
military at their last permanent base rather than their state of induc-
tion, etc. The most important challenge, however, was that related
to the adjustment of the census for errors of closure (undercount
and/or overcount). The results of the 1990 Post Enumeration Survey
and of demographic analysis techniques used to assess coverage
show that the 1990 Census likely missed at least 5.0 million people,
a substantially larger number than the 2.2 million missed in 1980
and that, as with previous censuses, the errors of closure were
largest for minority populations (see U.S. Bureau of the Census
Press Release CB91-131, April 18, 1991a and CB91-221, June 13,
1991b). Although, the Secretary of Commerce has decided not to
adjust the 1990 Census counts, the adequacy and coverage of census
results will continue to be a topic of debate artd litigation.
In sum, the range of data available from decennial censuses is
extensive. It should be clear that knowledge of the data from the
decennial censuses is critical to the applied analyst and must be a
continuing area of study for applied demographers.
The Suroeys oft~ U.S. Bureau oft~ Census. The Bureau of the
Census also conducts a large ongoing survey program. Among
these surveys are:
Current Population Survey (CPS)
American Housing Survey (AHS)
Current Construction Survey

Current Business Survey
Survey of Minority-Owned Businesses
Current Industrial Survey
Foreign Trade Survey
Survey of Income and Program Participation (SIPP)
93
These surveys are often conducted in conjunction with other agen-
cies and are the source of such frequently cited annual statistics as
those on population, household, and family characteristics, housing
starts, international balance of payments, business failures, and
wholesale and retail business inventories.
The most widely used of these surveys are the Current Popula-
tion Survey (CPS) and the Survey of Income and Program Participa-
tion (SIPP). Conducted every month with expanded versions or
supplements being administered in different months of the year, the
Current Population Survey is the major source of data for the
Current Population Report Series. The Current Population Survey
obtains data from about 60,000 households, but it generally provides
data representative at only the national and census region level. Its
sampling design is now being analyzed to determine if it can be
redesigned to provide data that are generalizable at the state level.
The Survey of Income and Program Participation (SIPP) was first
· conducted in 1983 after years of development and testing of its
content and sampling procedures. The survey uses a panel design
so that persons and households are interviewed once every 4
months over a 2 1/2 year time frame. The survey provides uniquely
detailed data on such characteristics as educational attainment, work
histories, health and disability characteristics, assets and liabilities,
pension plan coverage, earnings and benefits, property taxes, and
expenditures for detailed items such as childcare, child support,
housing, etc. It provides the most detailed data available on income
from roughly 50 separate sources. Although it provides data that
are representative only at the national level, it is an important
source of information for analysts doing research on income and
economic benefits.
Among the recurrent reports (popularly called the P-Series)
produced from these surveys are:
P-20: Population Characteristics-Periodic summaries of
trends in demographic characteristics for the
United States (e.g., age, race, sex, household
composition, migration).

94
P-23: Special Studies--Periodic data summaries on
demographic, economic, social, and other charac-
teristics of the United States and other popula-
tions (e.g., alimony, minority populations, world
population, youth, etc.) produced as a result of
special analysis efforts.
P-25: Population Estimates and Projections-Estimates
for the United States (monthly), states, counties,
and incorporated areas (periodically) and projec-
tions for selected periods.
P-26: Federal-State Cooperative Program for Population
Estimates--Population estimates produced
through the Federal-State Cooperative Program
involving state as well as federal demographers.
P-28: Special Censuses--Basic population counts for
special censuses conducted by the Census Bu-
reau.
P-60: Consumer Income-Data on individual, family,
and household income and poverty levels.
P-70: Household Economic Studies-Data on family and
household income and other economic, demo-
graphic, and social factors from the Survey of
Income and Program Participation.
Another widely used product of the Bureau's survey activities is
County Business Patterns. This publication, and associated data
files, provide information on business establishments, employees,
and payrolls for major industry groups. It is the only Census
Bureau publication providing such data at the county level that is
available on an annual basis.
In addition to its domestic programs, the Bureau produces .
extensive data on other nations. The Bureau has several major
programs in the area of international statistics. The products of
these programs include such items as periodic population estimates
for all major countries in the world, country-specific profiles, maps,
and other useful data items. A recently expanded set of actjvities
related to international statistics has been the Bureau's establish-
ment of a Foreign Trade Division. It publishes information on
foreign exports and imports for major metropolitan areas and states
in the United States. The international programs of the Bureau are
described in brief form in Factfinder for the Nation Number 21
(U.S. Bureau of the Census, 1991c) and its foreign trade data are

95
usefully summarized in a summary document entitled, A Guide to
Foreign Trade Statistics (U.S. Bureau of the Census, 1991d).
The Bureau also publishes a wide number of references that
describe census data collection efforts and serve as aids in using
census data. These include:
catalogs of publications and directories of data files
(computer tapes);
guides and manuals for the use of major census products
and for specific types of data and jurisdictions (e.g., direc-
tories for using economic data or data for local areas);
manuals explaining its major classification and other meth-
odological systems (e.g., Standard Industry and Occupa-
tion Oassification Manuals);
statistical summaries or compilations;
specialized methodological studies; and
general works describing census products or history.
Among the most useful of such general census products are the
Census Bureau's guides to the Population and Housing Census and
to the economic censuses. The Users' Guide to the 1980 Decennial
Census, for example, presented a listing of every variable available
in the Census, displayed every way it is used, independently and in
conjunction with other variables, and indicated where in the Census
each use could be found. A similar guide is under preparation at
this time for the 1990 Census. Guides are also available on special-
ized topics such as the Guide to Census Data on the Ederly.
The U.S. Bureau of the Census is the major data collection and
dissemination agency in the Federal Government and knowledge of
its data is crucial to the applied analyst. However, given the
volume of data produced, it is difficult to remain current on data
available from the Bureau. In this regard, several services and
publications provided by the Bureau may be especially helpful.
These include the Bureau's on-line service which provides direct
access (via computer terminal) to lists of new census data sets and
publications. This on-line service is called CENDATA. Its content
is developed and maintained by personnel from the Bureau but
public access is offered only through several private on-line services.
Information on gaining access to it can be obtained by contacting the
Bureau's Data Users Services Division.
Among the most useful of publications for keeping track of the
Bureau's activities is the Census and You (formerly called the Data
User News) which is published monthly. It describes major new

%
services and data sets. Among its most useful items is a frequently
published list of telephone numbers for subject matter specialists in
the Bureau. Another useful publication lists new data sets and
publications available from the Bureau. It is called the Monthly
Product Announcement. It presents brief lists and descriptions of
new Census Bureau data sets. Other useful series include the Fact-
finder for the Nation and the Statistical Briefs series. Information
on these references can be obtained by writing the Data Users Serv-
ices Division, U.S. Bureau of the Census, Washington, D.C. 20233.
In sum, the U.S. Bureau of the Census is the major federal data
source. Basic familiarity with its products and programs is essential
for anyone doing applied demographic analysis.
Bureau of Economic Analysis
The Bureau of :Economic Analysis (BEA) like the Census Bureau
is part of the Department of Commerce. The BEA is a major pro-
vider of economic projections of employment and income. The
function of BEA is to prepare the economic accounts of the United
States and to interpret economic developments in the light of these
accounts and other information. Among the major types of products
and programs it provides are
data on national income and the gross national product
(GNP);
data on business and other components of national
wealth;
balance of payments data giving details on U.S. transac-
tions with foreign countries;
data on interindustry relationships showing how the
Nation's industries interact to produce GNP; and
detailed data on economic activity by region, state,
metropolitan area, and. county.
These data are supplemented by various forecasts of
· investment outlays and programs of U.S. business; and
· leading, lagging, and coincident business cycle indicators.
The Bureau's major publications include the monthly publica-
tion, Survey of Current Business, the biennially produced Business
Conditions Digest, and the periodically produced Long Term
Economic Growth. Among its most widely used data items are its

97
yearly estimates of income and employment by sector for counties.
These projections are extremely useful in providing baseline data for
projecting the potential impacts of new economic development
programs on an area's economy.
Bureau of Labor Statistics
The Bureau of Labor Statistics (BLS), located in the Department
of Labor, is the chief agency providing data on labor and price statis-
tics. Its major programs and products are described in its periodical-
ly produced publication, Major Programs of the Bureau of Labor
Statistics.
Its major activi~es include programs and products on
current labor force--size of the labor force and labor
force projections;
employment structure and trends--industrial and
occupational data;
prices and living conditions--including the Consumer
and Industrial Price Indices;
wages and industrial relations--including data on work
stoppages, area wage surveys, and similar data;
productivity and technology--including productivity
measures and statistics;
occupational safety and health statistics--including
data on occupational injuries and illness;
economic growth--including patterns and projections of
economic growth; and
general information on labor-related data use and
analysis, including publications such as:
Monthly Labor Review,
Occupational OuUook Quarterly,
Handbook of Labor Statistics, and
BIS Handbook of Methods.
The Bureau of Labor Statistics is clearly the major federal source for
labor-related data.
The National Center for Health Statistics (NCHS) is a major
source of data on health and health services. Among the major

98
references of utility for accessing the data of the center are its Cata-
log of Public Use Data Tapes From NCHS and its periodically
published Catalog of Publkations of the NCHS.
The center is a major provider of data in the following health
and related areas:
morbidity
mortality
natality
· use of health services (e.g., hospitals,
physicians, and clinical care nursing homes)
health care costs and insurance
marriage and divorce
health planning data
Among the most useful data compilations on health published
periodically by NCHS is its Health: United States. The Center is
also the source of data tapes and other information from numerous
health-related surveys collected by NCHS including the:
-National Health Interview Survey
-National Health And Nutrition Examination Survey
-National Survey of Family Growth
-National Hospital Discharge Survey
-National Ambulatory Medical Care Survey
-National Nursing Home Survey
-National Health Care Survey
-National Maternal and Infant Health Survey
For the data user interested in health statistics, familiarity with
NCHS is essential.
National Center for Eduation Statistics
The National Center for Education Statistics (NCES) plays a
similar role to NCHS in the area of education. It is a general-pur-
pose federal statistical agency responsible for collecting, analyzing,
and disseminating statistics about education. Among the key refer-
ences essential for accessing NCES data are its periodically produced
Catalog of Publiations and Directory of Computer Tapes.

are:
99
Among the characteristics described by NCES data and products
public school enrollments (elementary,
secondary, and higher education),
public school finance,
teacher characteristics and salaries,
school facility characteristics,
vocational education characteristics,
adult and continuing education characteristics,
special education programs, and
correlates of educational attainment.
The NCES is a major source for data on the educational status,
attainment, facility and personnel conditions, and characteristics of
education in the United States.
Other federal agencies provide data in their given areas of
responsibility. The Department of Justice and the Federal Bureau of
Investigation are key sources for crime and criminal justice system
data. The Energy Information Center in the Department of Energy
is the major source of data on energy conservation and use and the
U.S. Geological Survey and the Bureau of Mines of data on energy
reserves and resources. The Federal Reserve Board, the Internal
Revenue Service, and the Federal Trade Commission are primary
sources of data on financial statistics. Transportation data can be
obtained from the Department of Transportation, and environmental
data from the Environmental Protection Agency. Housing data not
available from the Bureau of the Census can be obtained from the
Department of Housing and Urban Development, and data on
income maintenance and human services can be obtained from the
Department of Health and Human Services. For each of these
agencies, a review of data indices will usually reveal a newsletter or
a guide to programs or data that can be used as an initial step in
accessing the agency's data.
Although no single source can provide copies of all government
publications, tapes, etc., the National Technical Information Service
(NTIS) in the Department of Commerce has been designated as the
central source for the public distribution of government-sponsored
research reports and data. It is an excellent source for governmental
reports, especially those for historical time periods. The NTIS can
be contacted by writing the U.S. Department of Commerce, National
Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22161.

100
Federal data sources provide a wealth of data for corporate
and governmental planning, management, and analysis. Although
it is a major task to gain familiarity with the data available from
even a few federal agencies, it is an essential task for an applied
data user and one likely to provide better data for the research
analyst.
State Data Sources
As with federal agencies, the number of state agencies produc-
ing data is extensive. Nearly every state's agencies publish at least
some information in an annual or biennial report, and many also
publish newsletters, fact sheets or fact books, bulletins or similar
items. As for federal agencies, knowing the name of the agency
likely to produce data in a given subject area is often an essential
first step in accessing the required data, and a simple list of agencies
can be useful. Many states have handbooks or similar publications
listing the names of all state agencies. A concerted attempt to locate
such a reference is often an excellent investment of time. Given the
number of state agencies in most states and the diversity among
states, the discussion of state data sources presented here must be
very general. In most states, however, there will be the following
agencies:
state department of agriculture;
state education agency;
state employment commission;
state department of health;
state industrial commission, economic development
commission, or department of commerce;
state departments of community affairs,
human resources, or human services;
state library; and
state data center.
Nearly all of these agencies in every state will provide data in both
published and tape form and for geographical units down to the
county. Each of these data sources ·are briefly described below.
Most states have a state department of agriculture. These
department's generally publish some data on the production and
sales of major agricultural commodities in the state. Most such
departments also complete several cooperative programs with the
U.S. Department of Agriculture including the work of a State Crop

101
and Livestock Reporting Service which completes the surveys used
to project the yearly production of key commodities in the United
States, the number of farms, etc. Such departments are also excel-
lent sources of directories of the members of commodity and other
agricultural interest groups.
State educational agencies are generally excellent sources of
information on school enrollment, the number of teachers and
administrators, school facilities, educational and facility standards,
and educational financing for school districts. They often publish
directories, annual reports, and various types of statistical briefs
presenting such information.
State employment commissions, agencies, or their equivalents
exist in all states. They provide information on unemployment and
employment obtained from offices which assist persons seeking
employment. These agencies generally have data on employment
by sector for substate areas such as metropolitan areas or counties,
as well as data on wage levels and similar factors. Most also do
some forecasting of employment outlooks for various industries and
occupations in their states.
All states have state health departments. Such departments
have a number of regulatory and other functions but they are also
excellent sources of information. They provide data on births and
deaths, marriages and divorces, medical and health facilities (e.g.,
hospitals and nursing homes) and personnel, for county and other
substate areas. They can provide information on morbidity such as
the incidence of heart disease, cancer, sexually transmitted diseases,
etc., and on such health concerns as water quality and inspection
standards. This source of information is one of particular impor'.
tance for those analysts charged with demographic analyses involv-
ing an examination of trends in vital (i.e., births and deaths) events.
Nearly all states have an agency responsible for economic devel-
opment. They may be referred to as departments of economic
development or commerce, industrial commissions, or by similar
names. Their responsibilities usually include disseminating data on
areas where prospective firms and businesses might wish to locate.
As a result, they often have information on labor availability, busi-
ness establishments, and community services, as well as other
demographic and economic information·compiled in profile form for
substate areas. They are also often willing to assist a local analyst in
establishing a program to collect such data for use at both the state
and local level.
Human services agencies exist in every state. They are
charged with different levels and types of programs in different

102
states but among their basic responsibilities is that of providing
services to those with limited economic resources. As a result, they
usually can provide information on such services as the number of
persons receiving mental health services, the number of families
receiving Aid to Families with Dependent Children, supplemental
school lunch services and other forms of financial assistance, and the
number and location of various types of human service centers.
State libraries provide not only the standard services of a general
library, but also some relatively unique services. These libraries are
often charged with maintaining state data and other records for long
historical periods. If one has a need for historical data for an area
within a state, for histories of individual places within a state, and
for detailed information on legislative events in a state, the state
library is often an appropriate place to start. In addition, as noted
earlier, since most libraries also serve as repositories of publications
from agencies within a state, they are often excellent places to begin
ones search for critical items published by state agencies.
All states have a State Data Center. These centers were estab-
lished as a result of a joint agreement between the Bureau of the
Census and the Governor of each state. Under this agreement, the
state agrees to make census information readily available and acces-
sible to all its citizens and the Bureau agrees to provide free and
low-cost data and training in the use of such information to the
state's data center personnel. In most states, these centers consist of
a lead state agency and one or more affiliate centers in universities,
regional councils of governments, and other organizations. These
centers provide ready access to census data and most also house
state agency and other data. The Bureau of the Census Catalog for
recent years provides the names of agencies and contact persons for
each state data center in each state in the Nation. For a wide varie-
ty of data needs these centers are a good .first source. They are
likely to either have the information needed or know where to direct
a potential user. Recently the state data center concept has been
expanded to include business data. These Business and Industry
Data Centers (BIDCs) exist in about one-third of the States and
promise to expand access to business and industry data.
State agencies and organizations then, like federal agencies, are
major data producers and disseminators. Although the data from
such agencies are often limited to the state and its component areas,
state agencies are invaluable sources for describing the demographic,
economic, and social characteristics of many geographical areas of
interest to business and governmental analysts.

103
Nongovernmental Data Sources
In addition to those data provided by federal and state sources,
there are also a large number of nongovernmental (private and
nonprofit) data providers. Because the number of such providers is
so large and their range of services so diverse, the discussion here
will focus on a description of the types of services provided by such
providers rather than on descriptions of the services of specific
providers. Several publications are available, however, that describe
the services and data provided by such sources. These include the
Green Book, one of several publications that lists consulting and
research firms in the United States. A similar publication is The
Marketing Services, Organizations and Membership Roster, pub-
lished by the American Marketing Association. American Demo-
graphics is a popular magazine dealing with demographic matters.
Its advertisers include the major private vendors of demographic
data, it contains extensive information on private firms, and it pub-
lishes frequent directories on the services available from private-
sector firms such as the Directory of Microcomputer Data and
Software Analysis and its The Best 100 Sources for Marketing
Information: Who's Who from American Demographics.
Nongovernmental data providers offer a wide range of both
secondary and primary data and a wide range of services that can be
tailored to the specific needs of data users. They can generally
provide quick turnaround times in the production and delivery of
specialized analyses and can usually provide a wide range of serv-
ices including data analysis and interpretation. They are then often
excellent sources of information, particularly for the specialized data
user.
Although nearly any type of data-related service can be obtained
from such entities, it is useful to discuss the general categories of
services usually offered by nongovernmental sources. These cate-
gories are not exhaustive, but exemplary. Those to be discussed in-
clude
primary data collection,
secondary data manipulation and area profiling,
secondary data analysis,
on-line data location and manipulation,
compilation of industrial and corporate directories
and economic indicators, and
socioeconomic trend analysis and interpretation.

104
Many nongovernmental data providers offer direct data collec-
tion services, such as survey research services. Mail, personal inter-
view, and telephone survey services are provided by a wide range
of such firms. These services can usually be obtained for all survey
phases--questionnaire design, sampling, data collection, analysis,
and report preparation--or for one or more of these phases. When
data on individuals or small population areas are required, direct
surveys are often the only means available to collect the necessary
data.
A large number of firms and other groups are also actively
engaged in the manipulation of secondary data and in the produc-
tion of specialized data profiles from secondary data. Many of these
firms can provide software packages for manipulating census and
other data and can provide prepackaged data profiles for many
geographical areas. Such services are usually provided on a per
data item per area basis. Such firms may provide the user with
access to a larger base of expertise and other services than could
otherwise be supported by the user.
Many governmental data providers have neither the mandate
nor the resources to provide specialized analyses of secondary data,
but many nongovernmental firms specialize in such services. They
can usually provide services, such as determining market and service
areas and locating target populations, and can bring together data
from numerous sources to address specific issues. These entities can
often provide such services more expeditiously than governmental
agencies and can be specifically contracted to supplement the staff
resources of the contracting agency or firm. The flexibility of the
services available from such providers often makes their services
nearly ideal for specific data users.
On-line bibliographic reference and data systems are increasingly
common. Although selected governmental sponsored data bases
and bibliographies are often available from governmental agencies,
such as public or university libraries, many nongovernmental enti-
ties can provide access to numerous frequently updated public and
private data bases. In addition, an increasingly large number of
such entities allow the contracting user the opportunity to directly
manipulate selected data bases or search for selected data items.
The efficiency involved in obtaining such on-line access to a large
number of data bases from a single source often makes such services
particularly useful.
Yet an additional service provided by many firms and business
related associations consists of the compilation of specialized indus-
trial directories and mailing lists and the publication of selected sets

105
of economic indicators. Several entities (e.g., Dunn and Bradstreet,
Sales Management, a.nd the American Marketing Association) pro-
duce such indicators and directories periodically. For users involved
in specialized analysis, such directories can be useful.
Many nongovernmental groups provide ongoing analysis and
interpretation of national economic, demographic, and social trends
and describe their implications for general or specific industries and
interests. These services, usually provide~ through subscriptions to
periodically produced papers, are used by many corporate groups to
anticipate product trends and changing acceptability. Because pri-
vate-sector providers may have access to a wide range of data and
communication networks, they can often provide analyses of trends
of importance to businesses and other users whose own staff's are
heavily involved in day-to-day management. Nongovernmental
groups are often able to provide information of utility to the long-
term as well as the short-term planning and management needs of
data analysts.
Nongovernmental data sources play a crucial role in the provi-
sion of data and data analysis services. They provide a wide range
of services and provide flexibility and timeliness in data products.
Although the costs for the services provided by such groups often
exceed those from governmental entities, the range of services
provided is extensive and such groups represent sources of data
likely to be critical to many types of corporate and governmental
planning and applied analyses.
Using Secondary Data
The sources identified in the preceding parts of this chapter·
provide a wealth of data for addressing applied demographic issues
related to business, governmental, and other areas. Locating such
data, however, is only the first step in using them to address specif-
ic questions. Factors affecting data use is therefore a necessary and
complementary topic of discussion in any examination of data
sources and is discussed briefly below. Because individual uses of
data are likely to be specific to given types of users, however, the
discussion presented focuses on issues likely to be of importance to
nearly all uses-principles for data use and generic types of data use.
Because of the general utility of census data, such data are a major
focus in the discussion. The intent is to provide guidance applicable
to a wide range of data uses and users.

106
Prindples of Data Use
Although the criteria for applying specific types of data to
address specific informational needs vary widely for different types
of data, nearly all data uses will require that certain procedures be
completed. These procedures or prindples for data use are essential
to any application of census or other secondary data. These princi-
ples include the need to carefully
select the specific variables of interest,
determine the level of aggregation desired in data items,
select the geographical focus for data collection and use,
evaluate the comparability of areal units,
evaluate the comparability of the time referents for data
elements,
examine the definitions of key data items,
evaluate the likely accuracy of available data, and
determine the data form to be used.
The obvious first step in data use is to select those variables to
be used to address a given question. If one is collecting primary
data, this process consists of properly wording a questionnaire or
otherwise designing the data collection instrument to correctly solicit
the information desired. In secondary data use, however, the task is
one of discerning which of a large number of available variables
most closely measures the variables of interest. For example, do
you wish to measure wealth by income, value of housing, or owner-
ship of certain items (e.g., cars, televisions)? H you select income as
an indicator, do you use per capita income, household income,
family income, gross area income, or some other measure? All of
these income measures are readily available, but they measure dif-
ferent aspects of an area's level of wealth. In like manner, if you
wish to obtain information on groups of people residing in common
housing areas, should household or family data be collected?
Households and families are quite different conceptually, and the
selection of one or the other as the focus of data collection must be a
careful process. Although there is little general guidance that can be
given for this process, careful selection of variables is critical to
adequate data use.
Oosely related to the selection of the variables of interest is the
selection of the level of aggregation for which information is de-
sired. That is, is information on individuals or on areas desired? If
data on specific individuals is required, then it will generally be

107
necessary to use either microdata or conduct a primary data collec-
tion effort. If one's interest is in identifying the effects of different
personal characteristics on buying habits or service use characteris-
tics, then microdata will be necessary. However, if the characteris-
tics of a specific area are to be described, then aggregated areal data
will be sufficient. The guiding principle in selecting the level of
data to be used should be whether or not one is likely to incorrectly
describe the factor of interest by using a given level of data. That is,
if the use of aggregate data could lead to erroneous conclusions,
then data on individuals should be used.
A third requirement is the need to carefully select the area for
which data are required. Is the appropriate area for data collection
the metropolitan area, the city, the block group, census tract, or
some other unit? As self-evident as this selection may appear to be,
users are continually misled by data for units that appear by title,
but are not by definition, the appropriate area of interest. There is
simply no recourse to obtaining detailed knowledge of areal units'
definitions. For the user of census data, then, knowledge of census
geography is critical.
It is also essential to evaluate the comparability of different areal
units, particularly if data for several periods of time are to be used.
Definitions of areas, particularly metropolitan areas, change fre-
quently, and, as a result, the subareas included in them also change
frequently. In like manner, due to annexation, many places change
their boundaries {and thereby their populations) periodically. It is
essential to know the exact boundaries of the areas being analyzed.
Yet another factor requiring careful consideration is the time
referent of data items. Are 1990 data sufficient or are more current
data required? Can one use 1987 data with 1990 data, or should the
1987 data be adjusted to 1990 before analysis begins? Such ques-
tions must be carefully considered. Although decisions concerning
the timeliness of data are a constant topic of concern among ana-
lysts, they are sometimes over emphasized. Many characteristics,
such as the socioeconomic characteristics of a population, change
relatively slowly. Data that are ten years old may be too old to use
to assess the socioeconomic characteristics of an area, but data that
are several years old may be adequate if only general patterns and
relative differences among areas are of interest. One of the guiding
principles is the likely degree of change that has occurred in the area
of interest. If it is a rapidly growing or declining area, then recent
data are essential, while for slower changing areas, older data may
be sufficient. Yet an additional point of guidance can be drawn
from the nature of the characteristics being examined and their likely

108
rate of change. Income and cost data, for example, change rapidly,
particularly during inflationary or recessionary periods, and recent
data are essential for the adequate description of such variables. On
the other hand, the age composition of an area generally changes
relatively slowly. The time referents of available data and of the
data necessary to address a specific question must be carefully
compared.
An additional factor requiring careful examination is the defini-
tion of key data items, particularly if data for several periods of time
are to be used. Some definitions, such as the definition of a metro-
politan area, a farm or of the poverty level, have changed several
times in recent years. As with areal definitions and time referents,
the definitions of data items must be carefully examined for each
data use.
It is also important to consider the likely magnitude of errors in
the data for an area of interest. Many data items, even those from
censuses, are based on samples rather than complete counts and as
such their accuracy is subject to sampling, coverage, measurement,
and other types of error. The likely effects of such errors must be
carefully considered for they can affect both the selection of varia-
bles and of the geographical areas for analysis. For example, the
selection of the specific measure of income used, the choice of block
versus tract data, and similar decisions should take such potential
errors into account. Most data sources, such as the Census Bureau,
carefully describe such errors and provide range of error estimates in
the appendices of their publications.
Finally, economical and efficient use of data require careful
consideration of the form of data to be used. Data for demographic,
economic, marketing, and other analyses are available in an increas-
ing number of forms. These forms differ in their ease of acquisition
(and usually their costs as well) and in terms of the ease of manipu-
lating the data contained within them. Some of the major forms
listed in order of their ease of acquisition (from the easiest to the
more difficult to obtain) and their potential for manipulation (from
those that are most difficult for the user to manipulate for special-
ized analysis to those which possess numerous options for such
manipulations) are
published data,
data on microfiche,
data on floppy disks,
data on high-density computer tapes, and
data on laser or optical (e.g., CD-ROM) disks.

109
In obtaining data, it is important to evaluate the economic feasi-
bility of obtaining information in various forms. Printed data are
usually relatively inexpensive; for example, but they cannot be
manipulated to obtain information for forms of variables or for areas
not contained in the publication. Microfiche allows one to have
access to a larger volume of data than could easily be obtained (and
stored) in paper form but does not allow for manipulation of the
data. Floppy disks for microcomputers allow some manipulation but
such disks can store only a small proportion of the data contained
on a tape. Computer tapes contain more information but require
access to a large microcomputer, a workstation, or a mainframe
computer. Laser and compact disks which can be accessed by
microcomputers and which can store as much data as four high-
density computer tapes are revolutionizing access to data for use on
microcomputers, but require readers as well as sufficient computer
memory to manipulate the data.
The key factors in choosing the form of data to be used are the
potential frequency of use and the need to manipulate the data for
addressing the question(s) of interest. If the data are to be accessed
for a one-time use, and the data are for the area and in the form
needed, it may be more cost-effective to use paper forms of the
data; while computerized forms are likely to be cost-effective if the
data must be accessed frequently and manipulated.
Each of the factors discussed above are performed relatively
routinely by most analysts, but even experienced analysts can occa-
sionally forget one of them with negative consequences. Their
systematic consideration (e.g., the use of a checklist of such princi-
ples) must be a standard part of the data use process. Many data
providers are increasingly aware of the need to address such con-
cerns in their publications. Census publications generally contain at
least four elements that address such concerns. That is, most census
publications contain a table-finding guide that shows the variables
covered in a publication and the geographic areas for which such
data are available. One appendix provides definitions of the areal
units used in the analysis. A second appendix provides definitions
of variables (e.g., of a farm or the poverty level), and a third appen-
dix presents data on the sampling and other potential errors in the
.data provided in the publication. Similar information can and
should be obtained on all data one wishes to use.
In sum, then, any user of data from the sources described in this
chapter should attempt to address the principles described above.

110
If these and other basic principles of data use are maintained, the
utility of such uses will be increased substantially.
Examples of Data Use in Addressing
Topics of Applied Analyses
Having described a variety of data sources and factors that must
be considered in the use of such data, it is useful to conclude this
discussion by indicating how the data noted above can be used to
address the types of analyses in which applied demographic analysts
are often involved. The intent is to assist the reader in recognizing
how different categories of the data provided by data sources noted
above can be used to address pragmatic issues. The uses to be brief-
ly discussed here include
· area profiling,
· determination of market potential,
· determination of market penetration,
· facility siting,
· program and data evaluation,
· product feasibility analysis, and
· projections of future markets and facility requirements.
One of the uses for which data such as those described above
are beneficial is in profiling the characteristics of an area's popula-
tion. Such profiles are, in turn, a useful source of general guidance
for marketing and management decisions, particularly if profiles are
created which show trends in characteristics across time, as well as
the characteristics of the area for a recent point in time. They can
serve as an initial cost-effective means of screening areas for facility
placement or product and service marketing. In this regard, data
compilations such as the County and City Data Book can often
provide basic profiles. In addition, with the ready availability of
computerized data, standardized software routines can be written to
provide quick compilations of such profiles. Such profiling is a
useful first step for many business and other applied analyses.
One of the most frequent uses of such data is in determining the
potential market for a particular product or the service population
for a particular service. By taking the size of populations from
decennial censuses or P-25 and P-26 estimates for given areas with
specific characteristics and applying estimates of the number of
purchases or clients per unit of population, the potential market for

111
a particular product or for a particular service can be roughly deter-
mined. In addition, data on a firm's share of the market in areas
presently being served can often be obtained from such censuses as
the census of retail trade. Given such data, one can make estimates
of the potential share of that market that a firm or agency might
receive (its market share).
A related use of such data is in the determination of market
penetration--the extent to which a firm or agency is obtaining the
desired proportion of sales of a given product or clients for a given
service. By using data from the censuses of business on the number
of firms of a given type in an area and data on total sales of the
products of such firms, one can determine whether a firm's market
share is as expected. Similarly, by examining the client populations
of other service agencies (usually available from state data sources)
and data on total expected and actual clients for a given service, the
extent to which a service agency is serving the needs of a given
clientele can be discerned.
Such data can also be useful in selecting the site for a facility or
business. By using census maps in conjunction with data on the
number of firms or agencies of different types in different areas from
the censuses of business and data from the census of population to
determine client or customer type, the best location for a facility can
be determined. Nearly every facility siting of a major business or
corporation relies on the use of such data, but the needs of even a
small business concern can often be met with such data because of
its relative low costs.
Census and other data can also be used .effectively to evaluate a
firm or agency's programs and to evaluate data obtained from
various sources. Given even limited data on the location of a firm's
or agency's clients or customers and their characteristics, and census
data on the characteristics of the population in the firm or agency's
primary service areas, it is possible to discern how its clientele
compares to the general population, and to identify additional types
of clientele who might be served (e.g., additional product markets).
In addition, census and other data can easily be used to evaluate the
likely representativeness of data contracted for or purchased from
another entity. If the characteristics of the respondents of such a
study differ significantly from those of the population or firms de-
scribed in the most recent population and economic censuses of an
area, then such data should be closely examined.
Another use of such data is in product feasibility analyses. That
is, determining whether there would be a market of customers with

112
the resources necessary to purchase a product if it could be pro-
duced at a given price. By using data on income from such sources
as the Bureau of F.conomic Analysis, particularly disposable income,
population, and the sales of particular products and other data from
recent censuses or surveys, one can discern the likely market for a
product in a given area or select a market area where such a product
might be marketable.
Finally, census data in conjunction with other economic data and
vital statistics data can serve as a base for projecting future markets
and facility requirements. By examining trends in population, sales,
business growth, and the projected characteristics of a population at
specific points in time in the future, one can estimate future mar-
kets. Many of the nongovernmental data sources noted above are
excellent sources of such projections. In addition, by examining
such data in conjunction with census maps and other data, it is
possible to identify potential future facility sites. For long-term as
well as short-term analyses, then, census and the other data de-
scribed above are of critical importance.
Summary and Conclusions
A large number of federal, state and nongovernmental concerns
produce data likely to be of utility for applied demographic analyses.
Although the magnitude of available data and its varied levels of
areal and temporal coverage are often confusing to even the sophis-
ticated data user, there are numerous indices and general compila-
t.ions of data that can assist one in identifying particular items of
interest.
Such data can be instrumental in addressing applied questions if
care is taken to adhere to certain principles in data use so that the
conclusions arrived at are accurate and correctly focused on the
topics and areas of interest. As a result, any investment made by an
analyst in obtaining detailed knowledge of such data is likely to be
well rewarded.

4
Basic Methods and Measures of Applied Demography
In this chapter, we examine several general methods and meas-
ures used in applied demography. These are methods and measures
that are used in descriptive studies and are basic to nearly any
demographic analysis. Although many are very simple measures
with which readers may already be familiar, they are essential to
any analysis and necessary for understanding the remaining materi-
als in this work. We begin with several general measures that are
used in the analysis of many different demographic factors and then
discuss measures as they relate to each of the individual demograph-
ic concepts and variables outlined in Chapters 1 and 2.
General Measures
The Use of Rates
Among the most basic measures in demography is the meas-
urement of rates of incidence and change. Perhaps the most widely
used of all measures of change is simply the percentage change from
one period to another. As shown in Figure 4.1, this is expressed as
the amount of increase or decrease in population per 100 persons.
This percentage change measure is thus a rate per 100 persons in the
population.
Rates are the most basic measures used to evaluate the incidence
of demographic factors and processes. Rates measure the relative
frequency of occurrence of an event in a population. In demographic
analyses, the most common form of a rate is simply a numerator
consisting of a number of events for a given time pe~od divided by
a denominator which is the population experiencing or exposed to
the risk of the event during the same time period as the occurrence
of the event. The value obtained after the numerator is divided by
the denominator is then multiplied by a constant, such as 1,000.
This constant places values on a common base and eliminates the
need to use small decimal values.

Percent
Change
Example:
Percent
Change
U.S.
Population
1980-1990
Figure
4.1:
Percentage
Change
in
Population
(Population
at
t
2
-
Population
at
t
1
)
---------------~
x
100
(Population
at
t
1
)
248,709,873
-
226,545,805
-
-
-
-
-
-
-
-
-
-
-
x
100
9.8%
226,545,805
......
......
II-

115
Because demographic events are measured for discrete time
periods and because populations change over time, both the numer-
ator and denominator for rates are often adjusted. In the numera-
tor, the most common adjustment is to take an average number of
events for several years rather than a single year. For example, a
birth rate for 1990 might employ a numerator that was the mean or
arithmetic average number of births for 1989, 1990, and 1991. This
is done because there can be substantial year-to-year fluctuations in
the number of events, and one wishes to obtain a rate that indicates
the usual incidence of an event in a population. For an area with
a small number of events, year-to-year fluctuations can lead to very
misleading rates if the time at which the events are measured is an
unusual period.
A choice of denominators is also likely to be required. That is,
rates are variously computed with denominators which are popula-
tion values at the beginning of the period of interest (e.g., 1980 in a
1980-to-1990 rate), at the midpoint (e.g., 1985 for a 1980-to-1990
rate), or at the end of the period (e.g., 1990 for a 1980-to-1990 rate).
When the beginning of the period population is used, the rate
expresses change in the event relative to the beginning population
base. The midpoint population (usually obtained by using an aver-
age of a beginning and end of period population) is the most often
used to compute basic rates and represents an attempt to measure
the average number of persons at risk of the event. The endpoint
population is often used to assess change relative to the population
remaining after a period of change. Net migration rates (discussed
below), for example, are often based on expected populations which
are end-of-period populations. Whatever procedure is used to
obtain the numerator or the denominator, it is essential that all rates
to be compared for various areas use values for equivalent time
periods.
Three types of rates are commonly employed in demographic
analyses. They are used to measure the incidence of demographic
processes and numerous other factors as well. These three types of
rates are crude, general, and specific rates. These rates are shown in
Figures 4.2 through 4.4. They differ in the extent to which they
measure an event relative to the population at risk of the event.
That is, a crude rate measures the occurrence relative to the total popula-
tion, only part of which is actually subject to the risk of experiencing
the event. For example, births occur only to females of certain ages,
while the crude birth rate shown in Figure 4.2 measures births rela-
tive to the total population. Crude rates can be misleading if a
population is composed of a disproportionate number of persons

Figure
4.2:
Crude
Rates
Number
of
Occurrences
Crude
Rate
X
Constant
Total
Population
Example:
To
compute
the
crude
birth
rate
for
the
United
States
in
1990
Crude
Birth
Rate
(CBR)
Number
of
Births
Total
Population
x
1000
CBR
U.S.
1990
[(Births
in
1989
+Births
~n
1990
+Births
in
1991)]
~~-=~~~~~~~~~~~~~~~~~~~~~~~~=--~~
x
1000
U.S.
Population
in
1990
[(4,021,000
+
4,17:,ooo
+
4,343,208a)]
248,709,873
4,181,069
x
1000
=
16.8
248,709,873
aValue
for
1991
estimated
using
values
for
1989
and
1990.
x
1000
.....
.....
°'

117
with or without the characteristics likely to lead to their experiencing
the event. As the name implies, crude rates only crudely measure
the frequency of occurrence of the phenomenon in a population.
General rates, such as the example shown in Figure 4.3, more
closely limit the measurement of the base to those persons actually at risk of
the event. The general fertility rate shown in this figure is a rate per
1,000 women in the ages in which child-bearing is most likely to
occur, 15-44 years of age. Specific rates, such as that shown in Figure
4.4, show the greatest specificity measuring events relative to the specific
population at risk. Thus, the events shown in this figure are the
births to women 20-24 years of age relative to the number of women
20-24 years of age. The advantage of the use of specific rates is
clearly that they more exactly measure the events relative to those
persons most likely to actually experience them. If the data are
available to obtain specific rates, they are usually pteferred because
they are less likely to mislead one relative to the incidence of the
phenomena in the populations of interest.
Descriptive Statistical Measures
Numerous widely used measures from general statistical analysis
are also commonly applied in demographic analysis to measure the
characteristics of the distribution of a variable within a population.
Among these are the three measures of central tendency, the mean,
the median, and the mode. The mean or simple arithmetic average is
widely used to measure demographic factors (e.g., age, income, etc.).
The advantage of using the mean is that its properties are well-
known statistically and associated measures, such as the variance
and standard deviation and measures of statistical significance, can
be used to describe the characteristics of a distribution. The mean is
often replaced in general analyses with either the mode or the
median because the value of the mean can be skewed by extreme
cases, while the mode and median are not affected by extreme
values. The mode simply indicates the value occurring most often, while
the median is the value that divides a ranked distribution in half (with 50%
above and 50% below the median value). Which of the three measures
should be used depends on the nature of the distribution and the
norms of use in an analytical area. The median is normally used to
describe age and income, the mean to describe such factors as age at
first marriage, and the mode to refer to such factors as the most
common occupation of employment in an area. Other descriptive
statistics and procedures such as histograms, graphs, charts, fre-
quency distributions, etc. are also widely used. The applied

118
Figure 4.3: General Rates
Number of Occurrences
General Rate
Population at Risk
Example:
Number of Births
General Fertility Rate
(GFR) Females 15-44
GFR (U.S. 1990)
4,181,069
~~~~- x 1000
58,483,000a
x 1~00
71. 5
aValue is derived from the 1989 population estimate for the United States
(Hollman, 1989).

Specific
Rate=
Example:
Age-Sex
Specific
Fertility
Rate
(ASFR)
for
Women
in
Age
Group
ASFR
for
U.S.
Females
20-24
(1990)
Figure
4.4:
Speciflc
Rates
Number
of
Occurrences
to
Persons
with
Specific
Characteristic
~~~~~~~~~~~~-
X
Constant
Number
of
Persons
with
Specific
Characteristic
=
Births
to
Women
in
Age
Group
------------~x
1000
Women
in
Age
Group
1,060,035
-------
x
1000
9,356,000a
113.3
aEstimated
from
Hollman
(1989)
and
Spencer
(1989).
......
tO

120
demographic analyst must thus develop a basic knowledge of gener-
al statistics as well as knowledge of methods unique to demography.
Each of these general measures is used in numerous areas of
demographic analysis and many will appear repeatedly in the
examples provided below. Although they are not measures unique
to demography, familiarity with them should not lead to hesitancy
to use these measures when they are appropriate. They are often
the most appropriate measures for an analysis.
Measures of the Major Demographic Processes and Variables
Population Change
In addition to the percentage change measure discussed in the
section on general measures, there are several other widely used
measures of population change. Among these are the arithmetic,
geometric, and exponential rates of change shown in Figures 4.5
through 4.7. In each of these figures the base measures from which
the rate is derived are shown as well as the basic formula for deter-
mining the rate of change. Because one may want to compare
change in areas using data for different lengths of time, the meas-
ures are shown on a per-year basis.
As is evident in these figures, arithmetic change is simply the
numerical change betwe.en two populations at different points in time. The
geometric rate of change is that determined 'by the compound interest formu-
la familiar to those who haoe calculated interest rates for financial analyses.
It computes rates based on fixed intervals of time. By contrast, the expo-
nential rate of change is based on continuous compounding. It is the rate
that is most commonly referred to when rates of population change
for areas are discussed because its continuous process characteristic
most closely simulates the continuous nature of demographic change
(e.g., the continuous patterns of births and deaths in a population).
When the world's population is indicated as increasing at a rate of
1.8 percent per year, it is the exponential rate of change in the
world's population being referenced.
One widely used means of describing rates of change is in terms
of the number of years it would take for an area to double its exist-
ing population at its present rate of change. By solving the expo-
nential formula for time (t) in years with the population set at
double its existing size (that is, at 2P), one finds that the formula
produced shows that the doubling time can be determined 'by dividing the
rate of change per year into 0.6932. Since no matter what the size of
the population, the value 0.6932 will be obtained by solving the

Figure 4.5: Arithmetic Rate of Change
pt pt + bn
2 1
b
r =
Where: population at second date
Pt = population for a base date
1
b mean annual numerical change
n years between base date and
second date
Example: To obtain arithmetic change for
the United States for 1980 to
1990
Given: U.S. Population, 1990
U.S. Population, 1980
248,709,873
226,545,805
b (248,709,873-226,545,805)/10
b 2,216,407
r = 2,216,407/226,545,805 x 100
r = .0098 x 100
r = 0.98
121

122
Figure 4.6: Geometric Rate of Otange
(1) P = Pt (l+r)n
t2 1
and
Where: pt
pt
popu 1at ion at the estimate
2 date
population at the base
1 date
r = rate of change
n = number of years between
base and estimate date
Example: To obtain geometric rate for the
United States for 1980 to 1990.
248,709,873
)
r = 1
226,545,805
r = 1.0093777 1
r = 0.0093777 x 100
r = 0.90

123
Figure 4.7: Exponential Rate of Change
(1) pt
2
pt
1
ern
and
(2)
,....c::)
r =
n log10e
Where: pt population at time 2 (t2)
2
pt population at time 1 ( t 1)
1
e a constant (2.71828)
r rate of change
n time period
between tl and t2
Example: To ob ta in U.S. change from 1980
to 1990
l ( 248,709,873)
oglO
226,545,805 .0405370
0.009333
r =
10 (.4342942) 4.342942
r = 0.009333 X 100 = 0.93%

124
exponential formula for 2P, one can always find the doubling rate of
a population by dividing its exponential rate of growth into 0.6932.
Thus, the doubling period for the world's population at an annual
growth rate of 1.8 percent per year would be 38.5 years (0.6932
divided by an exponential rate of growth of 0.018). When expressed
in rates per 100 so that 0.6932 becomes 69.32, this relationship is
sometimes referred to as the rule of 69, because one can obtain the
doubling period by dividing the annual rate of growth (expressed as
a percent) into 69. (This rule is also sometimes referred to as the
rule of 70, since if the geometric formula of change is used to com-
pute the doubling rate, the value obtained is 0.6968 which, when
multiplied by 100 and rounded, is 70). Whichever form is used, the
doubling period is a quite useful way of describing the implications
of a specific rate of population change.
Measures of the Demographic Processes
The three demographic processes of fertility, mortality, and
migration use many of the general measures noted above as well as
several unique measures. Crude, general, and specific rates are
widely used to describe the processes. The three rates shown in
Figures 4.2 through 4.4 are the rates most often used to measure
fertility. The number of births used in the values is obtained from
vital statistics departments in state departments of health or from
the National Center for Health Statistics with births being those by
the place of residence of the mother (rather than place of occur-
rence). For mortality and migration, crude and age-specific rates are
used with both applying to such events by place of residence.
However, since all persons are subject to the risk of death and of
migrating, there is no counterpart to the general rate for mortality or
migration.
Fertility Measures. In addition to the crude, general, and specific
fertility rates shown in Figures 4.2 through 4.4, two additional
measures will be examined here. These are the child-woman ratio
and the total fertility rate. The formula for the child-woman ratio is
shown in Figure 4.8. This ratio simply shows the number of persons O-
to-4 years of age divided by the number of females of child-bearing age
(note that in this example we have used women aged 15-49 as being
those of child-bearing ages; whereas 15-44 was used in Figure 4.3,
alternative ages from 10 or 15-to-44 or 49 are variously used to
indicate women in child-bearing ages). This rate is only generally
indicative of fertility levels in a population because both mortality

Example:
Figure 4.8: Child-Woman Ratio (CWR)
Chi Id-Woman Ratio =---- X 1000
Where: P0 _4 population ages 0-4
Pp 15 _49 females ages 15-49
CWR (U.S. 1990)
18,408,000a
65,872,000a
x 1000 279.5
3Estimated using data from Spencer (1989).
125

126
and migration may also have affected the number of persons ages 0-
4. The reason for including this rate in this discussion is that it can
be computed using only census or other count data and does not
require vital statistics data as do the other rates shown above. As
such, it is often useful for measuring fertility in small areas for
which vital statistics data are not available.
Perhaps the most widely discussed measure of fertility is the
total fertility rate shown in Figure 4.9. It is the sum of the age-specific
fertility rates for all women in the child-bearing ages, and when adjusted to
be per-person-specific, indicates the number of children that the average
woman would have in her reproductive lifetime if she aged through her
reproductive years exposed to the age-specific rates prevailing at a specific
point in time. In the example shown in Figure 4.9, the rate indicates
that the average woman would have had 1.96 children during her
reproductive lifetime. Among the most widely discussed levels of
total fertility is the rate of 2.1, referred to as the replacement rate of
fertility. This is the total fertility rate that must prevail in a popula-
tion (with survival rates similar to those of the United States) for it
to replace itself because the average woman must replace both
herself and her mate. The value required for replacement is slightly
larger than 2.0 because some children do not survive to reproductive
age.
Mortality Measures. In addition to the crude death rate and
age-specific death rates delineated above, the measurement of the
incidence of death in a population tends to center on the incidence
of deaths at certain ages, on the causes of death, and on the effects
of a given set of death rates over the life-cycle of a population.
Death rates among infants are of particular interest because
infant mortality is often indicative of the general level of health care
in a society and because, as noted in the discussion in Chapter 2,
the death rate is higher during the first year of life than for any
other age prior to about age 55. Figure 4.10 presents three widely
used measures of infant deaths. Infant mortality is simply the number
of deaths occurring to persons less than one year of age. Since persons
less than one year ofage are those born during the last year, the
number of infant deaths (i.e., those deaths to persons less than one-
year of age) is divided by the number of births to obtain the infant
mortality rate. The infant mortality rate is also often examined
in terms of two components, deaths to infants less than a month old, re-
ferred to as the neonatal death rate, and deaths to infants one month to one
year of age, referred to as the post-neonatal death rate (see
Figure 4.10). The reason for the use of these two rates is that deaths

127
Figure 4.9: Total Fertility Rate (TFR)
i=45-49
Total Fertility Rate (N ) :E ASFR. X 1000
j i=15-19 I
Where:
Example:
For
the
age group
ASFRi = age-sp~cific fertility rate for age
group 1
Ni number of years in age group i
Age ASFRa
1990 for 15-19 52.4
United States 20-24 113. 3
25-29 117 .5
30-34 = 76.1
35-39 .. 27.1
40-44 5.2
45-49 .. 0.2
E= 391.8
TFR = 391.8 X 5 1959.0
TFR per woman = 1.959
aValues from Spencer (1989).

128
Figure 4.10: Selected Measures of Infant
Mortality
Infant Mortality Rate (IMR)
D0-1.
IMR - - -
1- x 1000
Where: D0_1.
I
deaths to persons less than
one year of age during year
Bi births during year i
Neonatal Mortality Rate (NMR)
NMR
Dl month.
~~~~~
1- X 1000
B.I
Where: D
1 month.
I
deaths to persons less
than one month of age
in year i
Bi = births in year
Post-Neonatal Mortality Rate (PNMR)
PNMR
01-12 monthsi
Bi
Where: Dl-12 months.
I
x 1000
deaths to person one
month to one year of age
in year i
Bi births in year

129
to infants less than a month old are often related to problems
in gestation and to such factors as the level of prenatal care received
by the mother during pregnancy. Post-neonatal mortality is likely to
reflect post-birth environmental factors rather than problems related
to gestation. Whatever measure of infant mortality used, it is some-
times necessary in areas with substantial year-to-year fluctuations in
births and infant deaths to adjust the numerators and denominators
of infant mortality rates to ensure that infant deaths in a year are
being measured relative to the correct base of births. In such cases,
it is necessary to use separation factors to separate births and infant
deaths into comparable annual periods (see Shryock and Siegel,
1980: 412).
Because different causes of death are more likely to occur to
persons in certain ages and to persons with different socioeconomic
characteristics, there is also considerable interest in the incidence of
deaths by cause. The cause-specific death rate, defined as the number of
deaths from a given cause in an area divided by the population of the area,
is commonly used. Such analyses generally show coronary disease
and cancer to be the major causes of death in nearly all areas of the
United States.
Among the most unique techniques used to measure the impacts
of mortality is the set of procedures referred to as life-table analysis.
Life-table analysis is a procedure that simulates the impacts of a given set of
age-specific mortality rates on a population over the entire lifetime of the
population. It simulates how many persons would die at each age
until the last person in the population dies. A hypothetical popula-
tion of 100,000 (called the radix) is used with elements of the table
being computed for each age. Figure 4.11 provides an example of a
life table and Figure 4.12 briefly defines the standard elements of a
life table. The n prefix before these elements and the x suffix after
them refer respectively to the size of the age groupings being
examined and to the initial age (x) of the age group being consid-
ered (e.g., 511.5 would refer to the five-year age group of 15-19).
As shown m Figures 4.11 and 4.12, a life table contains informa-
tion on the proportion of persons dying and the number living at
each age, given a particular set of age-specific mortality rates. Two
particularly important elements of the table are the nLx and the gx
values. The latter represents the value for life expectancy at age x
with the value at birth (or age 0) being commonly referred to

......
Hgure
4.11
Abridged
Life
Table
for
the
Male
Population
of
a
Hypothetical
Area,
1990
~
Number
Total
Person
Average
Living
Number
Number
of
Years
Lived
N.umber
of
Years
Age
Proportion
at
Dying
Person
in
This
and
of
Life
Remaining
Interval
Dying
In
Beginning
of
During
Years
Lived
All
Subsequent
at
Beginning
of
(in
Years)
Interval
Age
x
Interval
In
Interval
Ages
Age
x
(nqx)
(ndx)
0
x
to
x+n.
(
1
x)
(nLx)
(TX)
(ex)
0
-
1
0
.01152
100,000
1,152
98,951
7,534,601
75.35
1
-
5
0.00214
98,848
212
394,891
7,435,650
75.22
5
-10
0.00129
98,636
127
492,824
7,040,758
71.
38
10
-
15
0.00140
98,509
138
492,227
6,547,934
66.47
15
-
20
0.00465
98,371
457
490,918
6,055,707
61.56
20
-
25
0.00630
97,914
617
487,996
5,564,789
56.83
25
-
30
0.00695
97,297
676
484,829
5,076,793
52.18
30
-
35
0.00779
96,621
753
481,298
4,591,964
47.53
35
-
40
0.00908
95,868
870
477,297
4,
110,666
42.88
40
-
45
0.01155
94,998
1,097
472,468
3,633,369
38.25
45
-
50
0.01741
93,901
1,635
465,666
3,160,901
33.66
50
-
55
0.02913
92,266
2,688
455,019
2,695,235
29.21
55
-
60
0.04756
89,578
4,260
437,671
2,240,216
25.01
60
-
65
0.07326
85,318
6,250
411,594
1,802,545
21.13
65
-
70
0.10299
79,068
8,143
375,391
1,390,951
17.59
70
-
75
0.15271
70,925
10,831
328,630
1,015,560
14.32
75
-
80
0.22217
60,094
13,351
267,759
686,929
11.43
80
-
85
0.31761
46,743
14,846
196,598
419,170
8.97
85+
1.00000
31,897
31,897
222,572
222,572
6.98

Figure 4.12: Elements of a Life Table
x to
x + n - the period of life between two exact ages.
(x and x + n where n - age interval)
n'lx
n~
- the proportion of the persons in the age group
alive at the beginning of an indicated age
interval (x) who die before reaching
the end of that age interval (x + n).
- the number of persons living at the
beginning of the indicated age interval (x)
out of the total number of births assumed as
the radix of the table.
- the number of persons who die within
the indicated age interval (x to x + n).
- the number of person-years lived
within the Indicated age Interval (x
to x + n) by all persons from age x to x+n.
- the total number of person-years
lived after the beginning of the
indicated age interval.
- the average remaining lifetime (in years)
for a person who survives to the beginning
of the indicated age interval. This ls
also referred to as life expectancy.
131

132
as simply life expectancy rather than by its full title which is life ex-
pectancy at birth. The nLx values are used to compute survival
rates which are widely used in mortality analysis.
Life tables vary in form and in coverage. Complete life tables
are computed for single years of age from 0 to 1, 1 to 2, etc. to some
terminal age, such as 75 years of age and older. An abridged life
table uses 5-year or some other set of multiple-age categories. Life
tables may also examine only the effect of rates of transition from
life to death (in which case they are referred to as single-decrement life
tables), or the effects of mortality and one or more other factor(s),
such as labor force participation, first marriage rates, or school
enrollment rates (in which case they are referred to as multiple-
decrement life tables). Life tables are usually used to trace the implica-
tions of a set of rates for a given point in time. This type of life table
is properly referred to as a current or period life table. A life table
can also be constructed using rates derived from the historical expe-
riences of a cohort as it aged over time. This table is referred to as a
generation or cohort life table.
Life tables can also be used to examine the implications of a set
of death rates for a stationary population (that is, a population with
an equal number of births and deaths). In this use, rather than
being seen as providing a mortality history of a group of people as
they pass through life, it is seen as the history of a population with
a new birth cohort of 100,000 persons entering the population each
year and 100,000 persons dying each year. The nLx values can then
be seen as indicating the number of persons who would be in each
age group in a population rather than the number of person years
lived by all pers0ns in an age group. Analyses of stationary popula-
tions are useful for determining the age composition that a popula-
tion would assume over time if exposed to a given set of age-specific
mortality rates.
Although it is not possible here to provide all of the computa-
tional components of a life table, if is important to describe how a
life table is constructed so that its elements can be more adequately
understood. To construct a life table, a set of age-specific death
rates referred to as nmx values in life-table descriptions must be
converted to nqx (proportion dying) values. The difference in these
values is that an age-specific death rate indicates the relative fre-
quency of occurrence of death in a population while the nqx value
refers to the proportion dying. Age-specific rates are computed by
dividing the number of deaths to persons of a given age by the
number of persons of that age in a population, while the proportion
dying value of q is obtained by dividing the number of deaths to

133
persons of a given age by the number of persons of that age plus
the number who have died. The denominator for the nmx values is
simply p, the population, while the denominator for the nqx value is
p, plus the deaths during the period. The nqx value is a measure of
the probability of dying during any given age interval. There are a
number of procedures for converting nmx values to nqx values
including several widely used standard tables (e.g., Reed-Merrell
and Greville's methods, see Shryock and Siegel, 1980 or Namboodiri
and Suchindran, 1987 for descriptions of these procedures).
However, with access to computers, nqx values can be readily
computed from nmx values using standard formulas (again see
Shryock and Siegel, 1980; Pollard et al., 1990).
All other elements of the life table can be derived from the nqx
values. That is, the Ix value (the number of persons still living at
the beginning of a specific age) is the difference between the
number beginning a preceding age interval and those who died
during the interval. The ndx value is simply the number dying
during a time interval and can be obtained either by multiplying the
nqx value for an interval by the Ix value at the beginning of the
interval or by subtracting two adjoining Ix values. The large nLx
value is the total number of person years lived (a person surviving
from 20 to 25 would have lived 5 person years) by all of the persons
in the age group during the interval and is obtained by averaging
the Ix and Ix+ n values, it being assumed that those who died
during the interval lived on average one-half of the x to x+n period.
The Tx value is the sum of all remaining years to be lived by all
persons in the population at the age of interest and is obtained by
adding all of the nLovalues from the bottom of the table to the
age of interest. The ex value is the average remaining lifetime in
years for a person at age x and is obtained by dividing the Tx value
for a given age by the Ix value for that age.
Life table techniques are, in fact, techniques with wide applica-
bility. They are used to determine the lifetime loss in earnings for
disabled workers, to examine the implications of school enrollment
rates on enrollment patterns over time, and to determine the
number of persons married at different points in time as a result of a
set of age-specific marriage rates. Furthermore, life-table techniques
are not limited to such demographic uses but allow one to examine
the lifetime depletion of any population or part of a population and
can be applied to the aging of any factor, good, or service that is
likely to be depleted over time. For example, such techniques might
be used to examine the aging and eventual demolition of housing

134
stock. Life-table techniques can be used to examine the implications
for a population of any set of age-specific death (or other form of
depletion) rates.
Among the most frequently used elements of a life table is the
nLx function which can be employed to compute life table survival
rates. The formula for this rate is shown in Figure 4.13. Figures
4.14 and 4.15 provide examples of the computation of these values
for middle, beginning, and terminal age groups. As the formula
and the examples suggest, the life table survival rate indicates the
probability of surviving from one age to another. It is a widely used
measure of the effects of mortality on a population and appears as a
component measure in many forms of demographic analysis.
Migration Rates. Migration is a difficult process to measure in
the United States because there is no direct registration system (as in
some nations) that requires residents of the United States to indicate
when they change residences. Migration must be measured primari-
ly with data from the decennial census, from periodic surveys, or by
indirect methods. In the decennial census, migration is measured
by asking respondents their place of residence five years ago. If
their residence at the time of the census and five years earlier are
different, respondents are considered to have moved, but if the
respondent's residence at the time of the census and five years earli-
er are in different counties (or countries), the respondent is then
considered to have migrated. As noted above, persons are consid-
ered to have inmigrated or outmigrated from the standpoint of a
given area, while the difference between inmigration and outmigra-
tion for an area is referred to as net migration. Figure 4.16 shows
several rates commonly employed to measure migration.
Migration is usually measured using what are referred to as
residual methods. These methods are equivalent to solving the
bookkeeping or population equation for the migration component.
Thus, the amount of total population change attributable to births
and deaths is accounted for and the remaining difference between
the total change and that due to births and deaths--the residual--is
assumed to be migration. Figures 4.17 and 4.18 show alternative
forms of the net migration formula for computing residual measures
of net migration.
Figure 4.17 shows the formula directly derived from the book-
keeping equation. This equation provides the formula for comput-
ing the net migration rate for the total population. This formula
could also be used to compute the rate for individual ages (with

s
Figure 4.13: Ufe Table Survival Rates
·---
x, x + n
Where:
sx, x + n - probability of a member of an age group surviving
from the time period, x to x + n
Lx + n - number of persons alive at the end of the period,
x+n
Lx ... number of persons alive at the beginnJng of the
period, x
Example: To obtain life table survival rates for ages 15-20 to 25-30
(given values In Figure 4.11).
L2s.30
515-20, 25-30 =
L15-20
484,829
00,918
0.98759
135

136
Given: s
Use: s
Example:
Figure 4.14: Procedure for Computing Survival Rates
for Multi-age Age Groups from a Life
Table for Single-Year Age Groups
x, x + n
x, x +
mean of L for
age in ye~r x + n
mean of L for
age in ylar x
Given Lx values as follows:
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
98,123
98,048
97,959
97,857
97,739
97,608
97,461
97,303
97,140
96,978
96,818
96,662
96,508
96,357
96,205
s
To compute survival rate from ages 15-19 to 25-29
(96,818 + 96,662 + 96~508 + 96,357 + 96,205)
x, x + n
sx, x + n
sx, x + n
(98,123 + 98,048 + 97~959 + 97,857 + 97,739)
96,510
97,945
0.98535

Figure
4.15:
Procedure
for
Computing
Beginning
and
Termlnal
Age
Survival
Rates
Beginning
Age
Group
L
xl
5
0,1
100,000
L
L
L
L
L
x
1
+
x
2
+
x
3
+
x
4
+
XS
5
o,s
..
Tennlnal
Age
Group
sx
Example:
Tx
+
n
TX
500,000
(~I
L
L
L
L
x
1
+
x
2
+
x
3
+
x
4
n
or
100,000
+
Lx5)
To
compute
survival
rates
for
persona
in
the
United
States
for
1980
for:
A.
0-1
years
of
age
B.
0-5
years
of
age
C.
65+
years
of
age
(for
a
10-year
period)
(continues)
~

Figure
4.15
(continued)
Given:
For
Age
0-1:
98,973
so,1
'
10
0,000
For
Age
65+:
L
xl
L
Lx2
X3
L
X4
5
65+
=
595,390
1,273,347
98,973
L
=
98,515
XS
98,694
T
=
1,273,347
x65
98,617
Tx
+
n
=
595,390
(if
for
10
years)a
98,560
For
Age
0-5:
98,973
+
98,694
+
98,617
+
98,560
+
98,515
0.98973
5
o,s
=
493,359
500,000
.986718
0.467579
(if
for
10
years)•
500,000
a
For
terminal
age
groups
the
numerator
used
is
the
Tx+n
value
for
the
time
period
n
years
after
the
terminal
date
used
for
analysis.
The
numerator
CTx+n
value)
Jn
this
example
is
that
for
75+,
but
had
a
different
time
period
been
used
(such
as
20
years),
the
Tx+n
value
would
have
been
the
value
appropriate
to
that
period
(that
is,
the
Tx+n
values
for
persons
aged
85+
).
.....
~

NMRt
2
Flgme 4.16: Migration Rates
lnmigration rate p
x k
Outmigration rate =
0
p
x k
Net migration rate
Where: I inmigrants
0 • outmigrants
P = popu la t ion
k constant
Figure 4.17: Net Migration Rate (NMR)
pt - ptl ) - ( Bt
t2
-
2 1 -
1/2 ( pt +
2
pt )
1
Dt
Where: pt population at an earlier
1 of time ( t 1)
1
t
- 2
period
pt population at a later period of
2 time (t2)
Bt -t = births between t 1 and t 2
1 2
Dt -t = deaths between t 1 and t 2
1 2
139

140
Figure 4.18: Residual Migration
Where: Mx + residual migration
x = an age or age group
p
x
the interval in years
between x and x + t
population aged x at
the first time period
the population at the
next time period at
age x + t
s = the survival rate
between x and x +
Figure 4.19: Population Density
Popu Iat ion
Density
Example:
Population
Density in
the U.S.in
1990
Total Population
Land Area (square miles)
248,709,873
3,539,289 70.3 persons
per square mile

141
births being ignored for all but the beginning cohort), but requires
death data by the age of the deceased and relatively detailed proce-
dures for adjusting ages and yearly death data. Figure 4.18 shows
the most widely used formulation for computing residual migration
for age groups (cohorts). In this formula, the impacts of deaths are
computed using a survival rate (usually computed from a life table)
so that migration is computed as the residual difference between the persons
actually counted at agiven date and the expected population of such persons
obtained by surviving the group of persons from an earlier date to the date
of the count. For example, to compute residual net migration for
persons who are 20-24 years of age in 1990, persons 10-14 years of
age in 1980 can be survived to 1990 when they are 20-24. The
difference between this survived population (referred to as the
expected population) and the count of persons 20-24 years of age in
1990, is the estimate of residual net migration. To obtain residual
migration estimates for the beginning ages of life, births from the
beginning period to the estimate date (usually the date of a popula-
tion count) are survived using the same formulation as shown in
Figure 4.18, but with births substituted for the Px component.
However it is computed, it is critical to remember that a residual
net migration measure uses a residual not a direct measure as the
estimate of migration. It requires one to assume that all of the
residual is due to migration. In fact, the residual difference may
also include such nonmigration factors as differences in the coverage
in the counts of the population in the two successive periods, errors
of various types in reporting or analyzing data, etc. If the existence
of such additional factors are known, they should be eliminated
before the rate is computed. Since other factors affecting the residu-
al are usually not known and cannot therefore be eliminated, it is
essential to recognize the limitations of residual migration measures.
Measures of Population Distribution
Measures of population distribution attempt to identify how a
population is distributed relative to the physical space or land area
its members inhabit. Among the most often used measures of
population distribution are simply the percentages or proportions of
persons in different types of areas, such as metropolitan or nonmet-
ropolitan, rural or urban, cities and towns versus open country, and
in places of different population six.es. Another widely used meas-
ure is the average number of persons per unit of land (usually square
miles) referred to as population density. Figure 4.19 shows

142
the standard measure of density with an example for the United
States in 1990.
· Several other measures of population distribution (that are also
used to measure other demographic factors) are shown in Figures
4.20 through 4.22. Figure 4.20 describes the population potential
measure which provides a relative measure of population distribu-
tion relative to two or more specific geographic sites. It indicates
the number of persons for whom each of several alternative geo-
graphic locations is the most accessible. In the example in Figure
4.20, the population potential of areas 1 and 2 are compared. These
areas could be census tracts, blocks, counties, or any other geo-
graphic unit, and any number of units could be compared. In this
example, the distances shown are the distances between the area of
interest (i.e., area 1 in the top panel and area 2 in the bottom panel)
and each of the other areas. Usually the distance is measured from
the center of one area to the center of the other. The distances
shown for the reference areas for which the population potentials
are being measured (i.e., the 3 miles shown for area 1 in the top
panel and the 2 miles shown for area 2 in the bottom panel) is the
average distance a person in the reference area would have to travel
to reach the reference point in the reference area (usually this is the
center of the area). In this example, area 2 has a larger population
potential than area 1 and is thus accessible to a larger number of
persons than area 1.
The population potential and related measures are often used for
site selection. If one is considering several alternative sites for a
commercial or public-service facility, one can use this measure (ad-
justing for physical features, transportation, and other factors) to
determine which of several sites is the most accessible to the largest
number of persons. The basic formula is used with its components
restricted to the items of interest. For example, population may be
replaced by households or by households or persons with given
purchasing capabilities, incomes, or other characteristics. Distance
may be replaced by travel time to the site or other relevant factors.
This measure is easily computed using procedures incorporated in
many standard geographic information systems and in many other
widely available software packages.
Figures 4.21 and 4.22 show several other widely used measures
of population distribution. The top panel of Figure 4.21 provides a
table containing a set of data for a hypothetical area. The first three
columns of this table show five population size categories of areas
(column 1), the total population accounted for by all areas in each
category (column 2), and the number of individual areas in each

Figure 4.20: Population Potential Measure with
an Example of its Application for
a Hypothetical set of Areas
n
Population Potential at L0 (Location) = I:
P.I
i=l D.I
Example:
Area 1
Area Population (P) Distance (D) P/D
1 50,000 3 miles 16,667
2 60,000 8 mi 1es 7,500
3 10,000 2 mi Ies 5,000
Total Population Potential for Area 1 29,167
Area 2
Area Population (P) Distance (D) P/D
1 50,000 8 miles 6,250
2 60,000 2 miles 30,000
3 10,000 5 miles 2,000
Total Population Potential for Area 2 = 38,250
143

144
Figure 4.21: Distribution of a Hypothetlcal Population by Size
of Place Category and the Related Lorenz Curve
Population
By Size Number Cumulative
of Place Total of Pe[cenl fe ts:eo 1
Category Pop. Areas (xi) (y i) (Xi) (Yi)
50,000 + 80,000 1 40 10 40 10
20,000-49,999 90,000 2 45 20 85 30
10,000-19,999 10, 000 1 5 10 90 40
5,000- 9,999 10,000 2 5 20 95 60
5,000 10,000 4 5 40 100 100
Lorenz Curve
100
90
80
70
60
Proportion
of 50
Places
40
30
20
10
0
10 20 30 40 50 60 70 80 90 101
Proportion of Population

Figure
4.22:
The
Glni
Coefficient
and
Index
of
Dissimilarity
Measures
of
Population
Distribution
Gini
Coefficient
(GI)
n
n
Gi
=
:E
X.
Y.+l
-
:E
x.
+1
y,
i
=1
I
I
i=l
I
I
Where:
X.
and
Y.
are
cumulative
percentage
dlstribulions
for
two
factors
Index
of
Dissimilarity
(ID)
ID
k
~
:E
i=l
lxi
Yi
I
Where:
x.
and
y.
are
percentage
distributions
f~r
two
!actors.
(rontinues)
......
i

Figure
4.22
(continued)
Example:
To
compute
Glni
Coefficient
and
Index
of
Dissimilarity
given
the
following
data:
Percentage
distributions,
cumulative
percentage
distributions,
and
cross
products
of
cumulative
percentage
distributions
for
a
hypothetical
population
(See
Figure
4.21)
Cumulative
Percentage
Percentage
Absolute
Di
s
tr
i
but
ion
Dist
r
i
but
ion
Proportional
Percent
Size
Cross
Products
Difference
of
Place
Places
Population
Places
Population
Category
(y
i)
(xi)
(Yi)
(Xi)
XiYi+l
xi+lyi
lyi-xil
50,000
+
10
40
10
40
0
.12
0.09
30
20,000-49,999
20
45
30
85
0.34
0.27
25
10,000-19,999
10
5
40
90
0.54
0.38
5
5,000-9,999
20
5
60
95
0.95
0.60
15
5,000
40
5
100
100
-
-
35
Sum
(~)
=
1.
95
1.
34
110
Gini
(Gi)
=
1.95
-
1.34
=
0.61
Index
of
Dissimilarity
(ID)
=
*
~
=
110/2
=
55
~

147
population size category (column 3). Thus, the first row of data for
these items shows that there was 1 area in the size category of
50,000 or more persons that had 80,000 persons, row 2 shows that
there were 2 areas in the size category with 20,000 to 49,999 persons
which together had 90,000 persons, etc. Columns 4 and 5 show the
simple percentage distributions of population and areas. That is, the
first row of data indicates that areas of 50,000 or more persons
accounted for 40 percent of the total population (of 200,000) in the
areas (column 4) and for 10 percent of the areas (of 10 areas) includ-
ed in the table (column 5). Columns 6 and 7 show the cumulative
percentage distributions, cumulating from the largest to the smallest
size of place categories. Ten percent of all areas were in the
50,000+ category, another 20 percent, or a cumulative percentage of
30 percent, were in places in the 50,000+ plus the 20,000-49,999
category (column 6). Forty percent of the population was in places
of 50,000+, another 45 percent in places of 20,000 to 49,999 for a
total cumulative percentage of 85 percent of the population in areas
of 50,000+ plus areas of 20,000 to 49,999 (column 7).
The Lorenz Curve shown in the bottom panel of Figure 4.21 shows a
graphical representation of the two cumulative percentage distributions
(columns 6 and 7) relative to one another. This curve was constructed
by connecting points which indicate the proportion of population
relative to the proportion of areas with a line drawn through the
points and connecting the two ends of the diagonal line. In this
example, 10 percent of the places accounted for 40 percent of the
population, 30 percent of the places for 85 percent of the population,
etc. The diagonal line is provided as a base for comparison because
it represents the condition in which the two cumulative percentage
distributions would be identical (e.g., 10% of the areas would have
10% of the population, 20% of the areas would account for 20% of
the population, etc.).
The distance between the diagonal line and the curve construct-
ed from the cumulative percentage distributions of the two factors
shows how similar the percentage distributions of the two factors
are given the size categories shown. The greater the distance
between the diagonal line and the curve, the greater the difference
in the distribution of the two factors. This curve may be drawn
either above or below the diagonal depending on whether the
cumulative distributions are cumulated from the highest to the
lowest category or from the lowest to the highest. For example, the
curve in Figure 4.21 would have shown the same area between the
diagonal and the curve had the percentage distribution shown in the
table been cumulated from the smallest to the largest population

148
size category (i.e., from the bottom up rather than from the top
down), but the curve would have been above instead of below the
diagonal (so computed, 40% of the areas or places would have
accounted for 5% of the population, and 60% of the places would
have had 10% of the population, etc.).
The Lorenz Curve is widely used because it presents an easy-to-
construct graph of the relationship between any two cumulative
percentage distributions. For example, it is often used in economic
analyses to indicate the distribution of income relative to the popula-
tion or the number of households and, when used as such, can be
seen as a graphical measure of income inequality.
An examination of the Lorenz Curve reveals that the area
between the diagonal and the curve indicates the extent of maldis-
tribution between the two factors graphed on the two axis. The
relevant measure is the proportion of the area between the diagonal and
the curve of the total area under (or over) the diagonal. The measure of
this area is called the Gini Coefficient. The formula for this coeffi-
cient is shown in Figure 4.22 along with an example of its use with
the data in Figure 4.21. This coefficient is simply the difference in
the cross products of the cumulative percentage distribution. In the
example shown, the Gini Coefficient indicates that roughly 60
percent of the area under the diagonal is between the diagonal and
the Lorenz Curve, indicating the population in places tends to be
concentrated relative to the size of place categories.
Another useful measure of distribution is the Index of Dissimilari-
ty which indicates the similarity of two categorical percentage distri-
butions (not cumulative, but simple percentage distributions). This
measure, which is simply one-half the sum of the absolute differences
between the percentage values in the categories of the two distributions, is
interpreted as indicating the proportion of population that would
have to change categories for the two distributions to be identical.
In the example in Figure 4.22, the Index of Dissimilarity is 55 indi-
cating that 55 percent of the population would have to change
categories for the two distributions to be identical.
Both the Gini Coefficient and the Index of Dissimilarity have
been extensively used to assess inequalities in distributions. The
latter measure, in fact, is often referred to as the segregation index
(see discussion below) because it is employed to measure the segre-
gation of racial/ethnic groups in cities and other areas. These and
related measures are among the most useful for assessing how two
factors are distributed relative to one another and are some of the
only simple summative measures available for measuring the differ-
ences between percentage distribution of two categorical variables

149
(see Massey and Denton, 1988 for a discussion of other segregation
measures).
It is important to recognize the wide applicability of these
measures. They can be used in at least three ways: (1) to assess the
difference between two factors for several different areas (such as
the distribution of customers and income among market areas or the
distribution of service centers relative to the number of clients for a
public service); (2) to compare two different areas relative to their
distribution across categories of a single variable (e.g., to compare
the income distributions for two different market areas); or (3) to
examine changes in the distribution of a variable over time (e.g., the
proportion of a product's users in different income categories in two
different years). These measures are ones that are not only usefully
applied to examine the geographic distribution of population relative
to land area, but can also be used to examine the distribution of
other factors likely to be of interest to the applied analyst.
Measures of Population Composition
Many of the general measures described at the beginning of this
chapter are also employed to measure the characteristics of a popula-
tion. For example, median age is a common measure of the age
structure of a population as are simple percentage distributions
showing the number and percentage of persons in each age group.
Similarly, median income and median years of education are widely
used measures. In the following discussion of measures of popula-
tion composition, only the relatively unique measures of each varia-
ble are delineated. Readers should be aware, however, that many
of the general measures can also usefully be applied to describe the
characteristics of a population.
Age and Sex Composition. The age and sex composition of a
population affect many other characteristics of a population from its
rates of fertility to the nature of the goods and services it is likely to
demand. Age is often measured by the use of simple percentage
distributions and the mean or median years of age. Sex is similarly
a key variable which is often measured in terms of the percentage of
the population that is male or female. Figures 4.23 through 4.25
provide other basic measures of these two variables.
Figure 4.23 shows the dependency ratio. This ratio indicates the
number of persons in dependent ages relative to the number in the working
ages. The dependent ages are variously defined as those 0-14 or 0-19

150
Figure 4.23: Dependency Ratio (DR)
DR =
Where: P0 _14 number of persons
0-14 years of age
Example:
Texas DR
(1990)
SR
number of persons
65 years ol age
and older
number of persons
15-64 years of age
4,080,580 + 1,716,576
~--------- x 100
11,189,354
Figure 4.24: The Sex Ratio (SR)
PM
-X 100
PF
Where: PM number of males
PF
Example:
SR Texas 1990
number of females
8,365,963
- - - - X 100
8,620,547
97.1
51. 8

Figure
4.25:
Population
Pyramid,
Texas
1990
Population
Pyramid,
Texas
(Population
shown
in
thousands)
Age
Group
85+
80-84
75-79
70-74
65-69
60-64
55-59
50-54
45-49
40-44
35-39
30-34
25-29
20-24
15-19
10-14
5-9
0-4
i
I
I
I
I
I
i
I
I
'
I
I
I
~
'
I
I
I
I
!
I
I
I
I
'
.
'
I
I
I
I
I
I
1000
800
600
400
200
0
200
400
600
800
1000
-
Male
•
Female
.......
01
.......

152
and those 65 years of age or older with those in the working ages
being all those at ages between the young and old dependent ages.
The ratio is sometimes computed separately for the young, in which
case it is referred to as the youth dependency ratio, or for the old,
referred to as the old-age dependency ratio. The dependency ratio
indicates how a population's age structure is likely to affect its abili-
ty to support itself and is therefore used both as a measure of age
and as a measure of the economic characteristics of a population.
Figure 4.24 presents perhaps the most widely used measure of
the sex composition of the population. This is the sex ratio, the
number of males divided 11y the number of females and multiplied 11y 100.
The sex ratio is extensively used in many forms of analyses because
of its consistency. In most developed countries, the sex ratio at
birth is approximately 105 males per 100 females, decreasing to
about 100 by age 20 to 30 and to about 60 by age 80. Wide variation
from these expected levels can be used to identify areas where
unique demographic events have occurred. For example, Bean et
al., (1982; 1983) used sex ratio differences between Mexican states
bordering the United States and those within the interior of Mexico
to estimate the number of illegal immigrants from Mexico in the
United States. In addition, the sex ratio has come to be increasingly
used as a factor which is indicative of conditions likely to lead to
particular patterns of behavior and family change (see for example,
Fossett and Kiecolt, 1990; Messner and Sampson, 1991).
One often employed technique to indicate the joint distribution
of age and sex in a population is the age-sex pyramid. Age-sex
pyramids are constructed simply 11y taking the number of males and females
of each age and graphing their numbers as shown in Figure 4.25 or by
using percentages in which the number of each sex in each age
group is divided by the total population and the percentages shown
graphically. By tradition, females are placed on the right and males
on the left side of the pyramid. In general, it is the width of the
base (beginning years) of the pyramid relative to its width at other
ages that is of interest in analyzing such pyramids. Pyramids with
larger bases reveal populations that are generally younger popula-
tions, while those with age categories that are more uniform in
width are likely to be indicative of an older population.
Race/Ethnicity. There are relatively few unique measures for
assessing the race/ethnicity composition of a population. Rather,
these characteristics are usually described in terms of simple numeri-
cal and percentage comparisons of the numbers and proportions of
persons in each race/ethnicity group in a population. However, two

153
measures that are used to measure the similarity in the patterns of
distribution of racial/ethnic groups across geographical areas are the
Index of Dissimilarity, or segregation index, and the Gini Coefficient
described under the discussion of measures of population distribu-
tion. If the proportions of persons in two different racial/ethnic
groups are compared for a set of areas, then the Index of Dissimilar-
ity and Gini Coefficient measures can be computed in the manner
shown in Figures 4.21 and 4.22. These can be interpreted as indicat-
ing the extent to which two racial/ethnic groups are physically
segregated from one another. A review of such measures (Massey
and Denton, 1988; 1989) shows that they are widely applicable
across areas and point to high levels of segregation among racial and
ethnic groups throughout the United States.
Household, Family, and Marital Composition. The measures of
household, family, and marital composition most used are simply
the number and percent of persons in specific categories of house-
holds and marital statuses. Other frequently used measures are
aoerage household siu (the number of persons living in households divided
11y the number of households), median or average age at first marriage,
and nuptiality (life) tables showing the numbers and proportions
married and single in populations with different levels of age-specif-
ic marriage and mortality rates. Figure 4.26 shows the formulations
for the crude, general, and age-specific marriage rates. Household,
family, and marital characteristics are measured by the use of quite
general measures.
Educational Characteristics. Shryock and Siegel (f980) note that
measures of educational characteristics can be grouped into those
that measure the inputs into the educational system, those that
measure progression in the system, and those that measure outputs
from the system. The measures of educational input most often used
are simply the crude, general, and age-specific rates of enrollment
with ages 5-34 used in the general rate. Similarly, the measures of
educational output most often employed are simply the crude and
age-specific illiteracy rates (with illiteracy variously defined either by
measured skills or less than 3 or less than 5 or some other designat-
ed number of years of formal education) and the attainment rate,
the proportion of the population achieving a given level of educa-
tion. Figure 4.27 shows two measures of educational progression.
Such retention and graduation rates can be usefully applied to
measure the progress of persons through an educational system.
These measure when combined with other descriptive measures

154
Figure 4.26: Crude, General, and Age-Specific
Marriage Rates
Crude Marriage 'Rate (CMR)
CMR ~ x 1,000
p
Where: CMR Crude Marriage Rate
M Number of marriages
during time period
(usually one year) of
interest
P Total population
General Marriage Rate (GMR)
GMR
M
x 1,000
p15+
Where: GMR General Marriage Rate
M Number of marriages
during time period of
interest
Population 15 years of
age and older
Age-Specific Marriage Rate (ASMR)
ASMR
Ma
p x 1,000
a
Where: ASMR
M
a
p
a
Age-Specific Marriage
Rate
Marriages to people of
age a during time
period of interest
Number of people of
age a

Figure 4.27: Measures of Educational
Progression
Grade Retention Rate (GRR)
GRR =--- x 100
Where: GRR Grade Retention Rate
Entrants to, or enroll-
ments in, grade g in
year y
Ey + x E 11
= ntrants to, or enro -
g + x ments in, grade g + x in
year y + x
Grade Graduation Rate (GGR)
GGR =
x + 1
x
x 100
Where: GGR = Grade Graduation Rate
EY =Entrants to, or enroll-
g ment in, grade g in
year y
Hy + x + 1 = Number completing
g + x school grade g + x in
year y + x + 1
155

156
of education, such as the median years of school completed, can
provide a comprehensive overview of the educational characteristics
of a population.
Economic Characteristics. The most commonly used means of
describing the economic characteristics of the population are such
descriptive measures as median income, per capita income (which is
simply the mean income per person in a population), the percent of
the labor force employed and unemployed, and the percent em-
ployed by occupational and industrial categories. Among the other
measures commonly used to describe the labor force are the labor
force participation rates shown in Figure 4.28. Although labor force
participation rates are widely known and used, the fact that these
rates are simply crude, general, and specific rates of labor force
participation is seldom recognized. In fact, the general labor force
participation rate is commonly referred to simply as the labor force
participation rate. Such rates, together with basic descriptive
measures, can provide a relatively complete description of the basic
economic characteristics of a population.
Selected Methods for Controlling
the Effects of Demographic Change
and Characteristics
Measures and methods that provide means of describing the
extent and form of demographic processes and characteristics in a
population have been presented. In the final section of this chapter,
procedures are examined that attempt to determine how much dif-
ference demographic factors make in the determination of a pattern
of events or behaviors. Viewed alternatively, analysts are some-
times interested in knowing how similar patterns of events or behav-
iors would be in two different populations if they had the same age
structure, ethnic composition, etc. For example, the sales for a
given product may be less in one area than in another, but the
populations of the areas may have very different age structures. Is
the difference in sales due to age structure differences or to other
factors? Service centers for a public service may have been estab-
lished on the basis of similar total populations, but the case load in
one center may be much higher than in another. Is the difference in
case loads due to differences in the ethnic, household, and income
compositions of the populations of the areas or due to other factors,
such as differences in staff interpretations of regulations? This
process of separating the effects of one set or type of factor from

Figure 4.28: Measures of F.conomlc Activity
Crude Labor Force Participation Rate (CLFPR)
LF
CLFPR - - X 100
p
Where: CLFPR Crude Labor Force
Participation Rate
LF = Labor force
P = Total population
General Labor Force Participation Rate (GLFPR)
GLFPR
LF
--- x 100
p15-64
Where: GLFPR = General Labor Force
Participation Rate
LF = Labor force
p15-64 =Population in economically
active population 15-64
(or 20-64) years of age
Age-Specific Labor Force Participation Rate (ASLFPR)
LFa
ASLFPR =- X 100
p
a
Where: ASLFPR
LFa
pa
Age-Specific Labor Force
Participation Rate
Labor force
age group a
.. Population in age group a
157

158
another can be seen as a process of controlling for the effects of such
factors.
Controls can be completed using several alternative procedures
ranging from very complex statistical procedures to the use of simple
rates and ratios. In this section, an introduction is provided to a few
general procedures likely to be of utility to the applied analyst.
Procedures for determining the statistical effects of demographic
variables and involving relatively complex multivariate modeling
(e.g., multiple regression, log-linear, path analysis, and hazard
models) techniques are not examined because such procedures are
too complex to be presented in the space available, and because they
are extensively descri}?ed elsewhere (Kerlinger and Pedhazur, 1973;
Snedecor and Cochran, 1967). At the same time, since rates and
ratios were examined earlier in this chapter, they will not be revisit-
ed here. Rather, we focus on widely used techniques that are rela-
tively simple to apply.
The specific procedures to be examined are:
(1) direct and indirect standardization;
(2) rate decomposition; and
(3) multiple-decrement life tables.
For each of these techniques, the basic approach and uses within
demography, computational procedures, and examples of applied
uses of the technique are presented. Readers should be aware that
more complete descriptions of these techniques are available from
other sources, and these sources should be consulted for more
complex applications of these techniques (see for example Shryock
and Siegel, 1980; Namboodiri and Suchindran, 1987; Das Gupta,
1978; 1990; Land and Rogers, 1982; Pollard et al., 1990).
Direct and Indirect Standardization
Standardization is among the most widely used methods to
control the effects of demographic variables. This technique involves
comparing two or more populations to determine whether or not differences
among them in the occurrence of an event or phenomenon are due to differ-
ences in population characteristics. The basic logic behind this tech-
nique is that if two or more populations being compared can be
standardized relative to the factor or factors believed to be leading to
the difference, then the effects of such differences can be deter-
mined. If the differences disappear when the factor is standardized,
then it can be concluded that it was the populations' differences

159
relative to the standardized factor that led to the differences in the
number of occurrences.
Differences in the occurrence of factors between populations can
result from two general sets of factors which provide alternative
procedures for standardizing demographic data. Such differences
may be due to the fact that the rate of occurrence of the phenome-
non is different in the two populations or because the compositions of
the populations are different. The same number of occurrences can
be obtained by either a high rate of occurrence in a small population
or by a low rate of occurrence in a large population. Two alterna-
tive forms of standardization are based on these two forms of differ-
ences.
Direct methods standardize two or more populations by comparing the
numbers of occurrences that one obtains in each population by applying the
specific rates for each population to the composition of a standard popula-
tion. This standard population can be any population but usually
that for a larger area of which the areas to be compared are a part,
or areas that are similar to the areas being compared, are used. For
example, a state may be used as the standard population to compare
counties and a county as the standard to compare cities within it.
Indirect standardization applies a set of specific rates from a standard
population to each of the population compositions of the areas to be
compared. As with the population used as the standard in·direct
standardization, the rates used as the standard rates are generally
obtained from a population that is either a parent area for the areas
to be compared or is similar to the areas being compared.
Figure 4.29 presents an example of the use of both direct and
indirect standardization to examine differences in home sales among
two sales territories which had populations with very different age
structures. In the first part of this example, age is standardized
using the method of direct standardization. Using this form of
standardization, rates for each of the areas to be compared are
applied to the population of a standard area, in this case the city in
which the sales areas are located. In the latter half of Figure 4.29,
indirect standardization is demonstrated with the rates for the city
being used as a standard which is applied to each area's population
by age. The analysis in Figure 4.29 shows that differences in the
age structures of the two areas are largely responsible for the differ-
ences in the sales observed between the two areas.
Unlike the example in Figure 4.29, in many potential uses of
standardization, data are not available on both rates of occurrence
for the factors being compared or on the detailed age structure of

Figure
4.29:
Direct
and
Indirect
Age
Standardization
Purpose:
To
determme
whether
apparent
differences
in
the
Incidence
of
an
occurrence
of
a
phenomenon
in
two
or
more
populations
are
due
to
differences
in
the
age
structures
of
the
populations
of
the
areas
or
to
other
factors.
Example:
To
determme
If
the
sales
of
single-family
homes
In
two
different
areas
of
a
hypothetical
dty
are
due
to
age
structure
differences
in
the
populations
of
the
two
areas
or
to
other
differences.
Given:
Two
areas,
1
and
2,
of
a
hypothetical
dty
with
popuJations
of
36,800
(area
1)
and
29,000
(area
2)
had
sales
of
3,570
and
1,932
respectively
in
January
through
May
of
1991.
You
wish
to
evaluate
whether
the
difference
in
sales
Is
because
the
population
in
area
2
is
concentrated
in
age
groups
less
likely
to
purchase
homes
or
whether
such
factors
as
your
advertising,
the
skills
of
sales
personnel,
etc.,
have
created
the
differences
in
sales.
Use
Direct
Standardization
in
which
age-specific
rates
for
the
areas
to
be
compared
are
multiplied
by
the
age
structure
of
a
standard
population.
Given
annual
age-specific
rates
of
single-family
home
purchasers
in
each
area
and
the
age
structure
for
the
dty
as
a
whole
used
as
the
standard
(note
the
standard
can
be
any
population
of
interest),
the
results
are:
(continues)
.....
~

Figure
4.29
(continued)
Annual
Standard
Age-Specific
Population
of
Purchase
Rates
for
the
City
as
a
Single-Familx
Homes
Whole
bv
Ae.e
Age
Area
1
Area
2
--
20-34
.30
.29
27,100
35-54
.19
.23
20,900
55-64
.07
.05
8,200
65
+
.04
.06
9,600
Total
expected
number
of
sales:
Difference
In
sales
between
two
areas
-
13,652
-
13,059
-
593
'
Expected
Annual
Sales
Area
1
8,130
3,971
574
384
13,059
Area
2
7,859
4,807
410
576
13,652
Use
Indirect
Standardization
In
which
a
standard
set
of
age-specific
rates
are
applied
to
the
age
structures
for
the
areas
being
compared.
Given
a
set
of
standard
annual
age-speclfic
purchase
rates
and
age
structures
for
each
of
the
two
comparison
areas
as
follows:
(continues)
I--'
°'
I--'

Figure
4.29
(continued)
Standard
Age-Specific
Purchase
Rates
for
Population
by
Age
Single-Family
Homes
Age
Area
1
Area
2
20-34
.
29
18,000
9,100
35-54
.21
14,
800
6,100
55-64
.06
2,000
6,200
65
+
.05
2,000
7,600
Total
expected
number
of
sales:
Difference
In
sales
between
two
areas
-
8,548
-
4,672
.,
3,876
Expected
Annual
Sales
Area
1
Area
2
5,220
2,639
3,108
1,281
120
372
100
380
-
-
--
8,548
4,672
Conclusion:
The
differences
between
the
sales
In
the
two
areas
are
primarily
because
of
the
concentration
of
the
age
structure
of
the
population
In
Area
1
In
younger
adult
ages
with
high
rates
of
home
purchasing
and
the
concentration
of
the
population
In
Area
2
In
older
age
groups
with
lower
rates
of
purchasing.
This
is
shown
by
the
fact
that
In
direct
standardization
the
expected
values
obtained
In
the
standardization
for
the
two
areas
are
not
nearly
as
different
as
the
actual
sales.
In
like
manner,
Indirect
standardization
clearly
shows
age
structure
effects
resulting
In
proportional
patterns
similar
to
the
differences
actually
occurring
(when
one
adjusts
for
the
fact
that
the
standardlied
sales
are
for
a
year
but
the
actual
period
observed
was
five
months
[by
dividing
8,548
and
4,672
by
5/12)).
r.
~

163
the populations being compared. As a result, both forms of stand-
ardization are seldom used simultaneously. Rather, since detailed
rates specific to given characteristics are less often available than the
age structure of populations, indirect standardization is most likely
to be used when data on rates specific to the demographic character-
istic to be standardized are not available for the populations being
compared. When such rates are available, direct standardization is
the technique most likely to be employed.
Standardization is an extremely useful procedure for addressing
the question of whether or not differences between areas are due to
specific characteristics. It can be used to control the effects of a
single characteristic or to control several characteristics simultaneous-
ly. To control multiple variables, the only change in procedure
necessary from that noted above is the need to obtain rates for the
areas to be compared that are specific to the combination of demo-
graphic variables to be controlled (e.g., age, sex, race/ethnicity, and
income specific rates if all these factors are to be controlled simulta-
neously and direct standardization is to be used) or the need to
obtain the number of persons (the populations) having each com-
bination of the characteristics to be controlled (e.g., the number of
20-24 year old Hispanic males earning $50,000 or more per year, if
age, sex, ethnicity, and income are the variables to be controlled and
indirect standardization is to be employed). Differences in a wide
array of demographic factors, such as the number of births, deaths,
or migrants, can be examined using standardization, but so can such
differences as those in sales, public service usage, the incidence of
diseases, and other factors. Since one often wants to eliminate the
effects of demographic variables in the search for other determi-
nants, standardization is a very useful technique.
Rate Decomposition
One of the difficulties related to the standardization of rates is the
fact that, although standardization allows one to discern whether or
not the standardized factors affect the differences between crude
rates of occurrence in two or more areas, it does not allow one to
identify the extent to which this difference is a function of the two
factors that potentially account for the difference-the differences in
population composition and the differences in the specific rates in
the populations. Rate decomposition is a technique which allows
such differences and the magnitude of such differences to be identi-
fied. This technique was introduced by Kitagawa (1955) and has
been extensively developed (Das Gupta, 1978; 1990; Clogg and

164
Eliason, 1988; Liao, 1989) for such uses as identifying the effects
of the distribution of age and marital status on the growth of house-
holds (Sweet, 1984), the effects of occupational structure and segre-
gation on the index of occupational dissimilarity (Bianchi and
Rytina, 1986), the effects of several demographic characteristics on
adolescent fertility (Nathanson and Kim, 1989), the effects of select-
ed demographic characteristics on the differential returns to labor
among blacks and whites (Lichter and Constanzo, 1987; Lichter,
1989), and the effects of age and education on outmigration rates
(Wilson, 1988).
Rate decomposition, at least as originally developed, involves
decomposing the difference between two crude rates of occurrence,
by using one or another form of weighted average of the composi-
tions and the specific rates of the populations being compared, to
analyze the sources of the difference. The proofs and computations
underlying this procedure are shown in the sources noted above,
but the general results for a sample rate decomposition are present-
ed here to more fully describe this procedure. Computer programs
for implementing rate decomposition are available from the authors
and several other sources.
The examples shown in Tables 4.1 and 4.2 involve decomposi-
tion of differences in crude rates of participation in outdoor recrea-
tional and tourism activities for the United States and Texas as
reflected in projections of the proportion of participants in the
populations involved in each activity (at least once a year) in 1990
and 2000 and 2000 and 2025 (see Murdock et al., 1990; 1991a;
1991b). The comparison is of crude rates for a given activity at two
different points in time. These comparisons are made for the popu-
lations of each of two areas, the United States and the State of Texas.
In each table, data for the total effect, the rate effect, the age
effect, and the race effect are shown in columns 1-4. The total effect
is equal to the difference between the crude rates for a population
(the United States or Texas) at the two different points in time being
compared (1990 and 2000 or 2000 and 2025). The rate effect is the
difference between the standardized rates for the two populations
with the effects of the standardized variables (in this case age and
race) having been removed. The effects for age and race are the
effects of these respective variables.
Several aspects about the results shown in Tables 4.1 and 4.2
require clarification in order for them to be adequately understood.
First, it should be noted that the rate effect and effects for the char-
acteristics will sum to the total effect. For example, an analysis of
the data on birdwatching for the 2000 to 2025 period in the first four

Table
4.1:
A
Decomposition
of
the
Projected
Difference
In
the
Rate
of
Partl.dpation
In
Different
Recreational
Activities
Among
Residents
of
the
United
States
by
Activity,
1990-2000
and
2000-2025
Percent
of
Olange
Percent
of
Absolute
Olange
Composition
Effect
Due
To
In
Total
Effect
Due
To
In
Total
Effect
Due
To
Reaeatl.onal
Total
Rate
Rate
Rate
Activlties
Effect
Effect
Age
Race
Effect
Age
Race
Effect
Age
Race
1990-2000
Backpadclng
-1.1795
0.0136
-1.0908
-0.1023
-1.15
92.48
8.67
1.13
90.39
8.48
Birdwatching
0.3623
-0.0170
0.3839
-0.0046
-4.69
105.96
-1.27
4.18
94.68
1.14
Camping
-1.4346
0.0249
-1.3899
-0.0696
·1.73
.
96.88
4.85
1.68
93.63
4.69
Fishing
-0.5946
0.0871
-0.6103
-0.0714
-14.65
102.65
12.00
11.3.1
79.39
9.28
DayHildng
-1.0468
0.0240
-0.9383
-0.1325
-2.29
89.63
12.66
2.19
85.71
12.10
Hunting
-0.3915
0.0832
-0.4149
-0.0098
-21.26
105.99
15.27
14.92
74.37
10.71
Plcniddng.
-0.4447
-0.0006
-0.4482
0.0091
1.26
100.78
-2.04
1.21
96.83
1.96
Wallclng
-0.1656
-0.0230
-0.1480
0.0054
13.89
89.36
-3.25
13.03
83.92
3.00
2000-2025
Backpadclng
-2.1632
0.0490
-2.0098
-0.1524
-2.26
95.22
7.04
2.17
91.10
6.74
Birdwatching
0.9590
0.1002
0.8606
-0.0018
10.44
89.75
-0.19
10.40
89.41
0.19
Camping
-2.8000
0.0928
-2.8318
-0.0610
-3.32
101.14
2.18
3.11
94.85
2.04
Fishing
-1.1418
0.3.%5
-1.3934
-0.1029
-31.05
122.04
9.01
19.16
75.28
5.56
DayHlldng
-1.7946
0.1580
-1.7171
-0.2355
-8.80
95.68
13.12
7.48
81.36
11.16
Hunting
-0.9988
0.2033
-1.0899
-0.1122
-20.35
109.11
11.24
14.46
77.55
7.99
Plcnf.cktng
-2.2732
-0.6185
-1.6220
-0.0327
27.21
71.35
1.44
27.21
71.35
1.44
Walking
-0.0642
0.2651
-0.3584
0.0291
-412.93
558.26
-45.3.1
40.63
54.91
4.46
,_.
°'
01

Table
4.2:
A
Decomposition
of
the
Projected
Difference
in
the
Rates
of
Participation
in
Different
Recreational
Activities
Among
Residents
of
Texas
by
Activity,
1990-2000
and
2000-2025
Percent
of
Oiange
in
Total
Percent
of
Absolute
Change
Composition
Effect
Due
To
Effect
Due
To
in
Total
Effect
Due
To
Recreational
Total
Rate
Race/
Rate
Race/
Rate
Race/
Activities
Effect
Effect
Age
Ethnicity
Effect
Age
Ethnicity
Effect
Age
Ethnicity
1990-2000
Bicycling
-1.7690
-0.4.?47
-1.4950
0.180'7
25.70
84.51
-10.21
21.34
70.17
8.48
Saltwater
Swimming
-1.2249
-0.9043
-0.5186
0.1980
73.82
42.34
-16.16
55.79
31.99
12.21
Golf
-0.7265
-0.4812
-0.0381
-0.2072
66.23
5.25
28.52
66.23
5.25
28.52
Horseback
Riding
-1.0294
-0.4372
-0.4998
-0.0924
42.47
48.56
8.97
42.47
48.56
8.97
Camping
-1.7128
-0.8990
-0.5432
-0.2706
52.49
31.72
15.80
52.49
31.72
15.80
Hunting
-1.0491
-0.8389
0.0646
-0.2748
79.96
-6.15
26.19
71.20
5.48
23.32
Nature
Study
-0.5688
-0.4163
-0.0722
-0.0803
73.19
12.69
14.12
73.19
12.69
14.12
Freshwater
Fishing
-1.8979
-1.1717
-0.3667
-0.3595
61.74
19.32
18.94
61.74
19.32
18.94
Saltwater
Fishing
-0.2366
-0.3917
0.0347
0.1204
165.56
-14.68
-50.88
71.63
6.35
22.01
2000-2025
Bicycllng
-4.3728
-2.8955
-2.0920
0.6147
66.22
47.84
-14.06
51.69
37.34
10.97
Saltwater
Swimming
-2.5683
-1.9317
-1.3735
0.7369
75.21
53.48
-28.69
47.79
33.98
18.23
Golf
-1.4480
-0.7653
-0.2424
-0.4403
52.85
16.74
30.41
52.85
16.74
30.41
Horseback
Riding
-2.0425
-1.0646
-0.8565
-0.1214
52.13
41.93
5.94
52.12
41.93
5.94
Camping
-3.9232
-2.3991
-1.0736
-0.4505
61.15
27.37
11.48
61.15
27.37
11.48
Hunting
-2.7451
-1.5026
-0.7073
-0.5352
54.74
25.76
19.50
54.74
25.77
19.50
Nature
Study
-1.3897
-1.0224
-0.2290
-0.1383
73.57
16.48
9.95
73.57
16.48
9.95
Freshwater
Fishing
-4.3809
-2.6930
-0.9981
-0.6898
61.47
22.78
15.75
61.47
22.78
15.74
Saltwater
Fishing
-0.9572
-0.8990
-0.4536
0.3954
93.92
47.39
-41.31
51.43
25.95
22.62

167
columns of Table 4.1 suggests that population change between 2000
and 2025 will increase the rate of participation (the proportion of
persons participating) in birdwatching by 0.95. This is composed of
positive effects for the rate and age effects (0.10 and 0.86, respective-
ly) and a negative effect (- 0.0018) for the race/ethnicity effect.
These results indicate both the type of findings likely to appear
in the use of rate decomposition and the fact that understanding
such results requires knowledge of the direction and nature of
population change and of the specific rates being examined in the
populations being analyzed. Thus, the projections shown were
completed by taking 1980-based participation rates and applying
them to population projections (from Spencer, 1986 and 1989 for the
United States and from Murdock et al., 1989a for Texas). For
example, the participation rates indicate that birdwatching is much
higher among older and majority populations and much lower in
younger age groups and among minorities, while the projections
point to a future population with an increasing number of elderly
and a population with an increasing proportion of minorities. The
aging of the population should increase the number of birdwatchers,
but the increase in the number and proportion of minorities would
decrease the number of birdwatchers. Only by knowing that the
projections used show an aging of the population and an increase in
the proportion of minorities and that the rates of participation for
birdwatching are higher for older and majority population groups is
it possible to interpret the results.
The results in the remaining columns (columns 5-10) of each
table show the percentage of the total effects due to the rate and the
characteristics' effects. The percentages in columns 5 through 7
show percentages with negative and positive values (which sum to
100 percent of the difference), while columns 8 through 10 show
absolute percent contributions in which the signs of the effects are
ignored. The values in these last three columns allow one to more
easily discern the relative impacts of the variables. These percent-
ages for the above noted example of birdwatching indicate that,
from 2000 to 2025, more than 89 percent of the change in the crude
rates of incidence is due to the effects of aging.
One remaining aspect of the results reported in a rate decompo-
sition requires clarification, the rate effect. The rate effect is the
difference between the crude effects remaining after the populations
have been standardized for the other variables in the analysis (in the
example in Tables 4.1 and 4.2 for age and race/ethnicity). It is large-
ly a residual of the effects of all other factors not standardized in the
rates and, although the factors determining a residual are clearly not

168
fully identifiable (Das Gupta, 1978; 1990), knowledge of the popula-
tions involved can also assist one in interpreting rate effects.
For example, in addition to the data shown in Tables 4.1 and
4.2, an analysis (not shown here) was completed using alternative
projection scenarios. This analysis indicated that the percent of the
effect due to the rate effect increased with the rate of in-or immigra-
tion in the populations. Since growth through migration involves a
disproportionate number of young adults, who generally have
higher rates of involvement in rigorous recreational activities, this
suggests why the rate effect is nearly always positive in the tables.
The rate effect involves increases (through migration) in young
adults who tend to increase the total rate of participation.
Although data on the same recreational activities were not avail-
able for both Texas and the United States as a whole, given the
information noted above, the data in Tables 4.1 and 4.2 point to
several findings likely to be of relevance for applied analysts. It is
evident in the data in both tables that, whether one examines data
for the United States as a whole or even for a relatively rapidly
growing state such as Texas, participation in recreational activities of
the types noted in these tables is likely to be decreased by the pro-
jected population patterns of the future because future populations
show changes in age and race/ethnicity characteristics that decrease
the incidence of recreational participation (i.e., older age and more
persons of minority status).
It is also evident, that both rate effects and race/ethnicity effects
are substantially greater in Texas than in the United States as a
whole. This is a result of the fact that the Texas population is pro-
jected to grow nearly three times as fast as the Nation as a whole (at
about 1.5% per year for Texas versus 0.5% for the Nation over the
projection period, 1990-2025) and to involve a much larger propor-
tion of minorities (e.g., by 2025, 50% of the Texas population would
be composed of minorities versus about 33% of the U.S.
population). These results suggest that Texas will continue to have
a market for recreational activities more similar to the historic pat-
terns showing a greater prevalence of young adult activities, while
the rest of the Nation will show a much faster evolution toward
middle- and older-age activities.
The results also show that different activities are likely to be
differentially affected by the projected patterns of population
change. For example, the rates of participation in walking, and to
some extent, picnicking in the United States are not substantially
higher for majority populations than for minority population as are
the rates for other activities. They are therefore less impacted by

169
changes in racial and ethnic patterns. In Texas, bicycling and salt-
water fishing show higher rates of minority participation than other
activities. As a result, their incidence in the population is increased
by the changes in the race/ethnicity composition of the population.
The interpretations delineated above demonstrate how rate
decomposition can be used to discern the impacts of likely future
patterns and the implications of different types of population change
for different activities and behaviors. Its use and the correct
interpretation of its results, however, require knowledge of the
underlying population patterns and of the differentials in rates for
the factors being examined.
Multip~e-Deaement Life Tables
Simple life tables were introduced earlier in this chapter. Their utili-
ty was indicated as allowing one to discern the long-term effects of
incremental loss (through death or an equivalent process such as
housing demolitions) on a population. In addition to allowing one
to trace mortality-related effects, life tables can also be used to trace
other life-course events that may involve repeated entrances into
and exits from a status. Life tables that delineate only the impacts
of mortality are referred to as single-decrement life tables. Those
showing the impacts of mortality plus one or more additional factors
are referred to as multiple-decrement life tables.
Among the most common multiple-decrement life tables are
nuptiality tables, tables of school life, and tables of working life
which examine marriage, enrollment, and labor force participation
patterns respectively over the life course. Below, the basic compo-
nents and uses of these three forms of multiple-decrement life tables
are discussed and an example of the use of a table of working life in
discerning the compensation for a person involved in an accident
which led to permanent disability is presented. Although only these
forms of multiple-decrement life tables will be discussed here, as
noted above, the reader should be aware that multiple-decrement
life-table methodologies are likely to be applicable to any phenome-
na in which there is a population with incremental loss (i.e., mortal-
ity) over time and for which rates that are at least age-specific can be
obtained for the mortality factor and for one or more additional
factors of interest.
Figure 4.30 shows the unique values computed for nuptiality,
school life, and working life tables. All forms of these tables use the
results of the mortality component of a life table and apply the rate
for the factor(s) being added to the mortality-related components of

170
Hgure 4.30: Unique Components of Nuptlality Tables, Tables of
School Ufe, and Tables of Working Ufe
Nuptlality Table
Columnn
x
,
ColumnV
x
,
ColumnNX
Column%N
x
O'
Column ex
- Percent of population with first marriages at age x
- Number of persons with first marriages occurring at age x
- Number of first marriages at age x and all older
ages
- Percent marrying at age x and all older ages
- Average number of years of single life remalnhlg to persons
alive and single at the beginning of age x
School Ufe Table
Columnsx
ColumnLSX
Columnlsx
Column T
sx
O'
Columnesx
- Percent of population enrolled in school at age x
- Number living and in the school (enrolled) population at age x
- Number alive and in school at the beginning of age x
- Number of years remalnhlg in school at age x
and all older ages
- Average number of school years remalnhlg to persons
alive and enrolled at beginning of age x
Working Ufe Table
•
ColumnLw
x
•
Column'IW
x
0 •
Columnewx
- Percent of population in the Jabor force at age x
- Number living and in the Jabor force in at age x
- Number alive and in the labor force at the beginning
of age x
- Number of years remaining in the Jabor force in age x
and all older ages
- Average number of years in the labor force remalnhlg
to persons alive and in the labor force at the beginning
of age x

171
the life table. The factors unique to each of these types of multiple-
decrement life tables can be seen as columns added to a standard
life table. As the information in Figure 4.30 suggests, each of these
tables uses a rate, number of occurrences per 100,000 in the radix of
the standard life table, to compute the number of events for the
factor (i.e., first marriage, enrollment, or number of persons work-
ing). They all contain measures of the number of persons in the
state (married, enrolled, employed) and a measure of the number of
years at each age that would be spent in that state.
Nuptiality and school life tables have numerous uses for those
involved in public- and private-sector planning and other activities.
Nuptiality tables are useful for discerning such factors as age at first
marriage and for discerning the proportion of persons marrying at
each age. This information can be used to focus marketing and
advertising of products related to marriage. In addition, it is useful
in estate and related forms of planning, since such tables can be
used to identify the number of years a woman or man is likely to
live in a single status at older ages. School life tables can be used
for segmenting marketing and advertising and to discern changes in
educational patterns over time. For example, knowing how patterns
of educational involvement are changing by age and the proportion
of persons at different ages involved in education, can be used to
plan for levels and types of educational services (e.g., to determine
the level of need and types of educational services needed for
persons in older ages).
Tables of working life are clearly among the most used life table-
related products. Figure 4.31 shows a very simplified example of
one use of a table of working life, that of determining income loss to
persons who have been injured or disabled. It uses data from Table
4.3 to determine years of working life. This table has six columns.
The first column simply shows the age groups. Column two pro-
vides age-specific labor force participation rates used to discern the
number of persons in the labor force. Column three, the lw,(
column, presents the results of the application of the labor force
participation rates to the Ix column of the standard life table. It
indicates the number of persons alive and in the labor force at the
beginning of each age x. The Lwx value is computed as an
average of adjacent values from the lwx column in the standard

172
Figure 4.31: Example of Using a Table of Working Life
to DetermJne Income Loss
Given: the Life Table in Figure 4.11 and the Working Life Table
Components in Table 4.3
Example:
To determine the appropriate monetary settlement for a male
worker who was permanently disabled in an accident at age 35 and was
making the average annuaJ income for someone in hJs occupation. lhis
income was $37,635 in 1990.
-If one assumes that the worker had the average number of years
of working life remaining, then the income loss in 1990 dollars
would be:
25.1 years @$37,635 per year - $944,639
-If one assumes that the worker had the average number of years
of working life remaining and would receive two promotions at
ages 40 and 50 which would each lead to a 10% increase in real
income (using 1990 constant dollars):
income from 35 to 40 years of age, 5 years@ $37,635 - $188,175
Income from 40 to 50 years of age, 10 years@ $41,399 - $413,990
Income from 50 to end of working life, 12.7 years @45,539 - $578,345
Total Estimated Income Loss - $1,180,510

Table
4.3:
Components
of
a
Working
Life
Table
Derived
Using
a
Standard
Life
Table
(see
Figure
4.11)
Number
of
Number
years
Living
.Average
Remaining
Remaining
and
in
Number
of
in
Labor
Years
of
Percent
of
Labor
Force
Persons
Force
in
Working
Life
Age
Population
at
the
Living
and
Age
x
and
at
the
Interval
in
the
Beginning
in
the
La-bor
Al
1
Older
Beginning
(in
years)
Labor
Force
of
Age
Force
Ages
of
Age
x
..
.
.
..
0
..
x
to
x
+
n
(w
)
(
1
w
)
(Lw
)
(Tw
)
(ew
)
x
x
x
x
x
0
-
1
1
-
5
5
-
9
10
-
17
18
-
19
74.1
72,995
147,508
3,274,783
44.9
20
-
24
76.1
74,513
379,905
3,067,275
41.2
25
-
29
79.6
77
,448
385,170
2,687,370
34.7
30
-
34
79.3
76,620
382,645
2,302,200
30.0
35
-
39
79.7
76,407
379,
115
1,919,555
25.1
40
-
44
79.2
75,238
365,570
1,540,440
20.5
45
-
49
75.6
70,989
341,480
1,174,870
16.6
50
-
54
71.1
65,602
304,645
833,390
12.7
55
-
59
62.8
56,256
244,305
528,745
9.4
60
-
64
48.6
41,465
154,860
284,
440
6.9
65
-
69
25.9
20,479
79,570
129,580
6.3
70
-
74
16.0
11,
348
39,190
50,010
4.4
75
-
79
7.2
4,327
10,820
10,820
2.5
80
-
84
85+
-
-
-
-
-
1-1
~

174
way and indicates the total number of person years lived by persons
in the labor force between age x and x + n. Twx indicates the
total number of person years remaining for persons of age x and all
older ages in the labor force and is simply the sum of the Lwx
column. The final column indicates the remaining years of active
life for a person at the beginning of each of the age groups.
As the example in Figure 4.31 demonstrates, the number of
years of working life can be used to indicate how many years are
likely to remain in the working life of a person whose labor force
activity has been curtailed. Although the use shown is a very
simple one, it is evident that the table of working life provides a
means of simulating work life that is highly useful for labor-related
planning and other analyses.
Multiple-decrement life tables are useful for a variety of pur-
poses. Those involved in the analysis of phenomena that have a life
course like rate of incremental decline should evaluate the potential
use of such techniques in considerable detail (see, for example,
Namboodiri and Suchindran, 1987).
Conclusions
The goal in this chapter has been to provide an overview of
some of the basic measures and methods commonly used in applied
demographic analysis. Although numerous measures were de-
scribed, no single discussion can be exhaustive of the possible
measures and methods that might be used. The reader should be
aware of the need to gain familiarity with additional methods and
measures for assessing each of the factors discussed in the chapter.
It should be evident, however, that the size and distribution of the
population and the characteristics of populations can be examined
and described in a variety of ways which together provide a relative-
ly complete description of a population. For nearly all types of ap-
plied analysis, such a description is the first step in completing an
adequate assessment of the demand and/or market for a public or
private good or service. Knowledge of these basic measures and
methods is therefore essential for the applied analyst.

5
Methods for Estimating and Projecting Populations
Population estimates and projections are among the most widely
requested products of demographic analyses. They make use of
nearly all of the concepts and procedures discussed previously in
this book and so are in some senses among the more complex tech-
niques used in applied demographic analysis. Although they use
numerous relatively complex procedures, complexity should not be
confused with accuracy. Estimation and projection procedures are
only as accurate as the assumptions on which they are based, and if
the assumptions underlying them are incorrect, the estimates or
projections resulting from them will be inaccurate. Since one has
only the past on which to base these assumptions, and conditions
which existed in the past often change, the record of accuracy for
estimates (National Academy of Sciences, 1980) and projections
(Ascher, 1978) suggests that they are frequently inaccurate.
In many instances, however, there are simply few alternatives to
using some value for population variables in public- and private-
sector planning. Estimates and projections of the total populations
and population subgroups of states, counties, and subcounty areas
are essential for planning for services such as health care, schools,
highways, water, sewer, and similar services. In like manner,
estimates and projections of populations form a major basis for
determining the present and future markets for a variety of goods
and services and for other aspects of private-sector planning and
marketing efforts.
This chapter provides an introduction to basic methods of small-
area population estimation and projection. Its emphasis is on pro-
jecting total populations, although some methods that also produce
projections of population subgroups with specific characteristics such
as age, sex, and race/ethnicity are also presented. The methods
described can be used to prepare estimates and projections for a
variety of geographical levels, but emphasis is placed on methods
appropriate for preparing estimates and projections for county and
subcounty areas rather than for larger areas such as states or the

176
Nation or very small areas such as census tracts and blocks. It
should be recognized that the methods presented are only some of
the many methods available for completing such estimates and
projections. More complete descriptions of the methods presented
here and descriptions of other methods can be found in the refer-
ences cited at the end of the book. The authors particularly recom-
mend the works by Shryock and Siegel (1980), Pittenger (1976),
Irwin (1977), Haub (1987), and Murdock et al., (1987b).
This chapter is organized into five parts. The first part presents
basic definitions, principles, limitations, and general processes and
procedures used in population estimates and projections. The
second section presents a description and examples of widely used
population estimation techniques, while the third section presents
major projection methods. The fourth section briefly delineates the
role of population estimates and projections as the bases for esti-
mates and projections of several other population-based statuses and
characteristics such as labor force involvement and householder
status. The final section presents a discussion of procedures and
measures for evaluating population estimates and projections to
assess the accuracy of alternative methods and discern the nature of
errors likely to be produced by the use of alternative methods.
Bask Definitions and Concepts, Principles
and Limitations, and General Procedures for
Use in Population Estimation and Projection
Definitions and Concepts
Foremost among the distinctions usually made in this area of
analysis are the differences among population estimates, population
projections, and population forecasts. Population estimates refer to
population data obtained for periods which fall between dates for
which actual population counts are available, such as estimates for
1985 obtained by using 1980 and 1990 Census data, or determina-
tions of population for dates since the last population census (e.g.,
1991) for which data on actual counts could hypothetically have been
obtained. In other words, estimates refer to data obtained on
populations for past or present periods for which population cen-
suses are not available. In addition, in most instances, estimates
involve the use of data for the estimate date for components of
population change (i.e., births, deaths, or migration) and for factors
that have historically been closely related to population size and/or
change (e.g., housing units, school enrollment, vehicle registrations).

177
Projections, on the other hand, refer to determinations of future
populations. They consist of computations of future levels of popu-
lation that will exist in an area if certain sets of assumptions prove
to be valid. Thus, a projection of the population of an area in the
year 2000 based on the assumption that 1980-1990 fertility, mortality,
and migration levels continue from 1990 to 2000 is an example of a
population projection. Such projections will be correct only if the
assumptions on which they are based are correct; they consist of
little more than the tracing of the logical consequences of a set of
assumptions.
A population forecast also refers to an attempt to determine
future population levels. Unlike a projection, however, the term
forecast has a connotation of certainty and judgment that many
demographers wish to avoid. As many scholars point out, this
distinction is often recognized only by demographers (Keyfitz, 1972;
1982), and the terms forecast and projection are used interchange-
ably in discussions of demographic assessments. In this chapter,
however, we shall use the term projection to refer to attempts to
determine future populations.
Principles and Limitations
Whatever the terms used, however, it is clear that any estimate,
projection, or forecast is likely to vary in accuracy in accordance with
the characteristics of the estimation or projection area and the esti-
mation or projection technique. Shryock and Siegel (1980) note
several general principles which bear on the accuracy of estimates
and projections. Perhaps the most important of these is that noted
above that any estimate or projection is only as accurate as the
assumptions on which it is based and will only be correct if its
assumptions are correct. Because of this, the assumptions underly-
ing an estimate or projection must be examined critically. In addi-
tion, Shryock and Siegel (1980) note that population estimates and
projections are generally more accurate if performed
1. for an entire nation or large geographic region rather
than for a small component area or subregion;
2. for total populations rather than for population sub-
groups;
3. with series of data directly related to the determinants of
population change (birth, death, and migration data)
rather than data that provide indirect or symptomatic
indicators of population change;

178
4. for shorter rather than longer periods of time;
5. for areas in which past trends are more likely to continue
rather than new patterns to arise; and
6. for areas undergoing slow rather than rapid change.
For areas undergoing rapid change or experiencing substantial
departures from past patterns, these principles suggest that estimat-
ing or projecting populations in such areas will be difficult. It is
essential to remain cognizant of these principles in preparing and
evaluating population estimates or projections.
General Demographic Procedures Used in
Estimates and Projections
To understand many of the estimation and projection techniques
described below, a common base of knowledge regarding several
basic procedures and methods not previously discussed must be
obtained. Although some of the techniques discussed are used in
other areas of analysis as well, those discussed here are most fre-
quently used in conjunction with population estimates and projec-
tions.
The Population Equation~ Although the population equation
was described in Chapters 2 and 4, we briefly review it here because
it provides a useful model for differentiating population estimation
and projection techniques. As shown in Chapter 2, a population for
a given period of time is a product of the population at an earlier
period of time and of the births, deaths, and patterns of migration
which have occurred between the two time periods. All methods of
population estimation and projection either estimate or project the
Pt2 value directly or take population for the earlier period, Pq and
use vital statistics and migration data to obtain the population
estimate or projection for a later period. It is therefore useful to
recognize that most population estimation and projection procedures
are oriented to solving this simple equation for Pt2.
Computational Adjustments. The completion of population
estimates and projections often requires adjustments to the data
used in the estimates or projections. Although such adjustments are
numerous and no comprehensive review can be provided here, two
widely used procedures are briefly examined: controlling to a total
and accounting for the effects of special populations.

179
Estimates or projections made for counties or subcounty areas, if
computed separately for individual areas and summed across sub-
areas of a larger area, can produce unrealistic totals for the larger
area. For example, the use of the sum of county population esti-
mates or projections for a state estimate or projection or the sum of
states for a national estimate or projection will often result in values
that imply rates of population growth which differ markedly from
those made in the assumptions for the estimates or projections and
from those that are logical given historical events. Because of this, it
is essential to control the totals estimated or projected for subareas
to the total for their parent area. Figure 5.1 presents an example of
controlling to a total for the counties in the St. Louis, Missouri Area.
In general, controlling to a total requires obtaining each subarea's
proportion of the sum of the subareas' populations and applying
these to the total parent area population to get subarea •controlled•
population values.
A second general procedure merits discussion here. This proce-
dure is that of accounting for the effects of special populations.•
Special populations refer to subgroups of populations that have
demographic patterns that are distinct from those for the population
as a whole. Examples of special populations include: college
populations, institutional populations (hospitals, prisons, etc.),
military base populations, or similar groups. Special population
procedures are commonly employed if the proportion of the total
population composed of a special population group is sufficiently
large (e.g., the Census Bureau commonly uses an estimate of 5
percent or more of the total population as indicating a sufficiently
large special population to merit the use of special procedures).
When special population procedures are employed, special
populations are usually separated from the remainder of the popula-
tion and treated in one of two ways. One common procedure is to
assume a fixed number of persons in the special population with a
fixed set of characteristics (such as a given age and sex structure).
This procedure is commonly used for such special populations as
college populations in which the total size can usually be estimated
or projected and the age structure is likely to be relatively stable
over time. The second procedure is to develop a separate model for
the special population in which the same factors used to estimate or
project the value of the total population must be developed. For
example, if a component procedure (as noted below) is being used
to estimate or project the total population, this would involve de-
veloping a separate set of fertility, mortality, and migration rates for
the special population. Whichever method is employed, however,

180
Figure 5.1: Example of Controlling to a Total
Given:
· Hypothetical independent estimate of population for the St. Louis, Mo.
area in 1988 of 1,583,600
· Hypothetical Independent estimates of population for the counties in
the St. Louis, Mo. area In 1988 as follows:
Step 1. Determine percentages In each area using uncontrolled values
(e.g., 204,000/1,457,700 - 14%)
Step 2. Apply percentages from uncontrolled values to control value to
determine the controlled value for each area (e.g., 1,583,600
x .14 = 221,704)
Control led
1988 Percent Estimate
Coun t;i Estimate of Sum Value
St. Charles 204,000 14.00 221,704
St. Louis 1,005,900 69.00 1,092,684
Jefferson 169,800 11.65 184,489
Frankl in 78,000 5.35 84,723
Sum of
Counties 1,457,700 100.0 1,583,600
Figure 5.2: Projections for a College-Dominated County by Age for
1980-2020 NOT Adjusting for Special Populations
Age
Groups 1980 1990 2000 2010 2020
15-19 13,347 5,837 11, 679 10,427 8,698
20-24 23,543 5,899 6,908 13,828 ·8,645
25-29 8,652 13,781 6,003 11,972 10,691
30-34 6,102 23,285 5,874 6,892 13,670
35-39 4,281 8,569 13, 544 5,938 11,771
40-45 3,271 6,025 22,887 5,771 6,786
45-49 2,959 4,193 8,398 13, 254 5,787
50-55 2,708 3,156 5,813 22, 105 5,512
55-59 2,689 2,780 3,932 7,870 12,383
60-64 2,313 2,435 2,847 5,223 20,018
65-69 2, 113 2,300 2,368 3,350 6,738
70+ 4,275 5,181 5,776 6,431 9,269

181
the central point is that it is essential to identify such populations
because a failure to do so is likely to result in substantial distortions
in estimated or projected populations, particularly those containing
age and other detail.
An example of the difficulty entailed if special population proce-
dures are not employed is shown in Figure 5.2. This figure shows
projections for a county of about 120,000 persons (in 1990) with a
large university with enrollment of more than 40,000 students.
Without the use of special population procedures, the large cohorts
in the college ages are assumed to remain in the population and the
projections incorporating them produce highly distorted and inaccu-
rate results since a majority of the college population leaves the area
after graduation (normally during the ages from 21to25).
Methods of Population Estimation
In this section, we describe several of the most widely used
procedures for population estimation. These methods include:
1. Extrapolative techniques (e.g., the use of exponential
trends);
2. Symptomatic techniques (e.g., the use of building
permits, school emollment);
3. Regression-based techniques (e.g., ratio correlation); and
4. Component techniques (e.g., cohort survival, compo-
nent method II).
In general, extrapolative techniques, as their name implies,
involve methods in which simple linear or other trends based on
past periods are assumed to apply to the period from the last popu-
lation count to the estimate date (commonly referred to as the
estimation period). Symptomatic techniques involve the use of
variables or factors with a known relationship to population.
Change in these variables is believed to be indicative (or symptomat-
ic) of population change. Commonly used symptoms include build-
ing permits, school enrollment, electric meter hookups, births,
deaths, and vehicle registrations.
Regression techniques are based on the use of the statistical
procedures of multiple regression with populations being estimated
using values for symptoms for which data can be obtained for the
estimate date together with regression weights established from

182
historical periods. Whereas the first three sets of techniques general-
ly attempt to directly estimate total population (the Pt2 value in the
population equation), component methods use data on births and
deaths and some means of estimating migration for the estimation
period. Given data on the population for a known period of time
preceding the estimate date (the Pt1 value in the population
equation), births, deaths, and net migration can be added to the
population for the estimation period to obtain an estimate of the
population at the estimate date. Each of these methods has unique
strengths and limitations which are discussed below. ·
Extrapolative Techniques
Extrapolative techniques are methods which use patterns of population
change established from past time periods to estimate the population for an
estimate date. In general, trends are derived from data for the last or sever-
al recent census periods and used in a direct (or slightly modified) form to
extend a population value from the last population count to the estimate
date. Three of the most widely used trends are the arithmetic, geo-
metric, and exponential rates of growth shown in Figures 4.6
through 4.8. The rates of growth derived in the manners shown in
these figures are simply applied to the last population count (usually
the last census) using the formula for the determination of popula-
tion (also shown in these figures) to estimate the population for an
estimate date.
In addition to the use of these three rates of growth, techniques .
utilizing other patterns of change can also be employed. For exam-
ple, whereas arithmetic, geometric, and exponential rates employ
patterns that, when graphed, approximate a linear (straight-line)
pattern, there are numerous types of polynomial curves that may
also accurately characterize patterns for selected periods. Among
these are the Gompertz and logistic curves which became popular in
the work of Pearl and others (Pearl and Reed, 1920) in the 1920s and
1930s. Each of these two techniques involves a curve that is asymp-
totic over time. Whereas the Gompertz curve is somewhat skewed,
the logistic curve provides a smoother curve more closely resembling
a normal curve. The use of these curves generally requires the
availability of data for numerous historical periods. In general,
then, the use of these curves requires a larger base of data than that
needed for other simple extrapolative techniques. The formulas for
these curves directly follow:

Gompertz curve:
Logistic curve:
A+ Bx
l+e
Where: Pt =estimated population for date, t2
2
x • time (year for which the estimate is to be
made, i.e., 1989, 1991, etc.)
K =upper or lower asymptote (maximum or
minimum population for an area deter-
mined by analysis of historical time series)
A,B =constants derived from fitting population
time series to the nonlinear equations (for
either the Gompertz or logistic curve)
e = 2.718281828.... (a constant)
183
Whatever specific extrapolative procedure is utilized, the accu-
racy of estimates produced will depend on how similar the estima-
tion period is to the historical period on which the extrapolative base
patterns used to extend the base population values to the estimate
date are based. As a result, the limitations of these techniques lay
in their dependence on historical time periods to characterize estima-
tion periods.
In addition, it is important to recognize that, although these
techniques usually do not incorporate population characteristics in
their procedures and do not make explicit assumptions about the
demographic processes of fertility, morta1,ity, and migration, they
involve implicit assumptions that the structure and rates of demo-
graphic processes in the population during the estimation period are
similar to those during the historical period from which the patterns
used are derived. If the population's age or other characteristics
have changed from the historical period to the estimation period,
such that they lead to different patterns than in the past, then the
estimates may be affected in ways that extrapolative procedures
cannot anticipate. For example, the size of the baby-boom genera-
tion has produced marked increases in other demographic factors

184
such as the number of births and the number of households. If one
had used extrapolative procedures, employing historical patterns
derived from years in which small birth cohorts were in their child-
bearing ages to estimate the population of an area during an estima-
tion period when the baby-boom population group was in its peak
reproductive ages, the estimates would likely have been too low. In
sum, then, a major weakness of these methods is the fact that they
cannot easily simulate differences due to changes in the characteris-
tics of populations at different points in time.
These procedures have the advantage of being computationally
simple and requiring only readily available data. Thus, the proce-
dures noted above require little more than the population at two or
more periods of time in order to obtain the rates (or patterns) neces-
sary to estimate a population. Such techniques are most likely to be
used for estimating populations for time periods immediately after
the last census and for completing estimates when time is limited.
They are also relatively widely used to estimate patterns for small
component areas within larger areas with estimates for the larger
areas being used to control the sum of the estimates for the compo-
nent areas.
Symptomatic Techniques
Symptomatic techniques use data on selected factors to estimate popula-
tion; change in these factors is seen as being indicative or symptomatic of
population change. Although a variety of symptoms are used with
these procedures, nearly all symptomatic techniques can be seen as
reflecting the basic formulation shown in Figure 5.3. We refer to
this as the censal-ratio method because it relies on a ratio of a
symptom to a population at a census date. It must, however, be
differentiated from the more elaborate sets of techniques the basis of
which were established initially by the work of Bogue (1950) and
have been extensively developed to include a range of rather com-
plex procedures (Voss et al., 1992).
As an examination of the formula in Figure 5.3 implies, censal-
ratio methods generally involve first establishing the ratio between
population and a symptom at a point in time (usually the last
decennial census) when there was an accurate count of both the
symptom and population for the estimation area. Data on the
symptom for the estimation date are then used together with the
ratio established for the known (census) date (either unchanged or
with the assumption that the ratio has changed in some known
manner) to obtain an estimate of the population.

Figure 5.3: Censal-Ratio Method with Symptomallc Data
Where: P = Population on estimate date
t2
S • Symptom value on last census
tl
U = Net change in symptom from census to
estimate date
Ratio of persons per symptom item at
St the last census date
1
185

186
Because data on a symptom for the estimate date are used to
estimate population, the accuracy of symptomatic techniques is
determined by the extent to which the ratio of the symptom to
population remains unchanged (or changes in a known manner)
between the base date and the estimation date and on how accurate-
ly the symptom is measured at the estimation date. This accuracy is
a function not only of one's ability to obtain accurate data on the
symptom, but also on obtaining such data for a time period and area
consistent with the date and location for which the population
estimate is to be made. Many symptomatic data are produced for
other purposes (than population estimation) and for areas that may
not coincide with the estimation area. For example, school enroll-
ment is often measured in the fall of the year whereas population
estimates are usually completed for either July 1 or January 1, also
school district boundaries often do not coincide with the boundaries
of estimation areas such as towns or cities. As a result, both the
time referent and the geographic area covered by a symptom often
must be adjusted before symptom data can be considered for popu-
lation estimation.
Housing Unit Methods. There are a number of symptomatic
techniques that use symptoms of household change to estimate
population. Among the most widely used of these techniques is the
housing unit method using indicators such as building permits and
electric meter connections. The steps in completing estimates using
these indicators and examples of their use are shown in Figures 5.4
and 5.5.
Although these steps need not be discussed in detail here,
several aspects of these two methods should be noted. In the use of
the housing unit method, it is essential that data on demolitions as
well as building permits be obtained. Local planning offices often
have more accurate records of new construction than of demolitions
so that care is likely to be needed to obtain complete and accurate
data on demolitions. It should also be noted that the accuracy of esti-
mates made using the housing unit method is dependent not only on accu-
rate data on changes in housing units, but also on accurate information on
the average size of households and on vacancy rates and changes in these
from the base date to the estimate date. It is usually the failure to obtain
accurate information on these latter two factors that has led to
problems in the use of the housing unit method of population
estimation. Among the best means of obtaining information on
changes in average household size and in vacancy rates is to use
information from such sources as the P-20 series from the Current

187
Population Survey which provides information on intercensal
changes in household size at the national level and/or to gather
information from periodic surveys of households in the estimation
area. However addressed, attempts must be made to update these
parameters if accurate estimates are to be made using the housing
unit method.
The use of electric meter and/or other utility data also requires
that care be taken to examine changes in average household size
and in the quality of data on households whose utilities have been
disconnected. In addition, it is essential to determine the number of
master meters (i.e., meters in which the utility use of several sepa-
rate households is recorded on a single meter) and the number of
units attached to them. It is also often necessary to reconcile the
area for which utility data are available with the estimation area. A
single utility may not cover an entire estimation area so that infor-
mation may be required from different utilities. In other locations,
the area covered by a utility will be larger than the estimation area.
Similarly, zip codes on customer addresses, which are sometimes
the only areal data available on utility customers, are often not suffi-
ciently precise to determine a customer's exact area of residence
without adjustments. As in the use of building permits, then,
numerous adjustments may be necessary to use utility data for
population estimation.
In sum, the advantages of the use of symptomatic techniques
using estimators of the number of households include the use of
data that can be readily updated and reliance on the relatively
strong and established relationship between the number of house-
holds and population. The disadvantages stem from difficulties in
obtaining data on change in average household size and on vacancy
levels. In fact, problems in obtaining data on these two critical
factors have historically been relatively severe. As a result, compari-
sons of the results of estimates for the 1970s and for 1980 to the
results of the 1980 Census showed such substantial errors in housing
unit-based estimates that the use of the housing unit method was
sharply curtailed in the years immediately after 1980.
Work done by Smith (1986), however, has shown that the accu-
racy of housing unit-based estimates can be substantially improved
by the use of separate vacancy and household size estimators for
different types of housing (i.e., single-family, multiple-family,
mobile home, etc.), by using an average of multiple indicators of
household change (e.g., building permits, electric meters, and/or
telephone connections), through the use of data from current sur-
veys to update estimates of household size and occupancy rates, and

188
Figure 5.4: Censal-Ratio Procedure with Housing Permit Data: To Estimate the Austin,
Texas Population for Aprill, 1984
Step 1.
Step 2.
Step 3.
Given the formula below and assumJng that Pt /Ht did not change since
the last census. 1 1
pt
1
Pt = (Ht + U) x -
2 1 Ht
1
where: Pt - population for estimate date
2
Ht • occupied housing units on the
1 last census date
U - net change in occupied housing
units between the census date
and the estimate date
pt
1
- - average number of persons per
Ht occupied housing unit at the
1 last census date
Obtain the census count of the total population and total number of
occupied housing units in the dty on the census date.
Pt (April l, 1980) .. 345,106 persons
1
Ht (April I, 1980) • 133,934 occupied units
1
Obtain the number of housing units added to the housing stock since the
last census date.
Number of Housing Units Added !o Housing Stock of Aust in
Type of Unit 1980 1981 1982 1983 1984
Single family 2,797 2,336 2,495 2,586 2,489
Mui t ipl e family 3,446 6,245 5,730 14, 336 9,341
Mobile home 2,436 3,358 3,208 6,600 4,614
Total 8,679 11, 939 11,433 23,522 16,444
Step 4. Obtain the number of demolitions since the census date (April 1, 1980)
Year 1980 1981 1982 1983 1984
Demo I i t ions 65 95 107 140 231
(amtinues)

189
Step5.
Step 6.
Step 7.
Adjust the figures so they are comparable.
Census figures for April 1, 1980 Included the housing changes for the
period from January 1, 1980 to April 1, 1980, so units added in January,
Februaly, and Man:h are Included both in the census counts and in the
housing stock data. These units (3/12 of all 1980 units) must be sub-
tracted from the total. Also since the housing stock values for 1984 are
for the entire year and the estimate is for April 1, 1984 only 3/12 of
1984 values are Included.
Adjust total number of housing units in housing stock for April 1, 1980
to December 31, 1980 and for January 1, 1984 to April 1, 1984:
9/12 (2,797 + 3,446 + 2,436) = 6,509
3/12 (2,489 + 9,341 + 4,614) = 4,111
Adjust number of demolitions in similar manner. Adjusted demolitions
for April l, 1980 to December 31, 1980 and for January 1, 1984 to April
1, 1984:
65 x 9/12 = 49
231 x 3/12 = 58
Step 8. Add housing units added to housing stock from Aprill, 1980 to April
1, 1984:
6,509 + 11,939 + 11,433 + 23,522 + 4,111 = 57,514
Step 9. Subtract demolitions since April 1, 1980 from the total units added
since April 1, 1980:
57,514 - (49 + 95 + 107 + 140 + 58) = 57,065
Step 10. Because we are interested in occupied units only, the number of vacant
units must be subtracted from the total number of housing units.
Assuming a vacancy rate of 9.0 percent, which was the local vacancy
rate at the census date of April 1, 1980:
57,065 - (57,065 x .09) = 51,929
Step 11. Add total number of occupied units on census date to number of occu-
pied housing units added since April 1, 1980 to determine total number
of occupied housing units at the estimate date, April 1, 1984.
133,934 + 51,929 = 185,863
Step 12. Determine the average number of persons per household at the census
date by dividing the population by the number of occupied housing
units on the census date.
345,106/133,934 • 2.57669
Step 13. Compute estimate of population,. Aprill, 1984, by multiplying number
of occupied housing units at estimate date by the average number of
persons per household:
185,863 x 2.57669 s 478,911

190
Figure 5.5. Censal-Ratio Method Using Electric Meter Billing: To Estimate the
Austin, Texas Population for April 1, 1988
Step 1.
Step 2.
Step 3.
Step 4.
Step 5.
Given the basic formula below and assuming no change in the ratio of
persons served per electric meter since the last census date.
pt
1
Pt • (Mt + U) x -
2 1 M
tl
Where: pt - population for estimate date
2
pt - population on census date, (e.g.,
1 April 1, 1980)
~ -number of residential electric
1 meter bllllngs on the census date
u - net change in electric meter
billings between the census date
and the estimate date
pt
1
average number of persons per
Mt occupied housing unit at the
1 census date
From the census, obtain the population (i.e., 345,106), and from the
utillty's office, obtain the total number of residential electric meter
b1lllngs (i.e., 146,338) that were active on the last census date, adjust-
ing as necessary for geographic and other differences.
Determine the ratio of population per meter on the census date:
345,106 + 146,338 - 2.35828
Obtain the current number of active meters adjusting for vacant
housing units with active meters, multi-family units using master
meters, and annexations since the census date. The local
electric/power company can provide this data for the month desired
for the estimate date. Active residential electric meter billings on
April 1, 1988 • 215,182
Calculate the total population by multiplying the number of active
meters on the estimate date by the ratio of population per meter:
215,182 x2.35828 - 50'7,4.59

191
by using ratios of the symptoms to total population rather than
using change in the symptoms to measure population change. The
incorporation of such changes has resulted in renewed emphasis on
the housing unit method.
Other Ratio-Based Methods. Another means of implementing
the basic logic of symptomatic techniques involves the use of the
average of simple ratios including those ratios which are involved in
vital statistics measures such as crude birth and death rates. Figure
5.6 shows an example of the use of an average of the change in the
ratios of births and deaths to estimate population at the estimate
date.
Figure 5.7 shows the formula for what is referred to as the vital
rates method. It uses the ratio of a vital rate for the estimation area to that
for a larger area of which the estimation area is a part, together with data
on the change in the rate for the larger area and the number of vital events
for the estimation area for the estimation date to estimate population. Since
one can determine the population in an area if both the vital rate
and number of events in the area are known, data on changes in the
rate for the larger area together with data on the number of vital
events (i.e., births or deaths) in the estimation area at the estimate
date can be used to estimate population.
Yet another example of a symptomatic technique involves using
the ratio of an estimation area's population to the population in a
larger or parent area. This ratio can be applied to a population
estimate for a larger area to obtain an estimated population for the
area of interest. That is, ratios are used to prorate the. larger area's
population to subareas. The formulation and example shown in
Figure 5.8 uses a simple proration technique to allocate part of a larger
area's population to a subarea on the basis of the share that the estimation
subarea's population was of the larger area's population at an earlier time
period. These prorationing techniques often also involve the use of
techniques in which the estimation area's share of the larger area's
population is trended relative to past trends in such shares. Prora-
tion, although a simple procedure, is among the most widely used
procedures, particularly for obtaining estimates for small areas such
as census tracts within counties and census blocks within tracts.
One final symptomatic technique to be discussed here employs
different symptomatic indicators to estimate the population in
different population segments. Figure 5.9 shows example compo-
nents of this method which is commonly referred to as the

192
Figure 5.6: Example of Simple Ratio Technique
To estimate population of New Orleans MSA for April 1, 1989
Given:
pt
pt
Population of New Orleans In 1980 -
Births In New Orleans In 1980
Births In New Orleans in 1986
Deaths In New Orleans in 1980
Deaths In New Orleans in 1989
20,192
1,256,668
21,783
20,192
10,483
10,873
= Births • = 0.92696
21,783
10,873
Deaths 1.03720
10 ,483
n
I: Ri
i=l
= x pt 1
2 n
1.96416
x 1,256,668
2 2
Estimated Population of New Orleans in 1989 - 1,234,149

193
Figure 5.7: Vital Rates Method
Uses crude vital rates for subarea and superarea and trends In rate for su-
perarea to estimate population In subarea
I I s s
Rt Rt1 x (Rt I Rt )
2 2 1
Where: R! 'Estimate of vital rate for local (sub-
t2 area) for estimate date
R! • Vital rate for local area for known
tl date (usually last census)
Rs • Yitai rate for superarea for
t2 estimate date
R5 • Yitai rate for superarea for known
tl date (usually laat census)
Total Number of Events
Since: Vital Rate (VR) -
Total Population
Total Total Number of Events
Population -
Vital Rate
If you know the rate for an estimate period and number of vital events for
that period, the total population can be derived

194
Figure 5.8: Example of the Use of a Proration Technique
Where: pt Population counted in last census
A 1
pt
2
Population estimate
t I Census date
s = Super area
l Local or subarea
Example of Proration Method:
To estimate the population of the St. Louis, Mo. MSA In 1989 given a
state population estimate for 1989. ,
Given:
Population In St. Louis MSA In 1980 - 1,778,504
Population In State In 1980 - 4,916,686
Population In State In 1989 - 5,159,000
Al pl
AS
pt tl pt
-- x
2 2
pS
tl
1,778,504
x 5,159,000
4,196,686
.4238 x 5,159,000
Estimated Population of St. Louis In 1989 - 2,186,384

Figure 5.9: Composite Method
Use change In different symptom Indicators to measure change In different
eohorts (subgroups) of the population
For example:
Symptom
5 years of age - Births for last five years
5-17 years of age - Persons enrolled In elementary and secondary
school
Females 18-44
years of age
Males 18-44
- General fertility rate
- Sex ratio (applied to number of females 18-44)
Population 45-64 - Crude death rate for persons 45+ (years of age)
65+ - Medicare data
195

196
composite method. As shown in this figure, different symptoms are
used to estimate the population in each age group or cohort with the
total population being simply the sum of composite estimates. This
method, although appealing from the standpoint that it attempts to
use population characteristics to increase the accuracy of its esti-
mates, is not widely used because of its extensive data require-
ments. However, composite techniques may be potentially useful if
estimates of population segments or cohorts are desired.
In sum, symptomatic techniques are quite useful for estimating
population because of the ready availability of many of the most
widely used symptomatic indicators (e.g., building permits, births,
deaths) and because they have a long history of use which provides
a basis of experience in their application. The major weakness of
symptomatic methods is that changes in the relationships between
the symptoms and population are difficult to identify. Continuous
examinations of the major premises underlying the relationships
between the symptoms and population are essential.
Regression-Based Methods
Regression-based methods employ the statistical procedures of
simple or multiple regression. The statistical formulas underlying
these techniques are normally expressed in the following form:
Simple regression: y = a + bx
Multiple regression: y = a+ btx1+ bix2+bJX3 ...
Where: y =dependent variable to be estimated
(i.e., population)
a = intercept
~ = regression coefficient, weight or
slope (unit change in y per unit
change in x) of independent vari-
able(s) i
Xi = value of independent variable(s) i
These formulas express a relationship between a dependent
variable and one or more independent variable(s) in terms of a
linear relationship that, if graphed, produces a straight line that

197
begins at the intercept (a) on the y-axis of a normal x-y chart and is
drawn so as to minimize the error of the line as an estimate of y-
values given the x-values. The b-value(s) indicate(s) the slope (angle
of incline) of the line that minimizes that error. When multiple
independent variables are used the minimization is equivalent to an
analysis using multiple dimensions of space rather than the two
dimensions commonly used in graphically representing information.
Multiple regression is thus, the procedure for finding the best fitting
line (that which minimizes error) given the values of the independ-
ent variables. The basic principles of regression will not be
reviewed further here because it is a well-known technique of suffi-
cient complexity that it cannot be described in detail in the space
available (however, readers unfamiliar with the technique may wish
to consult one of numerous excellent texts on this technique such as
those by Pedhazur [1982], Ierlinger and Pedhazur [1973], Ott (1988],
and Snedecor and Cochran [1967]). Those wishing to use regres-
sion-based techniques for completing population estimates should be
aware that all of the assumptionS that underlie the use of multiple
regression in other areas of analyses (e.g., the assumptions of linear-
ity, low multicollinearity, homoscedascity) must also be met to use it
for completing population estimates.
As used in completing population estimates, multiple regression
involves the use of multiple independent variables (the x-values) to
estimate population (they-variable). Independent variables would
normally include factors, such as those used above in symptomatic
techniques, which are seen as affecting or varying with population
change. Among the numerous commonly used independent varia-
bles are births, deaths, school enrollment, employment, building
permits, and electric meter hookups. These are similar to those
variables used in symptomatic techniques and regression can be seen as
a method of population estimation using multiple symptomatic indicators
with variable weights for the indicators. The major difference is that
whereas the symptomatic techniques using multiple variables give
the values of symptomatic variables equal weight (that is, they
generally use simply the average of the variables' values), regression
procedures use standard regression procedures so that the weights
used are the b-values or regression coefficient weights that indicate
the relative weights to be used for each symptom.
The use of multiple regression to estimate population can be
seen as a three-step procedure. The first step involves obtaining
regression coefficient values (weights) for the independent variables
and the intercept value for a period of time believed to be similar to
the estimation period. Often the period used is the period between

198
the two most recent censuses with a regression computation being
completed showing the effects of the independent variables on
population. The results from the regression model for the historical
period are assumed to apply to the relationship between the varia-
bles and population at the estimate date.
The second step involves obtaining independent variable values
(x-values) for the independent variables for the estimate date. These
values would be those for such independent variables as the num-
bers of births, deaths, or whatever the specific variables used in the
historical regression model. The third step, given the variable
values for the estimate date and the regression weights for the
independent variables and for the intercept value, is simply to use
the values and weights to estimate population using the multiple
regression formula shown above. Thus, one adds the intercept (a)
value to the product of the first independent variable value (x1)
multiplied by its regression weight (b1), plus the product of the
second independent variable value (x2) multiplied by its regression
weight (b2), etc.
Although multiple regression is sometimes used directly to
estimate population, the most widely used regression-based tech-
nique for population estimation is the ratio-correlation technique. The
ratio-correlation method is a multi-variable proration technique in which the
ratios of variable shares (subarea to superarea) are used as independent
variables in a regression formulation. Thus, as shown in the example in
Figure 5.10, the regression formula is expressed in terms of ratios.
The ultimate goal of a ratio-correlation analysis is to determine how
to allocate a population estimate for a larger area (usually obtained
from a source other than yourself) to the subareas of interest.
The example in Figure 5.11 shows an example for determining
the estimated population of the city of Waco, Texas for 1982 given
the population of its parent county, McLennan. In this example, a
regression analysis of the ratios of the independent variables for
cities in McLennan County was completed for the historical period
from 1970 to 1980 with the values for the intercept and independent
values shown in the figure (0.72 for a, - 0.01 for bl, etc.) derived
from an analysis of the data for 1970 and 1980 shown in the table
within this figure. That is, the values for each city were placed
within the formulation shown in Figure 5.10. The historical regres-
sion completed using 1970-1980 data for all places (Waco, Bellmead,
etc.) in McLennan County was the basis for the intercept and regres-
sion coefficient values shown in Figure 5.11. Given the regression

Figure 5.10: Steps for Completing an Estimate Using the
Ratlo-CorreJatlon Method
199
The standard steps tn completing a ratto-correJatlon estimate are:
Step 1.
+ b2
+ b3
Step 2.
Use regression analysis for an historical period (such as 1970-80)
to obtain coefficients using a formulation such as the following:
Pop. (City, 1980)/Pop. (County,1980)
Pop. (City, 1970)/Pop. (County, 1970)
Births (City, 1980)/Births (County,1980)
Births (City, 1970)/ Births(County,1970)
Deaths (city I 1980)/Deaths (County,1980)
Deaths (City, 1970)/Deaths (County,1970)
Sch. Enr. (City, 1980)/Sch. Enr. (County,1980)
Sch. Enr. (City, 1970)/Sch. Enr. (County,1970)
Apply coefftdents to change during the recent time period (the
estimation period), substitute symptomatic Indicators' values, the
Intercept and coefficients' values for the estimation period Into the
ratio-correlation equation and solve the equation for the popula-
tion of the subarea of interest (e.g., population of the city tn 1980).

Figure
5.11:
Ratio-Correlation
Method:
To
Estimate
Population
of
Waco,
Texas,
1982
Step
1.
Input
data
on
births,
deaths,
school
enrollments,
and
total
population
for
McLennan
County
and
cities
within
the
county
for
1970
and
1980
and
obtain
an
Intercept
and
coefficients
through
the
completion
of
an
historlcal
regression
analysis.
The
variable
values
used
as
Input
In
the
regression
are
the
ratios
of
the
cities'
shares
of
county
values
for
each
variable.
For
example,
the
Input
value
for
births
for
Waco
would
be
(1,9'1212,867)1(1,73912,430)
-
0.96
1970
1980
School
School
Area
Births
Deaths
Enrollment
Population
Births
Deaths
Enrollment
Population
McLennan
County
2,430
1,6U
19,155
147,
553
2,867
1,684
21,911
170,755
Cities
Waco
1,739
1,
188
11,
870
95,326
1,972
1,223
12,152
101,
261
Bellmead
105
46
838
7,698
105
45
633
7,569
Lacy-Lake-
View
15
20
248
2,558
34
13
274
2,752
Robinson
19
35
960
3,807
32
24
1,278
6,074
Woodway
12
14
804
4,819
20
11
1,711
7,091
McGregor
63
48
674
4,365
93
26
703
4,513
The
coefficients
derived
from
the
regression
analysis
of
1970-80
patterns
are:
b
1
(for
births)
=
-0.01
b
2
(for
deaths)
=
-0.12
b
3
(for
school
enrollment)
=
0.42
a
0
(the
intercept)
=
0.72
(amtinues)
~

Figure
5.11
(amtinutd)
Step
2.
Substitute
Intercept
and
coefficients
obtained
from
the
Step
1
regression
equation
Into
the
ratio-correlation
formula
for
the
Oty
of
Waco,
1980-82.
Input
Data:
1980
1982
School
School
Area
Births
Deaths
Enrollment
Population
Births
Deaths
Enrollment
Population
McLennan
County
2,867
1,684
21,911
170,755
2,819
1,717
22,498
172,888
Cities
Waco
1,972
1,223
12,152
101,261
1,889
1,252
13,011
Bellmead
105
45
633
7,569
131
44
749
Lacy-Lake
View
34
13
274
2,752
32
14
280
Robinson
32
24
1,278
6,074
41
30
1,423
Woodway
20
11
1,
711
7,091
19
5
1,778
McGregor
93
26
703
4,513
80
25
655
(continues)
~
....

Figure
5.11
(continued)
Step
2
Continued:
Solve
the
equation
for
the
population
of
Waco
for
1982
Pop.(City,
1982)/Pop.(County,1982)
Births(City,
1982)/Births(County,1982)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
=
a
+
b
Pop.(City,
1980)/Pop.(County,1980)
0
1
Births(City,
1980)/Births(County,1980)
Waco
Pop.,
1982/172,888
101,261/170,755
Waco
Population,
1982
-
105,370
Deaths(City,
1982)/Deaths(County,1982)
+
b2
Deaths(City,
1980)/Deaths(County,1980)
School
Enroll.(City,
1982)/
School
Enroll.(County,1982)
+
b3
-
-
-
-
-
-
-
-
-
-
-
-
-
School
Enroll.(City,
1980)/
School
Enroll.(County,1980)
1,252/1,717
0.72
+
(-0.01)
1,889/2,819
1,972/2,867
13,011/22,498
+
(-0.12)
1,223/1,684
+
(0.42)
12,152/21,911
s

203
weights and intercept value and the overall ratio-correlation formu-
lation for the regression model, the population for 1982 can be
estimated simply by £illing in the data for all variables in the formu-
lation (as shown in Figure 5.10 except. values for 1980 and 1982
replace those for 1970 and 1980) for which the values are known.
Only the value for the population of Waco in 1982 is unknown, and
by solving the equation for the Waco population in 1982 (or for any
other of the cities in the county for which an estimate is desired),
one can obtain the estimated population value for Waco.
The ratio-correlation technique is among the most widely used
methods of population estimation, because it employs multiple indi-
cators of population change and allows one to use estimates for
larger areas for which estimates are likely to be more accurate and
which may be available from other sources. Its weaknesses are
those inherent in the use of multiple regression and in the assump-
tions that ratio relationships between indicators and variables either
remain the same as at the base period or have patterns of change
during the estimation period that can be determined.
Ratio-correlation techniques have been extensively analyzed with
numerous alternatives to the basic formulation having been recom-
mended. Namboodiri and Lalu (1971) have suggested the use of the
mean of several simple (one independent variable) regressions in-
stead of multiple regression. Rosenberg (1968) and Pursell (1970)
have suggested the use of multiple strata of areas grouped by
common features as a means of increasing accuracy. Ericksen (1973)
has recommended the use of current survey data to update the
ratios and O'Hare (1976; 1980), Swanson (1978; 1980), and Swanson
and Tedrow (1984) have suggested that using the differences in
shares rather than the ratios of shares may provide superior esti-
mates. Despite extensive analyses, however, no one formulation
has been found to provide better estimates in all types of areas. The
users of regression-based techniques (including ratio-correlation
methods), should examine regression results for historical periods to
determine which of the alternative formulations is likely to be the
most accurate for the estimation area they are analyzing.
Component Methods of Population Estimation
Component methods of population estimation use the components of
population change-births, deaths, and migration-togdher with a base
population value to estimate population at the estimate date. In large part,
then, this method entails obtaining the values for the elements in
the population equation. This method is used both for the total

204
population and for completing estimates for individual age, sex, or
other cohorts in a population (in which case it is referred to as a
cohort-component model). It is a relatively data-intensive method,
particularly if used in a cohort form, since death and migration data
must be obtained for each cohort for the estimation period. Howev-
er, if such data can be obtained, a cohort-component population
estimate can be used to estimate not only the total population but
also the population of individual or groups of cohorts that may be
useful for planning services for specific market segments and clien-
tele.
In general, component and cohort-component methods use vital
statistics data on births and deaths from health department records
together with an indicator of the migration component. Component
methods of population estimation tend to differ in terms of the
factors used to estimate the migration component. Although
numerous methods are available, the following are the three most
widely used component methods:
1. Administrative records method
2. Component method II
3. Cohort-component method.
We briefly discuss each of these methods below.
Administrative records methods, as the name implies, use records ob-
tained for other administrative purposes to estimate migration. Perhaps
the most developed of such methods is the administrative records
method used by the U.S. Bureau of the Census to estimate county
and place populations. The specific administrative record used by
the Bureau is income tax returns showing dependents claimed for
tax purposes. By comparing matched Internal Revenue Service (IRS)
forms for adjacent years, migrants are identified as those persons
who file in two different areas in adjacent years. Because of strict
guidelines on confidentiality, IRS data are available only to person-
nel within the bureau. However, aggregate-level IRS data are avail-
able at the county-level and this method provides a useful model of
how administrative records can be used to estimate the migration
component. Those interested in more details on this method, as
used by the U.S. Bureau of the Census, may wish to obtain a paper
by van der Vate (1988) which provides a relatively detailed descrip-
tion of this and other methods used by the U.S. Bureau of the
Census.
Component method II uses school enrollment data as a means of esti-
mating the migration component. Specifically, this method uses an estimate

205
of the migration rate of school-age children to estimate the migration rate of
the total population. The ratio of school-age migration to total popula-
tion migration is established for a base period (usually the last
census) and used with estimates of school-age qiigration for the
estimate date to estimate the migration of the total population.
Component method II has been widely used and the details of
its computational steps are readily available in numerous sources
(see, for example, Murdock et al., 1992). Any review of this method
clearly indicates that the key factor underlying the accuracy of
estimates made with this method is the stability of the ratio of
school-age to total population migration, referred to as the migration
adjustment factor. Han area has undergone very rapid inmigration
and its population is one which has relatively few children, the
migration adjustment factor may account for a very large percentage
of the migration estimated.
The strength of component method II lies in the fact that it
provides a relatively straightforward method for population estima-
tion that is well developed and widely tested. Its weakness lies in
its assumption that the relationship between the school-age migra-
tion and the migration of the total population is sufficiently stable
that migration of school-age children can be used to estimate total
migration. As with all methods of population estimation, an exami-
nation of the validity of the key assumptions of the method is essen-
tial prior to its use in any specific area.
The colwrt-component method uses data on migration and mortality for
individual age cohorts (or cohorts that are age as well as other characteristic
specific) and data on birth rates to estimate population for the estimation
period. As shown in Figure 5.12, populations are moved from a base
date forward to an estimate date by applying the components in the
manner designated in the population equation. Deaths are account-
ed for by using age- and (in this example) sex-specific survival rates
derived from a life table to estimate the effects of mortality. Migra-
tion levels were assumed to follow the age-specific patterns of the
1970s, and births, which become the beginning-of-life cohorts, are
assumed to reflect the rates of 1980 applied to the women of child-
bearing age during the estimation period.
An examination of the step-by-step example presented in Figure
5.12 reveals several important factors about the use of cohort-
component models. First, it is clear that the cohorts selected for use
determine the extent and types of data required to implement the
method. The level of specificity in the data required must be equiv-

Figure
5.12:
Steps
in
and
Example
of
the
use
of
a
Cohort-Survival
Method
of
Population
Estimation
to
Estimate
the
Popula-
tion
of
McLennan
County,
Texas,
April
1,
1988
Steps
for
calculating
male
and
female
population
with
the
cohort-survival
method
are
provided
below.
Note
that
although
the
estimate
date
is
1988,
we
complete
an
estimate
for
1990
and
interpolate
between
1980
and
1990
values
to
obtain
the
value
for
the
estimate
year
of
1988.
An
estimate
for
1990
was
completed
because
only
rates
for
5-year
cohorts
were
available
so
that
the
use
of
an
estimation
period
divisible
into
5-year
elements
was
desirable.
Step
1:
Ust
the
population
enumerated
in
the
past
census
(in
this
example,
1980)
in
column
1.
Step
2:
Ust
10-year
life-table
survival
rates
computed
from
a
life
table
or
obtained
from
a
standard
source
in
column
2
and
10-year
net
migration
rates
in
column
4.
Step
3:
Obtain
expected
survivors
(column
3)
to
1990
by
multiplying
the
age-specific
survival
rate
(column
2)
by
the
1980
population
values
(column
1).
Add
the
product
derived
by
multiplying
the
population
in
the
last
3
cells
in
column
1
by
their
respective
survival
rates
to
obtain
the
value
for
the
last
cell
in
column
3.
Step
4:
Project
net
migration
(column
5)
between
1980
and
1990
for
each
cohort
by
multiplying
expected
survivors
in
column
3
by
the
migration
rates
in
column
4.
Step
5:
Project
population
to
1990
for
each
age
cohort
(column
6)
by
adding
columns
3
and
5.
To
obtain
the
1990
projected
population
for
the
0-4
and
5-9
age
cohorts,
use
the
appropriate
age-specific
birth
rate
multiplied
by
the
number
of
females
in
the
child-bearing
years
(ages
10
to
49)
as
shown
in
Panel
C.
Since
5-year
cohorts
are
being
used,
births
for
two
separate
5-year
periods
must
be
computed
to
obtain
the
population
values
for
the
0-4
and
5-9
cohorts.
Multi.ply
the
total
numbers
of
births
by
5
to
account
for
5
years
of
births.
Use
the
proportion
of
male-to-female
births
to
allocate
birth
to
sexes,
in
this
case
51
percent
for
males
and
49
percent
for
females.
Apply
the
five-year
survival
rate
and
the
five-year
net
migration
rates
appropriate
for
males
and
females
for
the
0-4
and
5-to-9
cohorts
formed
from
the
births.
For
example,
of
the
16,285
children
born
during
the
period
1980-1985,
8,305
would
be
males.
Use
the
male
survival
rate
for
the
Oto
4
cohort
(.9'725)
and
multi.ply
it
by
8,305
to
obtain
8,077.
Multiply
the
migration
rate
of
.0253
by
8,077
to
obtain
204
inmigrants
and
add
this
to
8,077.
The
result,
8,281,
is
then
survived
and
migrated
by
the
5-to-9
cohort
rates
to
obtain
a
final
value
of
8,708
for
the
5-9
age
group
which
is
placed
in
the
second
cell
of
column
6.
To
obtain
the
second
5-year
increment
of
births
from
1986-1990,
after
the
child-bearing
female
population
is
survived
and
migration
rates
have
been
applied,
multiply
each
cohort
by
their
age-specific
fertility
rate.
The
result
(shown
on
Panel
q
is
that
18,315
children
will
be
born
in
the
1986-1990
period,
of
which
9,341
will
be
male
and
placed
in
the
first
cell
of
column
6.
Step
6:
To
obtain
the
1988
estimate,
it
is
necessary
to
interpolate
between
the
1980
census
count
and
the
1990
estimate.
To
do
so
multiply
column
1
by
.20
and
column
6
by
.80
and
add
the
two
products
together.
Place
the
result
in
column
7.
The
exceptions
to
this
step
are
the
first
of
two
cells
In
column
6
which
are
not
interpolated.
(amtinues)
~

Figure
5.12
(amtinuetl)
Panel
A:
Estimate
of
Female
Population
Net
ProAected
Projected
Estimated
Females
Survival
Expected
Migration
et
Population
Population
1980
1990
1980
Rate
Survivors
Rate
Migration
1990
1988
1
2
3
4
5
6
7
(1)
x
.20
+
(2)
x
(1)
(3)
x
(4)
(3)
+
(5)
(6)
x
.80
0
to
4
x
.9787
x
.0253
x
9,005
9,005
5
to
9
x
.9967
x
.0600
x
8,463
8,463
0
to
4
10
to
14
6,126
.9768
5,984
.1406
SU
6,825
6,685
5
to
9
15
to
19
5,958
.9951
5,929
.1332
790
6,719
6,567
10
to
14
20
to
24
6,211
.9959
6,
186
.1765
1,092
7,278
7,065
15
to
19
25
to
29
8,681
.9937
8,626
.1816
1,566
10,
192
9,890
20
to
24
30
to
34
9,440
.9918
9,363
.
1337
1,252
10,615
10,380
25
to
29
35
to
39
6,638
.9901
6,572
.1302
856
7,428
7,270
30
to
34
40
to
44
5,607
.9870
5,534
.1091
604
6,138
6,032
35
to
39
45
to
49
4,634
.9812
4,547
.0903
411
4,958
4,893
40
to
44
50
to
54
3,953
.9720
3,842
.0777
299
4,
141
4,103
45
to
49
55
to
59
4,024
.9591
3,859
.0696
269
4,
128
4,
107
50
to
54
60
to
64
4,577
.9U5
4,309
.0638
275
4,584
4,583
55
to
59
65
to
69
.
4,
775
.9162
4,375
.0642
281
4,656
4,680
60
to
64
70
to
74
4,091
.8781
3,592
.0508
182
3,774
3,837
65
to
69
75
to
79
4,015
.8203
3,294
.0511
168
3,462
3,572
70
to
74
80
to
84
3,426
.7286
2,496
.0511
128
2,624
2,784
75
to
79
85+
3,003
.5918
2,925
.0511
149
3,074
3,060
80
to
84
1,797
.4203
x
x
x
x
x
85+
1,422
.2759
x
x
x
x
x
Totals
88,378
81,433
9,163
108,064
106,976
(continues)
r3
'
I

Figure
5.12
(continued)
N
0
Panel
B:
Estimate
of
Male
Population
00
5-Year
Net
ProAected
Projected
Estimated
Males
Survival
Expected
Migration
et
Population
Population
1980
1990
1980
Rate
Survivors
Rate
Migration
1990
1988
1
2
3
4
5
6
7
(
1)
x
.
20
+
(2)
x
(1)
(3)
x
(4)
(3)
+
(5)
(6)
x
.80
0
to
4
x
.9725
x
.0253
x
9,341
9,341
5
to
9
x
.9956
x
.0561
x
8,708
8,708
0
to
4
10
to
14
6,494
.9697
6,297
.1437
905
7,202
7,060
5
to
9
15
to
19
6,140
.9921
6,091
.1409
858
6,949
6,787
10
to
14
20
to
24
6,053
.9841
5,957
.1822
1,085
7,042
6,844
15
to
19
25
to
29
8,999
.9783
8,804
.2099
1,848
10,
652
10,321
20
to
24
30
to
34
9,922
.9774
9,698
.1912
1,854
11,552
11,
226
25
to
29
35
to
39
6,718
.9752
6,551
.1620
1,061
7
,612
7,433
30
to
34
40
to
44
5,372
.9669
5,194
.1409
732
5,926
5,815
35
to
39
45
to
49
4,163
.9500
3,955
.1224
484
4,439
4,384
40
to
44
.
50
to
54
3,873
.9243
3,580
.0914
327
3,907
3,900
45
to
49
55
to
59
3,685
.8865
3,267
.0779
254
3,521
3,554
50
to
54
60
to
64
4,250
.8283
3,520
.0723
255
3,775
3,870
55
to
59
65
to
69
4,170
.7626
3,180
.0682
217
3,397
3,552
60
to
64
70
to
74
3,721
.6743
2,509
.0621
156
2,665
2,876
65
to
69
75
to
79
3,
177
.5513
1,
751
.0423
74
1,
825
2,095
70
to
74
80
to
84
2,644
.4214
1,
114
.0423
47
1,161
1,458
75
to
79
85+
1,480
.2950
723
.0423
31
754
899
80
to
84
988
.1707
x
x
x
x
x
85+
564
.2094
x
x
x
x
x
Totals
82,413
72,192
10,
118
100,428
100,
123
(continlll!S)

Figure
5.12
(amtinued)
Age
in
1980
10
to
14
15
to
19
20
to
24
25
to
29
30
to
34
35
to
39
40
to
U
45
to
49
Panel
C:
Estimate
of
Births
Female
Population
1980
6,211
8,681
9,UO
6,638
5,607
4,634
3,953
4,024
Annual
Birth
Number
Rate
of
Births
.00172
11
.07430
6'5
.13759
1,299
.12104
803
.06582
369
.02421
112
.00434
17
.00035
1
Annual
Births
1980
to
1985:
3,257
Total
Births
1980
to
1985:
16,285
Male
Births
1980
to
1985:
8,305
Female
Births
1980to1985:
7,980
Annual
Births
1986
to
1990:
3,663
Total
Births
1986
to
1990:
18,315
Male
Births
1986to1990:
9,341
Female
Births
1986
to
1990:
8,974
~

210
alent to the specificity of the cohorts. If age-, sex-, and ethnicity-
specific cohorts are used, then the birth, death, and mortality data
used must be age-, sex-, and ethnicity-specific. It should also be
evident from this example that cohort procedures not only have
extensive data needs, they also have detailed outputs that are useful
for detailed planning and marketing efforts. Finally, the data in this
example should make it apparent that the most difficult aspect of
using cohort-component procedures is determining the fertility,
mortality, and migration rates that one assumes to apply from the
base date to the estimate date. These are the critical assumptions
that must be examined in the use of the cohort-component method
of population estimation. The limitations of cohort-component
methods are their extensive data requirements and the difficulty of
obtaining such data for the estimation period. Their strengths lie in
their explicit inclusion of the characteristics of populations and
components of population change in the estimation process and the
detailed data they provide.
All of the methods of population estimation noted above have
particular strengths and limitations. In general, the choice of
method must be dictated by the data available for implementing an
estimation procedure, the time available to complete the estimate,
and the nature of the detail needed in the estimate. H estimates for
very short periods since the last census are needed and no character-
istic detail is required, then one may wish to use either an extrapola-
tive technique or some very simple symptomatic technique. If
several years have passed since the last census and data are avail-
able on one or more symptomatic indicators but no data on charac-
teristics are required, it may be advantageous to use symptomatic
methods. If one intends to employ available estimates for larger
geographies to control his/her estimates for smaller component
areas, then ratio-correlation methods may be appropriate. If one
needs age, sex, or other detail in the estimates, some form of cohort-
component technique may be appropriate, if the necessary time and
data are available for its implementation. The choice of the best
estimation method requires a careful balancing of needs and data
availability.
Methods of Population Projection
Population projections involve techniques to determine future
populations. Because they cannot use symptoms or other indirect
indicators of population change (since no such factors exist for future
periods), they are, in some ways, more difficult to complete than

211
population estimates. A wide variety of methods, many of which
are closely related to those used in completing population estimates
are commonly used in population projections. Descriptions of each
of several major categories of methods are presented below (Irwin,
1977; Leistritz and Murdock, 1981; Rives and Serow, 1984; and
Murdock et al., 1987b):
1. Extrapolative, curve-fitting, and
regression-based techniques
2. Ratio-based techniques
3. Land-use techniques
4. Economic-based techniques
5. Cohort-component techniques.
Extrapolative, Curve-Fitting, and Begression-
Based Techniques
The techniques described in this chapter include a wide range of
procedures which attempt to predict the path of future population
growth on the basis of past trends in total population growth.
Included among such techniques are (1) arithmetic, geometric, and
exponential growth rate techniques; (2) curve-fitting techniques
(including polynomial, Gompertz, and logistic curves); and (3) re-
gression-based techniques (linear and nonlinear). Basic to such
techniques is the tendency to project only total population size using
assumed levels, rates, or trends in growth over time. Since these
have been discussed above in the examination of methods of popu-
lation estimation, only those unique aspects of these techniques as
used as methods of population projection are presented here.
The extrapolation techniques discussed above in the examination ·
of methods of population estimation can also be used in projecting
populations. By simply using the rates of growth for past periods as
shown in the formulas for the arithmetic, geometric, and exponen-
tial rates of growth in Chapter 2 and applying them to a base popu-
lation, one can obtain a projection for a future period. For projec-
tions for very short periods of time, these techniques may be quite
useful. For longer periods, particularly for areas that may show
substantial population changes, these techniques are likely to be less
acceptable.
Polynomial growth techniques can also be used. In fact, they
form the basis of many of the curve-fitting techniques. Unlike
arithmetic techniques, they involve patterns or trends which form

212
curves that, when graphed, are nonlinear in form. Whereas a linear
model would be of the form:
Y=a+bx,
a polynomial would include one or more additional terms, such as:
Y = a + bx + cxf. . . . zxn
and would form a curve rather than a straight line when graphed.
The Gompertz and logistic curves described above are examples
of such nonlinear curves. To use any of these curve-based patterns
to project population, one discerns the pattern of change for past
periods, fits the appropriate curve to that pattern, and extends that
curve to find the projected value for a latter period. Such curves
have not been widely used in population projections, particularly for
small areas, because they have seldom been found to be sufficiently
accurate to characterize future population patterns (Ascher, 1978)
and the data required to implement them are often not available.
More commonly used are regression-based techniques in which
the relation5hip of various factors to population growth are known
and used to predict future population levels. As noted above, these
techniques require establishing a set of factors or independent varia-
bles that accurately predict population levels for some past period
and assuming that the past relationships between these predictor or
independent variables and population levels will persist in the
future. Whereas one can use values for symptomatic variables for
the independent variables when such techniques are used for esti-
mating population, no such values are available for future periods;
and so, the values for the independent variables must be projected
before the values of population can be projected. Usually the pro-
jection of independent variable values for future periods is complet-
ed by extrapolating or completing additional regressions on past
historical patterns for the independent variable. The fact that the
use of multiple regression for projecting populations requires projec-
tions of independent variables' values and future values of regres-
sion coefficients substantially restricts the use of regression for
population projections.
Whatever the specific form of the extrapolation techniques used,
the advantages of using such techniques lie in the fact that they use
historical data, which are relatively easy to obtain, and use generally
easy-to-complete computational forms.

213
On the other hand, the dependence on past patterns can also be
a major source of error in projections for rapidly changing areas. In
addition, data needs for some techniques, particularly the projec-
tions of predictor variables needed to determine future populations,
may place considerable data collection demands on the research
analyst. Finally, these techniques seldom provide sufficient detail
on the demographic characteristics (e.g., population of school age)
necessary for some public service and planning needs. For such
methods, then, the data needs can vary from little more than total
population figures for two past censuses (arithmetic or exponential
techniques) to data on multiple variables for past and future time
periods (multiple regression based methods). These methods may
provide adequate short-term projections for past periods and for
populations whose compositions are unlikely to alter rapidly over
time. These methods, however, should be used carefully and with
full knowledge of their limitations.
Ratio-Based Techniques
Ratio techniques consist of procedures in which the population of a
subarea is projected on the basis of its proportion of a larger area's projected
population. In general, ratio techniques are subarea techniques that
are used in conjunction with other projection procedures. They are
frequently used in allocating projected regional or county popula-
tions to municipalities (Murdock et al., 1979a; 1979b).
Although the proportions or ratios of the subareas' populations
to the larger area's population may be assumed to remain constant
over time, it is more common to trend areas' ratios over time and to
adjust the sum of the areas' projections to the projection of the total
area's population (Pittenger, 1976; Murdock et al., 1987b). The
trends in shares are usually determined either by an extrapolation of
baseline patterns or by regression or a similar procedure. When
regression techniques are used with the subarea's share serving as
the dependent variable and the subarea's population attributes as
the independent variables, the technique is basically the ratio-
correlation technique discussed above.
Ratio techniques are most widely used in projecting population ·
for subareas of cities and municipal populations from county or
region totals. Their utility as a major projection technique is clearly

214
limited, but their use in subarea analysis is extensive. The advan-
tages of such techniques lie in their relatively limited data
requirements and simple computational procedures. Potential
disadvantages stem from the need to assume a given ratio or trend
in ratios of subareas to the total area's population over time and
from the lack of demographic detail provided by the outputs of such
procedures.
Figure 5.13 presents a simple example of the use of a ratio-based
technique for projecting the population of Bryan and College Sta-
tion, Texas in the year 2000 given a projection for 2000 for Brazos
County which contains these cities. Three alternative ratios are
examined in the example, but numerous other alternatives for pro-
jecting such ratios are available (see Pittenger, 1976).
Land-Use Techniques
Irwin (1977) delineates two separate types of land-use approach-
es: (1) the saturation approach, in which projected populations for
an area are limited by the number of housing units that can be built
in the area, and (2) density methods, in which limits are placed on
the population in an area on the basis of predetermined levels of
population per unit of area. Both techniques are most often used in
projections of subarea populations in urban areas (Portland State
University, 1975; Greenburg et al., 1978). These techniques, like the
ratio techniques, are seldom used except as part of more compre-
hensive procedures.
The saturation method is usually employed by assuming a
standard number and type of housing units per unit of area and
then computing population on the basis of an average number of
persons per unit. Among the problems with this method are the
determination of the upper limit for housing units per unit of area
and the need to obtain projections of average household size for
future periods.
The density method may be particularly useful for projecting
subarea populations within urban areas undergoing rapid growth.
In such areas, extrapolation of past trends may quickly lead to
unreasonable population levels. Controls of subareas' populations
to the total population are essential.
An example of this technique using a hypothetical city and a
density limit of 1,000 persons per square mile is shown in Figure
5.14. It demonstrates only one very simple means of controlling for
density and reallocating population to less dense areas. The use of

Figure
5.13:
Example
of
a
Ratio-Based
Technique
To
project
the
popu1ation
of
the
dttes
of
College
Station
and
Bryan
for
2000,
given
a
projection
for
Brazos
County
for
2000
Brazos
County
Bryan
College
Station
Remainder
Bryan
College
Station
Remainder
1980
Population
93,588
U,337
37,272
11,979
~
of
County
Population
-
47.4
39.8
12.8
1990
Population
121,862
55,002
52,456
U,404
~
of
County
Population
-
45.1
43.1
11.
8
2000
Projected
Population
lU,693
Projection!
Ysing:
Same
Proportions
Arithmetic
Trend
Exponential
Trend
as
in
1990
in
Share
in
Share
65,257
(45.
a)
62,218
(43.0~)
61,929
(42.U)
62,363
(43.a)
67,138
(46.U)
67,572
(46.7~)
17,073
(11.U)
15,337
(10.
0)
15,192
(10.5~)
~

216
Figure 5.14: Example of a Land-Use Technique
Given 1980 and 1990 Census populations and 2000 projected populations for five
census areas in Anyville Oty:
Area
1
2
3
4
5
Population
1980 1990
3,000
4,000
4,000
3,000
6,000
4,000
3,000
7,000
4,000
7,000
~ Change
1980-90
+33.3
-25.0
+75.0
+33.3
+16.7
Projected
2000
Population
5,332
2,250
12,250
5,332
8,169
Area in
Sq. Mi Jes
10
12
8
9
15
Hyou determine that density cannot exceed 1,000 persons per square mile, area 3
has too many people. You need to redistribute the population (4,250) that cannot
locate in area 3. There is no fixed procedure, but one could simply redistribute the
excess area 3 population proportionately.
For example:
Area
1
2
3
4
5
~ of
Remainder
of Popu I at ion
(not in Area 3)
in 2000
in Each Area
25.3
10.7
25.3
38.7
Di s t r i but ion
of Area 3
Population
Excess (4,250)
1,075
455
1,075
1,645
New
Population
Projection by
Census Area
6,407
2,705
8,000
6,407
9 ,814

217
more sophisticated techniques are described in the available litera-
ture (e.g., see Pittenger, 1976).
The advantages of these methods are clearly their utility in limit-
ing the rate of growth in component areas to feasible levels, while
their disadvantages lie in the difficulty encountered in determining
the density limits for housing units for an area and in the lack of
demographic detail produced by such procedures. Particularly for
rural areas that are not geographically confined, growth limits may
be extremely difficult to determine. On the other hand, in some
rural areas, topographic features or land ownership by federal or
state governments may limit the potential geographical expansion of
a jurisdiction (city). In such cases, land-use models may be appli-
cable.
Economic-Based Techniques
Economic-based techniques, as the name implies, project population on
the basis of assumed relationships between economic patterns and population
change. As the name also suggests, they tend to be the population
projection techniques most widely used by economists. They have
been widely used in the OBERS's (U.S. Bureau of Economic Analy-
sis, 1991) national and economic models. Their use is particularly
attractive when population growth in an area is expected to result
largely from economic development.
The basic methodology of such projections involves using an economic
model to determine employment change and then, either a direct or an
indirect method to determine either total population change or the level of
change within a key demographic component (usually migration) resulting
from the projected employment demand. In the simplest procedure,
projections of population are determined by applying a population-
to-employment ratio to the projection of employment. This
technique, however, relies on some very simple assumptions. In
particular, the assumption of a constant number of persons per
employee is often questionable because of wide variation in depend-
ency rates among populations in different areas. Thus, the use of
this simple application of the technique is of decreasing significance
as a means of population projection.
A more widely used procedure is to match the economic projec-
tions of labor demand with projections of labor supply to determine
migration levels. In this mode of use, an economic-based technique
is usually used to project labor demand in conjunction with a cohort
technique (cohort-survival technique) to project all but the migration

218
component of population change. Labor supply is usually deter-
mined by applying total, age, or age- and sex-specific labor force
participation rates to population projections (total, age, or age- and
sex-specific populations). Labor demand is then matched with labor
supply to determine migrating workers. If the labor supply exceeds
demand, workers are assumed to outmigrate. If demand exceeds
the labor supply, then workers are assumed to inmigrate. Inmigrat-
ing or outmigrating workers are then converted to population
estimates by the application of various assumed demographic char-
acteristics for migrating workers. In specific applications, however,
a detailed set of procedures and extensive sets of assumptions are
required for each of the following major steps:
1. Projecting labor demands over time
2. Projecting labor supplies over time
3. Matching labor supplies and labor demands
4. Determining levels of migration
5. Projecting the total population changes
accompanying the migration of labor
Each of these steps for standard models is briefly reviewed below.
The projection of labor demands is usually done using an input-
output model, an export base model, or some form of shift-share
analysis to project labor demands resulting from economic activity.
The projection of labor supplies usually involves the projection
of at least two major dimensions: (1) a baseline or closed popula-
tion to serve as the base to which estimators of employment supply
must be applied and (2) the expected levels of labor force participa-
tion of persons in the closed population during the projection peri-
od. The baseline population is often simply the last count of
persons adjusted for mortality and fertility changes since that count.
The levels of labor-force participation assumed for the projection
period are the key part of this technique, and if they are in error,
the level of migration predicted will be in error.
In general, the participation rates assumed to prevail are allowed
to vary over time. For local areas, these trends over time are often
linked to national projections of labor-force participation rates pub-
lished by the U.S. Bureau of the Census and the Bureau of Labor
Statistics. This patterning of local to national rates may be done by
calculating a ratio of local to national rates at a known (census)
period and then assuming this ratio will be maintained over time or
by altering the ratio in a prescribed manner over time. The fixed or

219
projected ratios for each period are then applied to the projections of
national participation rates to obtain local rates for use in projecting
employed population. This technique can be used with total popu-
lation labor-force participation rates or can be made characteristic-
specific (e.g., age-specific, age- and sex-specific, or age-, sex-, and
race/ethnicity- specific). Whatever technique is used, the participa-
tion rate, when applied to the baseline or closed population value,
becomes the major determinant of labor supply. This supply is
usually further adjusted by the local level of unemployment or
underemployment before being matched with labor demand.
The matching of labor demands and labor supplies may involve
relatively simple or highly complex procedures. That is, both labor
demands and supplies may involve one type of demand and supply
or several. In a procedure developed by Hertsgaard et al., (1978)
and Murdock et al., (1979b; 1984; 1987a), for example, at least four
separate types of demand and supply are used, and supplies are
examined with age and sex detail. Whatever the level of complex-
ity, however, the key assumption is that demands that cannot be
met by the local population will be met by inmigration, while excess
supplies will lead to outmigration. Research in economics and
demography points to a general relationship between employment
and migration (Sjaastad, 1962; Long, 1988), but there is some evi-
dence that this relationship is weaker and less pervasive than at
previous periods, and that employment changes are more directly
related to inmigration than to outmigration (Greenwood, 1985;
Ritchey, 1976; Long, 1988).
In recognition of the fact that not all migration behavior is
economically motivated, the level of migration resulting from the
matching of labor supply and demand is often altered by incorporat-
ing noneconomic procedures or by adjusting the basic matching or
interfacing procedure. For example, the OBERS's projections main-
tain three separate population groups: (1) those under 15 years of
age; (2) those 15-64 years of age; and (3) those 65 years of age and
over. Only the under 15 and 15-64 age groups' levels of migration
are determined by the employment matching procedure. The age
group 65 years of age and older is projected largely on the basis of
past trends with little regard being given to area employment pat-
terns. In other procedures, some populations, such as those at
military installations, colleges, and universities are treated in spe-
cial population• procedures and are exempt from employment
matching routines. Finally, some techniques have been developed

220
which allow the labor supply in an area to exceed labor demands or
demands to exceed supplies by predetermined rates before outmi-
gration or inmigration occurs (Hertsgaard et al., 1978; Murdock et
al., 1979a; 1979b; 1987a). In sum, then, the step of determining the
level of migration resulting from labor market changes has come to
use techniques that are increasingly complex and sensitive to differ-
ences in demographic composition.
Given that the matching procedure has been completed and the
number of migrating (in or outmigrating) workers determined, the
last step is to convert projections of migrating workers into projec-
tions of population. This usually involves applying a set of assumed
worker-related population characteristics to the projections of the
number of migrating worlcers. Though simple computationally, the
characteristics assumed for worlcers (such as family size, dependent
characteristics, etc.) will markedly affect the levels of population
projected. As with the use of data on average size of household or
other characteristics, the characteristics assumed for migrating work-
ers must be carefully determined.
Figure 5.15 presents a relatively simple example of the use of an
economic-based population projection model. As this example
suggests, population growth is assumed to be largely a function of
economic growth.
Economic-based techniques are relatively widely used (American
Statistical Association, 1977; Murdock et al., 1984). Their
advantages lie in the fact that, unlike many purely demographic
techniques, economic-based techniques allow the economic changes
expected to take place in an area to be taken into account. They
represent important attempts to integrate demographic and
economic factors that are clearly interrelated.
Their weaknesses must also be recognized. The number of
explicit assumptions on which such projections are based is large.
Accurate projections of both economic and demographic factors and
their interrelationships are required by such techniques. Since the
errors made in assumptions for basic factors at the beginning of such
computational procedures may be magnified as the computations
proceed (Alonso, 1968; Leistritz and Murdock, 1981), such a large
number of sequentially linked assumptions may be problematic. In
addition, the data requirements of such models are often extensive.
Data on economic and demographic trends such as labor force partic-
ipation rates, family size, and many other dimensions must be
obtained for the projection period. Finally, because they have been
developed relatively recently, these techniques have received even
less validation than other procedures, and it is unclear whether such

Figure 5.15: Hypothetical Example of a Simple F.conomic-Based Population
Projection Method
221
Given projected business acttvity of $19,511,775,000 and projected productivity per
worker of $3.5,440 in 2000 from an economic model (input-output, export-base, etc.),
project the population in Bexar County, Texas in 2000.
Steps in Projection Process
1. Project Jabor demand. If business activity Is $19,511,775,000 and productivity is
$3.5,440, then Jabor demand can be determined by dividing business activity by
productivity.
Business activity - $19,511,775,000
Productivity - $3.5,440 per wo:rker
Number of employees required - $19,511,775,000/$3.5,440 • 550,558
(Labor force demand)
2. Project Jabor supplies in 2000 given cohort survival projection of popuJatton in
2000. This is the popuJatlon for the preceding time period as altered by births
and deaths occurring between the preceding time period and 2000. The effects
of mtgratlon-reJated change have not been taken into account in the cohort-sur-
vived population. The Jabor supply Is projected by multiplying Jabor force partic-
ipation rates by the cohort-survived populations of the appropriate ages.
Bexar County Population by Age, Labor Force
Participation Rates by Age, and Projected
Labor Supply Age in 2000
Labor Force Projected
Cohort-Survived Part i c i pat ion Labor
Age Po2ulation Raiu Su22lx
0-4 124,825
5-9 120,516
10-U 105,266
15-19 105,629 40.U 42,463
20-24 101,638 64.U 65,861
25-29 121,998 69.7~ 85,033
30-34 121,388 69.7~ 84,607
35-39 90,034 69.7~ 62,754
40-44 83,429 69.7~ 58,150
45-49 62,540 69.7~ 43,590
50-54 50,030 69.7~ 34,871
55-59 46,171 50.2~ 23,178
60-64 43,405 50.U 21,789
65-69 41, 191 13.H 5,355
70+ 78,952 13.0~ 10,264
Total 1,297,012 537,915
(amtinutS)

222
Figure 5.15 (rontinued)
3. Match labor demand and supply and determine difference between them.
Labor demand - 550,558
Labor supply - 537,915
Difference - 12,643
4. Determine labor force migration. If one assumes that the difference between
labor demand and supply are those who must migrate, 12,643 workers would
migrate to Bexar County to take jobs In 2000.
5. Determine total population migration associated with labor force migration. If
the number of lnmigratlng workers equals 12,643 and each worker has a house-
hold size of 2.2 (Including the worker) the total number of lnmigratlng persons
would be 27,815.
6. Determine the projection of the total population for Bexar County In 2000. This
would be equal to the sum of the cohort-survived population or 1,297,012 and
the lnmigratlng population, 27,815; thus the projected total population for Bexar
County for 2000 would be 1,324,827.
In order for additional iterations of the model to be completed to obtain projections for
subsequent time periods, characteristics of new lnmigratlng persons (i.e., the 27,815)
would have to be assumed or determined. Thus, an assumed age distribution of the
lnmigratlng population would be necessary In order to obtain the lnmigratlng popula-
tion by age to merge with the base population to produce the new base population for
subsequent iterations of the cohort-survival model.

223
techniques provide more or less accurate population projections than
demographic techniques alone (Kendall, 1977; Murdock et al., 1984).
However, economic-based techniques represent an important set of
methods that are worthy of consideration, particularly in areas
where population change is closely tied to economic growth.
Cohort-Component Techniques
Cohort-component projection techniques are perhaps the most
widely used techniques for determining future population levels.
They are often seen as the most complex and sophisticated of the
purely demographic techniques and are usually preferred by profes-
sional demographers because they involve the direct simulation of
the demographic processes of fertility, mortality, and migration that
produce changes in population size.
As the name implies, the basic characteristics of these techniques are the
use of separate cohorts, persons with one or more common characteristics,
usually similar ages (i.e., persons born during the same period), and the
separate projection of each of the major components of population
change--fertility, mortality, and migration--for each of these cohorts.
These projections of components for each cohort are then combined
in the familiar •demographic bookkeeping equation• noted in
Chapters 2 and 4 (Barclay, 1958; Bogue, 1974; Murdock et al.,
1987b):
Where: Pt
2
pt2 = pt1 + Bt1 - t2- Dt1 - t2 +Mil - t2
• the population projected at some future date
t years hence
• the population at the base year tl
• the number of births that occur during the
interval tl - t2
Dli _l:z • the number of deaths that occur during the
interval tl - t2
M; _l:z • the amount of net migration that takes
·place during the interval tl - t2

224
When several cohorts are used, Pt may be seen as:
2
n
PL = l: Pc. t
-2 i .. 1 1' 2
Where: Pt is as in the equation above
2
Pt = population of a given cohort at time t2,
2
Where: all terms are specific to given cohorts q.
In general, single-year or five-year age and sex cohorts are used
in conjunction with age- and sex-specific survival rates, fertility
rates, and migration rates. The technique is seldom used for geo-
graphic areas smaller than counties because of the difficulty of
obtaining birth, death, and migration data for smaller areas and
because of the widely known problems of applying rates (or per-
centages) to small population bases (Irwin, 1977; Murdock et al.,
1987b). Whatever the geographical level of analysis, however, this
procedure can be seen as having four basic steps:
1. The selection of a baseline set of cohorts for the area of
study.
2. The determination of appropriate baseline migration,
mortality, and fertility measures for each cohort for the
baseline period.
3. The determination of the method for projecting trends
in fertility, mortality, and migration rates over the
projection period.
4. The selection of a computational procedure for apply-
ing the rates (from 3.) to the cohorts over the projec-
tion period.
Each of these steps involves consideration of numerous alternatives
that are discussed briefly below.

225
Selection of Baseline Cohorts. The selection of baseline cohorts
is usually done by selecting data from the last population census.
The data so selected are usually age- and sex- or age-, sex-, and
race/ethnicity-specific cohorts in single- or five-year age groups. Of
all the data requirements, the baseline cohort data required for the
procedure are the most readily available. In addition, the major
adjustments to such baseline data that may be necessary (in addition
to those noted below) are relatively simple, such as the adjustment
of cohorts to appropriate age groupings (Pittenger, 1976; Irwin,
1977; Haub, 1987).
Determination of Appropriate Baseline Measures. The selection
of the appropriate migration, fertility, and mortality rates to be used
in the projection is the key step in the projection process. The
accuracy of the assumptions about these rates and their trends over
time will determine the accuracy of the projections. The selection of
these rates involves numerous considerations.
Determination of Mortality Rates. Mortality levels can generally be
readily determined because of the availability of data on mortality
and the relatively slow rate of change in mortality levels over time
(at least in developed areas of the world). Life tables for states and
other areas are published periodically (for example, see National
Center for Health Statistics, 1975; 1986), and generally, state-level
rates can be assumed to be applicable to local areas without marked-
ly affecting the accuracy of the analysis. Given a life table, the
mortality measure most often used in projections is the age- and
sex-specific survival rate which indicates the probability of persons
of a given age and sex living from period (x) to the next period (x +
t). So considered, the survival rates for any age group can be
computed from the nLx column of the life table (as noted in
Chapter4).
An alternative to life table derived survival rates are national
census survival rates, computed from census data at the national
level. National-level data are used to control for the confounding of
mortality and migration factors. Thus, when national data are used,
the effect of migration on age groups can be assumed to be negligi-
ble because immigration is a very small percentage of total national
population change. To compute these rates, age groups at two
consecutive censuses are examined in the following computational
form:

226
Where:
Px+ t
Sx,x+t =--
Px
Sx X + t = is the survival rate from x to x + t
I
Px + t = population in a given cohort at the
second census period
Px =population in a given cohort at the
first census period
The problem with this method is that the national rates computed
are less likely than rates derived from state life tables to reflect local
conditions and hence these rates are generally used only when
appropriate life tables are not available. One additional advantage
of census survival rates is that because one is comparing counts for
two censuses at the national level, values are largely unaffected by
errors of closure (i.e., the extent of undercount or overcount).
Determination of Fertility Rates. The methods for determining
fertility levels fall into three general categories: (1) period fertility
measures; (2) cohort fertility measures; and (3) marriage-parity-
interval progression measures (Shryock and Siegel, 1980).
Period fertility measures are among the most often used meas-
ures of fertility in projections. They involve the use of rates show-
ing the number of births likely to occur to a group of women during
the projection period. The rates used most often are general fertility
rates and age-specific fertility rates computed in the manner noted
in Chapter 4. The distinctive characteristics of these rates are that
they are rates computed at a given point in time. They do not take
into account the fact that the time period covered by a set of projec-
tions will involve the fertility experiences of women as they age
over the projection period. Rather, these period measures are based
on the experiences of women of different ages at one point in time.
Cohort fertility measures attempt to overcome the limitations in
period measures by attempting to simulate a set of rates that will
characterize the actual experiences of a cohort of women as they age
through the life cycle. The most widely used form for simulating
these experiences is to choose a set of age-specific fertility rates that
would result in the average female giving birth to a given number of
children by the completion of her reproductive years. Among the
targeted values often chosen is the Total Fertility Rate of 2.1 births
per female. This replacement level of fertility, as noted in Chapter

227
4, is that number of births necessary for the women in a population
(with levels of mortality similar to those in the United States) to
replace themselves and their mates, taking into account (the .1) that
some children will die prior to reaching reproductive ages.
The advantage of using total fertility rates is that they allow
analysis in terms of family size and other similar concepts that are
familiar to a wide range of persons who may use the projections.
Although they are often based on the experiences of actual cohorts
of women who have completed their childbearing years, the obvious
disadvantage of using these rates is the difficulty encountered in
choosing the set of rates that will correctly characterize the experi-
ences of future cohorts of women.
Marriage-parity-interval progression measures refer to the use of
sets of sequential probabiUty measures that take into account the
probability that women with different marital statuses and complet-
ed family sizes will give birth to another child during the projection
period. Although this and similar techniques may be more widely
used in the future (Pittenger, 1976) and have been used in some of
the recent U.S. Bureau of the Census projections (U.S. Bureau of the
Census, 1979; Spencer, 1986; 1989), it is a relatively complex proce-
dure with extensive data requirements (numbers of women by
marital status, age, and parity, births by parity, etc.). Marriage-
parity-interval progression procedures have received relatively little
use in small-area projections and will not be discussed in further
detail here. They are, however, worthy of further examination (see
Shryock and Siegel, 1980: 789-90), and their use may become more
prevalent as the availability of detailed local area data increases.
In whatever manner fertility rates are determined, the goal at
the end of this step of the cohort-component procedure is to have
determined a set of fertility rates for each female cohort that can be
used to determine the number of births likely to occur during the
projection period. These procedures, then, are ones aimed at pro-
viding the Bt1- t2 function in the bookkeeping equation.
Determination of Migration Rates. Migration is the most difficult
demographic process to predict and the most difficult on which to
obtain current data when cohort-specific values are required. The
difficulty is further increased by the fact that migration may involve
two different forms with opposite effects on population change.
These forms are inmigration and outmigration. Any time an area
changes from a predominance of one of these patterns to the other
(thus changing from a positive to a negative or from a negative to a
positive value), the increased potential for error in the projections is

228
evident. Assumptions regarding migration are usually the major
area of contention in population projections.
Methods for projecting migration fall into two broad categories:
(1) net migration projection procedures and (2) gross migration
procedures. Whereas net migration procedures attempt only to
discern the net difference between the levels of in- and outmigration
in an area, gross migration procedures project inmigration and
outmigration separately.
Net migration procedures usually involve determining migration
using residual methods. The formula for the residual method of
migration was provided in Chapter 4. The advantage of the use of
residual methods is that these methods are ones for which it is rela-
tively easy to obtain the necessary data. The disadvantage of these
methods is that net migration is only a statistical difference between
the two processes of in- and out-migration and, as such, can result
from highly different patterns of activity in an area (for example, an
area that had inmigration of 50,100 and outmigration of 50,000 and
one which had inmigration of 200 and outmigration of 100 would
both have net migration of 100, but the level of mobility in the areas
would be very different). Thus, net migration may not accurately
characterize migration behavior in a projection area.
Gross migration measures are used less often but are more at-
tractive conceptually because they simulate the behavior of actual
individuals. The difficulty with the use of gross migration measures
is that the necessary data to determine them are often not available
for local areas or, when made available, are likely to be extremely
dated (U.S. Bureau of the Census, 1977; 1984). When appropriate
data are available, the projection involves projecting outmigration
for each area, usually on the basis of past patterns, and then, pro-
jecting the pool of outmigrants as inmigrants to each area on the
basis of past trends in the ratio of inmigrants in the local area to
total inmigrants in the pool (Shryock and Siegel, 1980; Irwin, 1977).
Although additional procedures for projecting local area migra-
tion levels have been suggested (Pittenger, 1974; Murdock et al.,
1987b), those discussed here are the main procedures presently in
use. Each method places a heavy reliance on the use of assumptions
based on past patterns. Unlike mortality or fertility patterns where
some theoretical limits can be set, the range of possible values for
migration is indeterminant and the reasons for changes in direction
from net in- to outmigration or net out- to inmigration are not
adequately understood.
Perhaps the most ambitious and theoretically complete means
for projecting migration have been those methods developed by

229
Rogers and others (Land and Rogers, 1982). In their most de-
veloped form, these multi-dimensional, multi-state models attempt
to simulate gross migration flows among all areas on a cohort-by-
cohort bases. Using matrices of probabilities of inmigration and
outmigration for each cohort for each combination of areas, these
techniques have been used to examine interstate migration and
selected flows among counties. They promise to be extended to
ever smaller levels of geography over time. At present, they are
seldom used for small-area projections because of their extensive
data requirements and the fact that such data {e.g., on flows of
persons in and out of all areas from and to all other areas) are
simply not available for small population areas in the United States.
Methods for Projecting Rates Over Time. Given that a baseline
set of mortality, fertility, and migration rates has been established,
the third major step involves developing procedures for projecting
the trends in these rates over time. There are three widely used
procedures: (1) continuation of baseline rates; (2) use of targeted
rates; and (3) trending of local area rates to regional, national, or
other standard• area's rates.
Continuation of rates determined for the baseline period may be
preferable in many instances, particularly if the area is large and is
not changing rapidly and the projection is for only a short period in
the future. For long-term projections, however, and particularly for
areas undergoing rapid development, such assumptions are seldom
warranted. Increasingly, then, projections using continuations of
past trends are being questioned and used only when projections
based on alternative assumptions are also used.
The use of targeted rates for specific periods or targeted levels of
change in rates over specific periods are more frequently employed.
In using these procedures, baseline rates are assumed to reach
predetermined rates by certain points in the projection period. The
U.S. Bureau of the Census has historically used rates of fertility that
are trended over time to reach a given level (1.6, 1.8, 2.1, 2.5, etc.
levels of 1FRs) by a specific year (U.S. Bureau of the Census, 1979;
Spencer 1984; 1986; 1989) and has also often used targeted rates for
migration--such as assuming immigration will be negligible by a
certain point in time (U.S. Bureau of the Census, 1977; Spencer,
1984).
The choice of rates using this procedure is usually tied to a
conceptual perspective on population, such as stable population
theory (a stable population being one with a fixed level of births and
deaths per year), or to assumptions that local area rates will con-

230
verge toward those of a larger area, such as the state or nation. The
rates chosen, then, are the targeted levels that will result in a given
stable population or that characterize a large area to which local area
rates will converge.
As should be evident, this procedure is also dependent on a
number of assumptions and requires the analyst to make projections
of long-term trends in each of the vital rates and assumptions about
the time period necessary for an area to reach a given level of fertili-
ty, migration, and mortality. The task involved is a difficult one.
The third approach is similar to the other two in that it involves
choosing a standard area after which local rates can be patterned.
However, its widespread use requires that it be given special
emphasis in this discussion.
In this third approach, the trending of local rates to a larger
area's rates, the analyst (1) selects a standard population to which to
relate the local area; (2) determines the ratio or relationship of the
local area's rates to the standard population's rates at a given point
in time; and (3) assumes that the local area's rates will either main-
tain a constant ratio or relationship to the standard population's
rates or change in a fixed manner over the projection period. Using
this procedure, the analyst can make widespread use of projections
made by various agencies and groups. The work of the U.S. Census
Bureau on projecting long-term national trends in fertility, mortality,
and migration have been used as the standard in many local area
projections (Tarver and Black, 1966; Murdock and Ostenson, 1976;
Hertsgaard et al., 1978; Murdock et al., 1987a). As with the first
two procedures discussed, the utility of this technique is heavily
dependent on the correctness of the analyst's assumptions about
projected long-term trends in vital rates in the standard• area and
about the comparability of local rates to those for other areas. It
shares the disadvantages and limitations of the other techniques, but
provides the researcher with the possibility of using the work of
analysts from agencies whose long-term projections and data bases
may be superior to his or her own.
Each of these three techniques for projecting trends in rates over
time requires the use of assumptions that are often quite heroic in
nature, given demographers' present abilities to predict mortality,
fertility, and migration phenomena. However these trends are
projected over time, the applied demographer should be the first to
view his or her assumptions with skepticism and should make the
speculative characteristics of such assumptions clear to potential
users of his/her projections.

231
Selection of Computational Procedures. Although all cohort-
component procedures compute their final population values on the
basis of the general summation procedure implied by the population
equation, several aspects of these procedures require brief considera-
tion. It should be made evident, for example, that few analysts
using the cohort-component technique feel confident enough of their
assumptions about vital rates to suggest that a single set of assump-
tions will be correct for all areas and periods. As a result, cohort-
component projections will generally involve making several sets of
alternative computations with different assumed rates resulting in
several alternative projection series.
A number of other considerations must also be addressed.
These considerations relate to adjustments required during the
computations and may be most efficiently examined by presenting a
standard set of steps used for deriving the values denoted in the
population equation. Although a number of analysts provide step-
by-step instructions for doing cohort-component procedures
(Murdock et al., 1987b; Irwin, 1977; Morrison, 1971; Pittenger, 1976;
Barclay, 1958; Tarver and Black, 1966; Shryock and Siegel, 1980;
Bogue, 1974), the general steps delineated below appear to be the
most useful for purposes of this discussion:
1. Adjust the baseline population cohorts for the correct
time periods and spatial referents.
2. Adjust rates of migration, fertility, and mortality
making sure that all rates are
a. based on consistent population bases;
b. adjusted to consistent time, place, and cohort
factors; and
c. specific to the characteristic detail desired in the
projections.
3. Survive baseline cohorts to the end of the projection
period to obtain their expected• populations.
4. Compute migration by applying cohort-specific migra-
tion rates to the appropriate expected population for
the projection date.
5. Compute births and add births to the initial cohorts of
the appropriate base population (if sex-specific cohorts
are used, births are allocated to sex groups in accord-
ance with sex ratios at birth).
6. Sum components for cohorts as desired to obtain the
total population or the population of subgroups.

232
7. Control the sum of populations for subareas to the
population total for the larger area.
Each of these steps entails adjustments that are briefly delineated
below.
In step one, it is essential to ensure that all data are made con-
sistent in terms of time and place referents. That is, all population
values should be adjusted for similar time frames. Population
censuses, for example, are for populations as of April 1 of the
census years. These figures should either be adjusted to be consist-
ent with the periods for which other data are available, such as
calendar years, or other data should be adjusted to be applicable to
April 1 of the year. Whatever geographical unit has been chosen for
analysis, all data must be adjusted to that unit by appropriate alloca-
tions or other procedures. It is particularly important to make sure
that constant boundaries are assumed across time and have been
taken into account in any historic data used. Special attention
should be given to such factors in urban areas where boundary
changes are frequent.
It is also essential in this initial step to consider what provisions,
if any, should be made for •special populations:• As noted above,
these are populations that are unlikely to be exposed to the same set
of demographic processes as the remainder of the population and
include such groups as college and university populations, military
base populations, and institutional populations. In general, such
populations are treated in one of two ways.
One commonly used procedure is simply to exclude them from
the cohort-component procedure and separately project their total
size for each projection date. For special populations in which the
population totals vary little from period to period, the age distribu-
tions are concentrated, and integration with the rest of the popula-
tion is limited (such as military bases and college populations), this
may be an adequate way to project the influence of such groups.
For other groups, their distinct demographic rates may be such and
their distributions across age groups extensive enough to merit a
second procedure--the development of separate fertility, mortality,
and migration rates and the use of separate cohort procedures. In
any case, it is in this initial step of determining baseline cohorts that
special populations must be designated.
Step two notes that the rates for each component must also be
adjusted. These adjustments include not only the same time and
place adjustments as for total population bases, but also those for.
cohorts. Whatever the level of detail-age, sex, ethnicity, etc.--for

233
which projections are desired, appropriate rates must be developed
for each detailed characteristic. Rates must also be made consistent
with the period of the projection and the size of cohorts. That is, if
the projection interval is one year and single-year cohorts are to be
used, rates must be single-year, not five- or ten-year rates and must
be for single years of age. Pittenger (1976), among others, provides
readily usable formulas for preparing adjustments of rates to appro-
priate periods, and Irwin (1977) provides excellent examples of
adjusting cohorts to be temporally and areally specific.
Similar concerns also relate to steps three, four, and five. One
such concern is the need to adjust the baseline populations to which
projected rates are to be applied. Some projection analysts (Pitteng-
er, 1976) recommend applying survival rates to the base period
population (e.g., 1990 in a projection beginning with 1990 data) and
then using the survived population as the base for fertility and
migration computations (for example, this procedure is used in the
example shown in Figure 5.16). Other analysts (Irwin, 1977; Tarver
and Black, 1966) recommend the use of an average number of
persons at risk. For example, if a five-year projection cycle is being
used with five-year age and sex cohorts, parts of at least two differ-
ent cohorts will be involved in each projection cycle. For example,
if the age·group of males 15-19 in 1990 is to be projected to 1995, the
five-year rates should be applied to an average number of persons
who are 15-19 during the 1990-1995 period. This will, in fact, in-
clude different parts of different cohorts being exposed to rates for
15-19 year olds for different lengths of time. Fifteen year olds in
1990 will be exposed to the rates for all five years (1990-1995), but 16
year olds will experience such rates for only four years, and 17 year
olds for only three years, etc. On the other hand, those 14 years old
in 1990 will experience the 15-19 year-old rates for four years, while
those 13 will experience these rates for three years, etc. To adjust
these cohorts, an adjacent cohort technique (Irwin, 1977) is neces-
sary in which the average of the two cohorts is used as the base for
projections. These adjustments should be made for all cohorts
before component rates are applied. Secondly, it should be noted in
step five that the births produced by adjusted sets of female cohorts
must be allocated to each initial sex distribution. This is usually
done by taking data on sex ratios at birth, available from state vital
statistics and health departments, and applying them to the total
number of births.
Finally, step seven points to the need to ensure that, if relatively
large areas with multiple subareas are to be projected, some attempt
to control the sum of local area totals to the total of the larger area

234
be made. If this is not done, the summation of subarea migrants or
births may exceed those that are reasonable for the larger area (see
Irwin, 1977; Murdock et al., 1987b for a discussion of this problem).
Although the adjustments noted in the seven computational
steps are all relatively minor, their omission can lead to serious
errors in computations. The computations as well as the assump-
tions underlying the cohort component procedure may, therefore, be
quite complex: Fortunately, however, a number of readily available
computer programs for performing such projections are available
(Bogue, 1974; U.S. Bureau of the Census, 1976, Strong, 1987;
McGirr and Rutstein, 1987).
Figure 5.16 presents an example of the use of a cohort-
component procedure to project the population of Harris County,
Texas to the year 2000. Although this is a comparatively simple
example, it demonstrates the detailed computations required to use
component projection procedures.
Cohort-component procedures are among the most developed
techniques available for population projections. The advantages in
the use of cohort-component procedures are that their use allows
demographic processes to be simulated and age, sex, and other
detail to be provided in the outputs from such procedures. The
disadvantages are equally evident. The data requirements are
extensive and a relatively large number of assumptions must be
made about each of the major components. The utility of such
procedures in making projections is dependent on their judicious
use and further development of our understanding of the determi-
nants of basic demographic processes.
Estimates and Projections of Population-Based
Statuses and Characteristics
Population estimates and projections are often used as the basis
for estimating and projecting other factors that are affected by the
size, distribution, and composition of populations. Among such
estimates and projections are those of the labor force; .school enroll-
ment (elementary and secondary as well as higher education);
households and households by tenure; incidences of various diseases
and other health-related conditions; demand for specific types of
goods and services (as measured by the number of persons with
certain use-related demographic characteristics); and numerous other
dimensions (Murdock et al., 1989a). Although such estimates and

Figure
5.16:
Steps
In
and
Example
of
the
Use
of
the
Cohort-Component
Method
to
Project
the
Poiulation
of
Harris
County,
Texas
by
Hve-Year
Cohorts
from
1990
to
2000
Assuming
1980
Age-Sex
Specific
Fertility
Rates
and
Age-Sex
Specific
Survival
Rates
and
1970-1980
Age-Specific
Net
Migration
Rates
1.
Record
base
population
age
groups
for
the
base
year
(In
this
example,
the
Harris
County,
Texas
population
by
five-year
age
groups
for
1990
(see
Column
1,
Panel
A)
and
for
the
projection
year
(see
Column
2,
Panel
A).
2.
Record
number
of
persons
by
age
for
the
base
year
(Column
3,
Panel
A).
The
first
entry
value
is
those
persons
born
between
the
base
year
and
the
projection
year
(see
Panel
B).
3.
Record
appropriate
age-specific
survival
rates
(Column
4,
Panel
A)
and
age-specific
net
migration
rates
(Column
6,
Panel
A).
4.
Determine
the
number
of
births
between
the
base
year
and
the
projectlon
year
as
shown
In
Panel
B:
a.
Recording
appropriate
female
age
groups
for
the
base
year
(Column
1,
Panel
B)
and
the
projectlon
year
(Column
2,
Panel
B)
and
the
number
of
females
by
age
(Column
3,
Panel
B).
Note
that
the
age
groups
of
females
shown
Include
not
only
those
In
the
childbearing
ages
15-44,
but
also
those
who
will
enter
the
childbearing
ages
from
1990-2000.
b.
Record
age-spedftc
survival
rates
(Column
4,
Panel
B)
and
birth
rates
(Column
6,
Panel
B)
for
females
and
age-specific
net
migration
rates
(Column
8,
Panel
B).
Note
that
only
age-spedflc
migration
rates
are
used
In
this
example,
but
age-sex-
specific
rates
would
be
preferable
In
order
to
maintain
consistency
and
Increase
precision.
c.
Survive
the
female
population
from
the
base
year
to
the
projectlon
year
by
multiplying
the
survival
rates
(Column
4,
Panel
B)
by
the
number
of
females
In
each
corresponding
age
group
(Column
3,
Panel
B)
to
obtain
the
expected
population
of
females
In
the
projection
year
(Column
5,
Panel
B).
Note
that
the
births
between
the
base
and
projection
years
(as
deter-
mined
In
d.
below)
are
also
survived.
d.
Determine
the
number
of
births
to
females
between
the
base
year
and
the
projection
year
(Column
7,
Panel
B)
by
multiply-
ing
the
age-speclftc
birth
rates
(Column
6,
Panel
B)
by
the
survived
females
In
the
corresponding
age
groups
(Column
5,
Panel
B).
As
shown
at
the
bottom
of
Panel
B,
because
the
projection
period
is
five
years
and
the
birth
rates
are
for
single
years,
the
number
of
births
obtained
by
multiplying
the
birth
rates
by
the
survived
female
population
must
be
multiplied
by
5
(i.e.,
the
number
of
years
In
the
projection
period)
to
obtain
the
total
number
of
births
occurring
between
the
base
and
projection
year.
Also,
note
that
In
order
to
obtain
the
number
of
males
and
females
born
during
the
projection
period
for
(.ontinues)
81

Figure
5.16
(rontinued)
use
in
the
next
projection
iteration,
the
total
number
of
births
is
multiplied
by
the
proportion
of
births
that
are
male
and
female
(51%
male
and
49%
female)
to
obtain
the
number
of
male
and
female
births.
Finally,
note
that
the
option
chosen
here
of
using
the
survived
population
as
the
base
for
calculations
is
only
one
of
several
options.
Other
acceptable
options
include
the
use
of
the
base
year
population
or
the
midpoint
population.
e.
Determine
the
number
of
females
who
will
migrate
between
the
base
year
and
the
projection
year
(Column
9,
Panel
B)
by
multiplying
the
age-specific
migration
rates
(Column
8,
Panel
B)
by
the
survived
female
population
in
the
corresponding
age
group
(Column
5,
Panel
B).
Note
that
a
negative
migration
rate
would
Indicate
net
outmigration
from
the
age
group.
f.
Determine
the
total
number
of
females
and
females
by
age
who
will
be
in
the
ages
0-44
in
the
projection
year
(Column
10,
Panel
B)
by
summing
the
number
of
survived
females
(Column
5,
Panel
B)
and
the
number
of
female
migrants
(Column
9,
Panel
B).
Note
these
values
become
the
beginning
population
values
for
the
next
iteration
of
the
projection
of
births
as
shown
in
Panel
D.
5.
Survive
the
population
from
the
base
year
to
the
projection
year
by
multiplying
the
age-specific
survival
rates
(Column
4,
Panel
A)
by
the
number
of
persons
in
the
corresponding
age
group
(Column
3,
Panel
A)
to
obtain
the
number
of
persons
by
age
surviving
to
the
projection
year
(Column
5,
Panel
A).
Note
that
the
births
between
the
base
year
and
the
projection
year
(as
computed
in
Panel
B)
are
also
survived
to
the
projection
year.
6.
Determine
the
number
of
persons
who
will
migrate
between
the
base
year
and
the
projection
year
(Column
7,
Panel
A)
by
multiplying
the
age-specific
migration
rates
(Column
6,
Panel
A)
by
the
corresponding
number
of
expected
persons
by
age
(Column
5,
Panel
A).
Note
that
the
use
of
the
survived
population
as
the
base
for
computing
migration
is
only
one
of
several
options
that
might
be
used.
Other
options
include
the
use
of
the
base
year
population
or
the
midpoint
population.
Also
note
that
if
the
migration
rate
is
negative,
this
indicates
outmigration
from
the
corresponding
age
group.
7.
Determine
the
total
population
and
the
total
population
by
age
for
Harris
County,
Texas
(Column
8,
Panel
A)
in
the
projection
year
by
summing
the
survived
population
(Column
5,
Panel
A)
and
the
number
of
net
migrants
(Column
7,
Panel
A)
for
the
corresponding
age
groups
and
across
age
groups.
This
population
becomes
the
beginning
(base
year)
population
for
the
next
iteration
of
the
projection
process
shown
in
Panel
C.
The
steps
delineated
above
will
be
repeated
for
each
iteration
of
the
projection
process.
(rontinues)
~

Figure
5.16
(amtinued)
Panel
A:
Projection
of
Population,
1995
Population
Survived
Age
of
April
1,
1990
Life
Table
Population
Net
Net
Total
Population
Plus
Births
Survival
April
1,
Milration
Migration
Population
1990
1995
1990-1995
Rates
1995
ates
1990-1995
1995
1
2
3
4
5
6
7
8
0-4
242,525
.9868
239,324
.0135
3,231
242,555
0-4
5-9
242,870
.9971
242,
166
.0390
9,444
251,610
5-9
10-14
230,837
.9986
230,514
.0641
14,776
245,290
10-14
15-19
209,144
.9968
208,475
.0632
13,176
221,651
15-19
20-24
206,843
.9934:
205,478
.0890
18,288
223,766
20-24:
25-29
223,612
.9919
221,801
.0962
21,337
24:3,
138
25-29
30-34
280,436
.9919
278,
164
.0797
22,
170
300,334
30-34
35-39
293,074
.9909
290,407
.0721
20,938
311,345
35.39
4:0-44
255,673
.9879
252,579
.0615
15,
534:
268,
113
4:0-4:4
45-4:9
211,781
.9818
207,927
.0526
10,937
218,864
4:5-49
50-54
158,343
.9714:
153,814
.0407
6,260
160,074
50-54
55-59
117
,728
.9562
112,572
.0354:
3,985
116,557
55-59
60-64
100,518
.9350
93,
984
.0326
3,064
97,0U
60-64
65-69
89,
118
.9058
80,723
.0310
2,502
83,225
65-69
70-74
73,807
.8635
63,732
.0266
1,695
65,427
70-74:
75-79
49,036
.8024
39,346
.0221
870
40,216
75+
80+
75,379
.5480
41,308
.0221
913
42,221
Tota
I:
3,131,434
(continues)
~

Figure
5.16
(continued)
Apri
1
1,.
1990
Age
Population
of
Females
Plus
Births
1990
1995
1990-1995
1
2
3
0-4
118,837
0-4
5-9
118,757
5-9
10-14
113,
105
10-14
15-19
102,484
15-19
20-24
100,791
20-24
25-29
110,
960
25-29
30-34
138,702
30-34
35-39
144,699
35-39
40-44
126,607
40-44
45-49
106,
183
Panel
B:
Projection
of
Births,
1990-1995
Female
Survived
Annual
Life
Table
Females
Number
Net
Net
Survival
April
1,
Birth
of
Births
Migration
Migration
Rates
1995
Rate
1990-1995
Rates
1990-1995
4
5
6
7
8
9
.9981
118,611
-
-
.0135
1,601
.9975
118,460
-
-
.0390
4,620
.9990
112,992
-
-
.0641
7,243
.9982
102,300
-
-
.0632
6,465
.9968
100,468
.0711
7,
143
.0890
8,942
.9961
110,
527
.1284
14,
192
.0962
10,633
.9957
138,106
.1124
15,523
.0797
11,007
.9946
143,918
.0597
8,592
.0721
10,376
.9919
125,581
.0204
2,562
.0615
7,723
.9874
104,
845
.0047
493
Total
Births/Year
19~0-95
=
48,505
Total
Births
1990-95
=
242,525
Female
Births
1990-95
=
118,837
(242,525
x
.49)
~
Female
Population
Ages
0-44
1995
10
120,212
123,080
120,235
108,765
109,UO
121,160
149,
113
154,294
133,305
(continues)

Figure
5.16
(amtinual)
Panel
C:
Projection
of
Population,
2000
Population
Sur:vived
Age
of
April
1,
1995
Life
Table
Population
Net
Net
Total
Population
Plus
Births
Survival
Apri
I
1,
Migration
Migration
Population
1995
2000
1995-2000
Rates
2000
Rates
1995-2000
2000
1
2
3
4
5
6
7
8
0-4
239,280
.9868
236,122
.0135
3,188
239,310
0-4
5-9
242,552
.9971
241,849
.0390
9,432
251,281
5-9
10-14
251,610
.9986
251,258
.0641
16,106
267,364
10-14
15-19
245,290
.9968
244,505
.0632
15,453
259,958
15-19
20-24
221,650
.9934
220,
187
.0890
19,597
239,784
20-24
25-29
223,765
.9919
221,953
.0962
21,352
243,305
25-29
30-34
243,
138
.9919
241,
169
.0797
19,221
260,390
30-34
35-39
300,334
.9909
297,601
.0721
21,457
319,058
35-39
40-44
311,345
.9879
307,578
.0615
18,916
326,494
40-44
45-49
268,
113
.9818
263,233
.0526
13,846
277,079
45-49
50-54
218,864
.9114
212,604
.0407
8,653
221,257
50-54
55-59
160,075
.9562
153,064
.0354
5,418
158,482
55-59
60-64
116,557
.9350
108,981
.0326
3,553
112,534
60-64
65-69
97,048
.9058
87,906
.0310
2,725
90,631
65-69
70-74
83,226
.8635
71,866
.0266
1,912
73,778
70-74
75-79
65,428
.8024
52,499
.0221
1,160
53,659
75+
80+
82,437
.5480
45,175
.0221
998
46,173
To
ta
I:
3,440,537
(continues)
~

Figure
5.16
(rontinued)
Age
of
Females
1995
2000
1
0-4
5-9
l0-14
15-19
20-24
25-29
30-34
35-39
40-U
2
0-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-U
45-49
Population
Apr
i
1
1,
1995
Plus
Births
1995-2000
3
111,
247
120'
211
123,080
120,235
108,765
109,UO
121,160
149,113
154,294
133'
305
Panel
D:
Life
Table
Survival
Rates
4
.9981
.9915
.9990
.9982
.9968
.9961
.9951
.9946
.9919
.9874
Projection
of
Births,
1995-2000
Survived
Females
April
1,
2000
5
111,024
119,910
122,957
120,019
108,417
108,983
120,639
148,308
153,0U
131,625
Birth
Rate
6
.0711
.1284
.1124
.0597
.0204
.0047
Annual
Number
Net
of
Births
Migration
1995-2000
Rates
1
7,708
13,993
13,560
8,854
3,
122
619
8
.0135
.0390
.0641
.0632
.0890
.0962
.0797
.0721
.0615
Total
Births/Year
1995-2000
=
47,856
Total
Births
1995-2000
=
239,280
Net
Migration
1995-2000
9
1,580
4,677
7,882
7,585
9,649
10,484
9,615
10,693
9,412
Female
Births
1995-2000
=
117,247
(239,280
x
.49)
Female
Population
Ages
0-U
2000
10
118,604
124,587
130,839
127,
604
118,066
119,467
130,
254
159,001
162,456
~

241
projections cannot be extensively described here (due to space limi-
tations), they generally involve applying a set of estimated or pro-
jected rates for the characteristic to be estimated or projected that
indicates the relative frequency of occurrence or incidence of the
characteristic in the population-such as rates of labor force participa-
tion, enrollment in school, householder status, persons with a given
health condition, or persons using a given product or service-to the
estimated or projected population. The rate may be a single rate for
the total population but more often occurrence or incidence rates are
used that are specific to given population cohorts (e.g., rates that are
specific to certain age, sex, race/ethnicity, or other characteristics).
Whatever the form of rate used, it is essential to recognize that
such population-based estimates and projections are subject to errors
due to both the assumptions underlying the population estimates or
projections and the assumptions about the rates of occurrence or
incidence for the population-based factor being estimated or project-
ed. Such population-based estimates and projections are of substan-
tial interest to those applied analyst who wish to discern the impli-
cations of a set of population estimates or projections for other
socioeconomic dimensions (Robey, 1985; Fosler et al., 1990) and
general knowledge of such procedures should be incorporated in the
knowledge base of applied analysts (Murdock et al., 1987b; Pitteng-
er, 1976).
Evaluation of Population Estimates
and Projections
The estimation and projection of populations is an inexact
science. In fact, nearly all evaluations of estimates and projections
relative to actual population trends suggest that errors are likely to
be substantial even for relatively short periods of time (Ascher, 1978;
Isserman, 1977, 1984; Keyfitz, 1982; Murdock et al., 1984; Stoto,
1983; Smith and Sincich, 1991). Because most estimates and projec-
tions utilize data on historical population patterns to determine the
assumptions underlying the estimates and projections, departures in
the demographic patterns of areas from those observed in historical
periods are likely to be unanticipated, resulting in estimates or
projections that are substantially different than the population pat-
terns that actually occur. Since there appears to be few bases (other
than historical patterns), for formulating assumptions about popula-
tion growth for periods for which population counts were not avail-
able or for future periods, marked departures from historical pat-
terns are likely to continue to negatively impact the accuracy of

242
population estimates and projections. The accuracy of estimates and
projections of populations for rapidly changing areas is thus likely to
continue to be problematic.
Despite the fact that estimates and projections have often been
found to be inaccurate even when carefully prepared, attempts to
evaluate the accuracy of estimates and projections are not without
merit. As noted by Murdock et al. (1989b; 1991c), the evaluation of
estimates and projections is one of the most essential elements in
the process of preparing population estimates and projections, par-
ticularly if the process is part of a long-term population analysis
program. Only by assessing the accuracy of past attempts to esti-
mate and project populations can the limitations of the methods and
assumptions underlying the use of a given method be identified for
any given projection area.
Evaluation Procedures
Although one cannot be certain (no matter how much prelimi-
nary analysis is completed), that a set of estimates or projections
will correctly estimate the population at an estimate date or predict
an area's future population, it is possible to assess the degree to
which a set of estimates or projections is logical relative to past
patterns. The following steps can be taken to assess estimates or
projections:
1. Examine them in comparison to historical patterns of
population change and to changes in the components of
population.
2. Evaluate them relative to other estimates or projections
that have been made for the estimation or projection area
or areas similar to the projection area.
3. Submit them to selected knowledgeable persons in the
estimation or projection areas for their assessment of the
validity of the assumptions and the estimated or projected
populations.
4. Complete sensitivity analysis of the effects of alterations
in key parameter assumptions.
5. Perform historical simulations in which the estimation or
projection model's accuracy in estimating or projecting
population change in past periods is evaluated.
Once a set of estimates or projections has been made and the
accuracy of the mathematical computations thoroughly validated, it

243
is useful to examine the trends suggested by the estimates or projec-
tions relative to historical patterns. In particular, by examining the
exponential rates implied by the estimated or projected changes in
the population relative to past patterns, the direction and magnitude
of changes for the total population and population subgroups rela-
tive to past patterns, and the assumptions used in computing the
estimates and projections, it should be possible to determine wheth-
er the projected values are consistent (or inconsistent) with historical
events. Although consistency with historical patterns does not
ensure accuracy (because the future may produce patterns different
than those of the past), departures from expected patterns that were
not intentionally induced by the analyst through the assumptions
made about the future in the methodology should lead one to
complete additional assessments of the computational accuracy and
consistency of the estimates or projections. ·
Comparisons to other available sets of estimates or projections
should also be made where possible. Although comparisons be-
tween the estimates or projections one has prepared and those
prepared by other persons or agencies do not provide certainty of
which (yours or the other) is most likely to be correct, such a
comparison, coupled with a comparison of the assumptions underly-
ing the estimates or projections, can indicate the effects of alterna-
tive assumptions on the estimates or projections. In addition,
comparisons with other estimates and projections made by other
sources can provide assurance that the parameter assumptions are
compatible with the judgment of other professionals involved in
estimation and projection activities.
Similarly, it is often useful to obtain reviews by knowledgeable
persons residing in the estimation or projection areas of both critical
assumptions underlying a set of estimates or projections and of the
final estimates or projections. In many cases, such persons have
observed and analyzed population changes in their areas over
extended periods of time and may be more knowledgeable about
local area population patterns than the analyst. If a large number of
geographically diverse areas is to be included in the estimates or
projections, consultation with a network of local persons such as city
or regional planners, demographers in local universities, and other
public and private service personnel can be useful in providing
information on the consistency of assumptions and of the estimates
or projections with the patterns experienced in the past in local
estimation or projection areas.
Sensitivity and historical simulation analyses are widely used
evaluation techniques (Alonso, 1968; Murdock et al., 1984; 1991c).

244
In sensitivity analysis, values for key parameters (demographic pro-
cesses or characteristics), such as birth or migration rates, are sys-
tematically altered and the results examined. If the changes which
occur are as expected, then some certainty that the estimation or
projection model is correctly simulating the key processes can be
obtained.
Historical simulations involve comparisons of estimation or
projections for census time periods to census counts for those same
periods. This usually involves using rates for historical periods
(e.g., 1970-80) to project population patterns for a known period of
time (e.g., 1980-90) and then assessing the accuracy of the estimates
or projections for a known date with the census count for the same
date (e.g., comparing estimates or projections for 1990 with 1990
census counts). The accuracy of estimates or projections relative to
counts can then be assessed using standard measures of error (see
below). Although accuracy in estimating or projecting populations
for past patterns does not guarantee the accuracy of estimates for
more recent periods or projections for future periods, an assessment
of the accuracy of estimates and projections relative to historical
patterns can provide at least some indication of the adequacy of the
estimation or projection procedure.
Historical simulation is an especially important process for an
ongoing estimation or projection program. Assessments of the
accuracy of past estimation and projection efforts relative to census
counts is essential to the refinement of procedures for such a pro-
gram. Such assessments should allow one to identify which as-
sumptions have been faulty for past periods and alert you to those
aspects of the estimation or projection procedure that require careful
monitoring in future estimation or projection activities.
Assessment Criteria and Measures of Accuracy
In any comparison of estimates or projections to census counts,
whether for historical or the most recent time periods, the assess-
ment of accuracy usually involves the use of certain criteria and
error measures. The standard criteria for assessing the results of
historical simulations is to examine estimation error; that is, the
difference between the estimated or projected population value and
the census value or count for a given date. In examining this differ-
ence, estimation error is usually evaluated in terms of rates (usually
percents) of error. Such rates of error are normally examined across
the entire population of areas (counties, cities, etc.) for which esti-
mates or projections have been made and evaluated in terms of

1. their absolute magnitude and relative magnitude
(compared to errors for other estimates or projections);
2. bias--that is, the extent to which the estimates or
projections overestimate or underestimate the census
values (i.e., the number or percent of areas that were
underestimated or overestimated and by how much);
and
3. patterns of variation in estimation error relative to:
-population size of areas,
-areas' rates of population growth, and
-type of area (counties, places, etc.).
245
In general, the smaller the estimation error and the less the bias,
as indicated by nearly equal percentages of areas being underesti-
mated and overestimated, the more acceptable the estimates or
projections. In addition, estimation errors would ideally be small
across all population size, growth rate, and type of area categories.
In most cases, however (as was noted in the introduction to this
chapter), rates of estimation error will be larger for areas with small-
er population sizes, areas that have shown the largest population
changes (either positive or negative), and (because of the association
between population size and type of areas) for subplace areas
(census tracts, etc.) and places than for counties or states.
Figure 5.17 presents the formulas for the most widely used
measures of estimation error. These measures differ in the manner
to which they take the direction of the error (that is, whether it is
positive or negative) and the population size of areas into account in
computing the rates of error. The mean percent eor (MPE) is a
simple mean of values in which negative and positive values cancel
one another. The mean absolute percent error (MAPE) measure's use of
absolute values does not allow positive and negative errors to cancel
one another and so provides a measure in which overall accuracy is
the focus. The mean percent absolute difference (MPAD) measure (also
referred to as the weighted mean absolute percentage error) controls
for both the effects of different types of errors (positive or negative)
and the effects of the population size of the estimation or projection
area. Whereas the mean absolute percent error gives all areas equal
weight (such that a 3-percent error for a city of 1,000 affects the
overall value by the same extent as a 3-percent error for a city of
1,000,000), the mean percent absolute difference measure weights all
areas by their population size. As shown in Figure 5.17, these

Figure
5.17:
Example
of
the
Use
of
Three
Commonly
Used
Error
Measures
Given:
Hypothetical
estimates
for
counties
in
the
St.
Louis
area
for
1988
and
using
Census
Bureau
prellmlnary
estimates
for
1988
as
the
standard
(assumed
to
be
the
correct
values)
County
St.
Charles
St.
Loui
11
Jefferson
Frankl
in
Sum
for
St.
Louis
Mean
Percent
Error
(MPE)
Mean
Absolute
Percent
Error
(MAPE)
Mean
Percent
Absolute
Difference
(MPAD).
1988
Estimate
206,000
961,000
172,500
78,500
1,418,000
n
I
isl
n
I
i=l
n
I
i=l
1988
Census
Estimate
Value
203,400
1,008,800
170,400
78,700
1,461,300
Error
(diffe-
rence)
2,600
-47,800
2,100
-200
-43,300
Percent
Error
1.
28
-4.74
1.
23
-0.25
-2.48
Estimate
Value
-
Census
Value
Census
Count
n
Estimate
Value
-
Census
Value
Census
Count
n
Estimate
Value
-
Census
Value
n
I
Census
Count
i=l
Absolute
Error
2,600
47,800
2,100
200
52,700
-2.48
4
7.50
4
52,700
Absolute
Percent
Error
1.
28
4.74
1.
23
0.25
7.50
-.62
1.88
3.61
1,461,300
~

247
measures produce different estimates of error. Which of these
measures is of most utility in evaluating a set of estimates or projec-
tions depends on the likely uses of the estimates or projections. In
general, the mean absolute percent error and the mean percent
absolute difference are likely to be of greater utility than simple
mean percent error measures since they provide a better indication
of the extent to which the average area is being correctly estimated
or projected.
The criteria and measures described above can be useful in
evaluating a set of estimates or projections, but none of them pro-
vides a definitive answer to the question of whether a set of esti-
mates or projections is sufficiently accurate. The answer to that
question is inherently judgmental. The question of whether a set of
estimates or projections will be sufficiently accurate also depends on
the likely uses of such estimates or projections. For example, if a set
of estimates or projections is to be used for facility planning and
error levels of up to 10 percent would not require changes in plans
for facility construction, then estimates or projections with errors of
10 percent may be acceptable; while if errors of 5 percent would
lead to population differences sufficient to require changes in facility
siz.e or location, then 10-percent errors would be unacceptable. The
acceptability of a set of estimates or projections must be ascer-
tained by the user in relation to specific needs (Murdock et al.,
1991c; lsserman, 1984).
In sum, the evaluation of estimates and projections is essential
to the estimation and the projection processes. Although an evalua-
tion can neither ensure the accuracy of a set of estimates or projec-
tions nor provide a definitive answer concerning the acceptability of
the level of error, it can provide important insight into the character-
istics of the estimates or projections and estimation or projection
methodologies and alert one to the potential strengths and weak-
nesses of such estimates or projections relative to events that have
occurred recently or are projected to occur in the future.
As noted above, although there is no fixed or single standard for
evaluating a set of estimates or projections, the procedures described
above are extremely useful for the applied analyst. More detailed
examinations of procedures for evaluating population estimates and
projections have recently been made by Murdock et al., (1989b;
1991c). These works provide some of the first systematic attempts
to develop a model for evaluating population estimates and projec-
tions. Readers interested in additional information on the evalua-
tion process may wish to review these publications.

248
Conclusions
In this chapter, we have described several estimation and projec-
tion methodologies for estimating and projecting populations. The
data requirements, assumptions, and computational steps for each of
several separate methodologies have been reviewed. In addition,
procedures for evaluating a set of estimates or projections relative to
past population patterns for the estimation or projection area, other
areas, and other available estimation and projection series have been
examined. In sum, a brief and basic introduction to the processes of
population estimation and projection has been presented. It should
be evident to the reader that in making or using population esti-
mates or projections, caution and discretion are essential. The fact
that no single technique has consistently produced accurate esti-
mates or projections and that any technique is only as accurate as
the assumptions upon which its procedures are based must be clear-
ly and continually stressed to potential users. _
The processes of population estimation and projection have
sometimes been referred to as being more similar to arts than to sci-
ences because of the need to utilize considerable judgment in speci-
fying assumptions for either estimates or projections. Therefore, it
is essential that any applied demographer attempting to complete a
set of population estimates or projections make a concerted effort to
master not only the mechanics of the population estimation and
projection processes, but also to obtain first-hand knowledge of the
estimation or projection area and of the limitations and the concep-
tual bases underlying the estimation and projection processes.

6
Summary and Condusions: The Future of
Population Change and Applied Demography
in the United States
This work has attempted to provide an introduction to the
concepts, methods, and data of applied demography. In so doing,
we have defined applied demography and delineated the key
elements and variables used in applied demographic analyses
(Chapter 1). The concepts of applied demography have also been
defined and recent trends in the variables used to measure these
concepts have been examined (Chapter 2). The major sources of
data for demographic and related areas were described, principles
for data use provided, and examples of typical applied uses of data
presented (Chapter 3). Basic measures of demographic processes
and characteristics as well as techniques for controlling the effects of
demographic variables were also examined (Chapter 4). Finally,
methods for completing and means for evaluating population esti-
mates and projections were reviewed (Chapter 5).
Throughout the description of these materials, the intent has
been to provide a concise overview of the major dimensions of
demography as used in applied analyses and to provide examples of
its application. Because it is only an introduction, anyone who
wishes to complete extensive applied demographic analyses will
need to utilize additional materials and references. Hopefully,
however, the work has provided both a useful introduction to the
field of applied demography and an initial indication of the wealth
of capabilities and insights that can be obtained by the applied use
of demographic perspectives and methods.
In this final chapter, we delineate those trends that are likely to
characterize future population patterns (and be the focus of applied
demographic analyses) in the coming decades. We also examine the
developments that are likely to characterize applied demography in
the future. Our intent is twofold. We wish to assist those involved
in applied analyses to anticipate future demographic patterns that
will impact the form and level of demand for future public- and

250
private-sector goods and services. We hope, as well, to assist ana-
lysts who may wish to pursue careers in applied demography to
discern those areas of the field that are likely to be the major areas
of growth and development in the coming decades.
Future Demographic Trends Impading
Products and Services
An Overview of Future Demographic Trends
Although it is difficult to anticipate future demographic trends,
an attempt is made here to (1) describe several major patterns of
change which are likely to have a pervasive impact on the United
States in the coming decades, and (2) delineate several demographic
patterns that, although unlikely to show the dramatic change antici-
pated for the preceding factors, appear likely to continue to effect
the population of the United States in the coming years. We con-
clude this section by examining the implications of these changes for
applied analyses.
Major Patterns Affecting the Population of the United States. At
least three patterns seem likely to be sufficiently pervasive to impact
future events in ways that will make them of importance for nearly
all applied analyses:
1. decreased rates of population growth;
2. an aging population base; and
3. an increasing number and proportion of minority residents.
Although these patterns cannot be examined in detail here, it is
possible to briefly delineate future trends related to these three
factors and examine some of their implications for selected areas.
Tables 6.1 through 6.3 provide projections of the total United States
population and of the population by age and race/ethnicity through
2050, and Tables 6.4 through 6.6 present data on the implications of
the projected population change for the future work force and for
college enrollment. These dimensions are only some of those likely
to be impacted by these demographic trends, but an examination of
them should be useful for delineating the general patterns likely to
impact numerous public- and private-sector goods and services in
the coming decades.
The data in Tables 6.1 through 6.3 indicate that many of the
current population and related patterns noted in Chapter 2 are

251
expected to continue. Population growth is projected to slow sub-
stantially such that the population would reach its maximum size
between 2040 and 2050 and begin to decline thereafter (Spencer,
1986; 1989). The minority population would grow rapidly, however,
with the proportion of the population composed of minority group
members increasing from about 25 percent in 1990 to more than 40
percent by 2050. Equally important, the growth in the minority
population is projected to account for nearly all net growth in the
United States population from now through 2050 with the Anglo
population declining (based on analysis of data in Table 6.1 in which
Hispanics have been subtracted from the white racial category to
produce Anglos).
The continued aging of the population is evident in Table 6.3.
The data in this table indicate that nearly 23 percent of the United
States population is projected to be 65 years of age or older by 2050
and 28 percent will be under 25 years of age. The data in this table
also show that there will be substantial variation in age among
ethnic groups. Although nearly 24 percent of the white population
is projected to be composed of persons 65 years of age or older by
2050, only 15 to 16 percent of Hispanics are projected to be 65 years
of age or older. On the other hand, while 35 percent of Hispanics
will be less than 25 years of age in 2050, only 28 percent of whites
will be of that age. Age differentials among ethnic groups will be of
critical importance in planning for future service and consumer
populations.
Equally important for near-term market and service analyses is
the need to recognize that, between now and 2010, the population
of the United States might best be characterized as middle-aged
rather than old. After 2010, the beginning edge of the baby-boom
generation will reach retirement ages, and as this generation enters
retirement ages, the population will age rapidly. Between now and
2010, the proportion of the population in elderly ages will change
relatively little but the proportion in middle-age age groups will
grow rapidly. The aging of the population is thus a long-term proc-
ess.
Tables 6.4 through 6.6 provide projections of some of the impli-
cations of the projected future change in population for the labor
force and for enrollment in higher education through 2025. Table
6.4 is derived from projections by the U.S. Bureau of Labor Statistics
(1989), while Table 6.5 uses data from the Bureau of Labor Statistics'
projections of future labor force participation rates (1987) and the
U.S. Census Bureau's population projections (Spencer, 1986; 1989)
to extend the projection of the labor force from 2000 to 2025. Table

252
Table 6.1: Historical and Projected Popu1ation
Growth In the United States ~ Race
and Spanish Origin, 1950-205
Population (in millions)
Total Other Spanishb
Year Population White Black Races Origin
1950 150.7 134.9 15.0 0.8
1960 180.7 160.0 19.0 1. 7
1970 205.1 179.7 22.8 2.6
1980 227.8 195.6 26.9 5.3 14.6
1990 250.4 210.6 31.1 8.7 19.9
2000 268.3 221.5 35.1 11. 7 25.2
2010 282.6 229.0 38.8 14.8 30.8
2020 294.4 234.4 42.1 17.9 36.5
2030 300.6 235.2 44.6 20.8 41. 9
2040 301. 8 232.0 46.2 23.6 46.7
2050 299.8 226.6 47.1 26.1 50.8
Percent Change from Previous Decade
1960 19.9 18.6 26.7 112.5
1970 13.5 12.3 20.0 53.9
1980 11. 1 8.9 18.0 103.9c
1990 9.9 7.7 15.6 64.2c 35.3
2000 7.2 5.2 12.9 34.5 26.6
2010 5.3 3.4 10.5 26.5 22.2
2020 4.2 2.4 8.5 21.0 18.5
2030 2. 1 0.3 5.9 16.2 14 .8
2040 0.4 -1.4 3.6 13.5 11.5
2050 -0.7 -2.3 1. 9 10.6 8.8
aValues for some years may differ from those shown in preceding
tables due to corrections made by the U.S. Bureau of the Census
following the reporting of the decennial census counts. The 1990
values shown are those projected in the source noted below, not
the 1990 census counts.
lpersons of Spanish origin may be of any race.
Cyalues are affected by the self-reporting by Hispanics as being of
other• racial group in 1980.
Source: Population values for 1950 are from the 1950 Decennial
Census from the United States Department of Commerce, Bureau
of the Census. Other values by race and ethnicity are from
Spencer (1986; 1989).

Tatie 6.2: Percent of Population by Race and
Spanish Origin In the United States,
1950--
Other Span is~
Year Total White Black .Races Origin
1950 100.0 89.5 10.0 .5
1960 100.0 88.5 10.5 1.0
1970 100.0 87.6 11. 1 1.3
1980 100.0 85.9 11. 8 2.3 6.5
1990 100.0 84. l 12.4 3.5 7.9
2000 100.0 82.6 13.1 4.3 9.4
2010 100.0 81. l 13.7 5.2 10.9
2020 100.0 79.6 14.3 6.1 12.4
2030 100.0 78.3 14.8 6.9 13.9
2040 100.0 76.9 15.3 7.8 15.5
2050 100.0 75.6 15.7 8.7 16.9
'1values for some years may differ from those shown In preceding
tables due to corrections made by the U.S. Bureau of the Census
following the reporting of the decennJal census counts. The 1990
values shown are those projected In the source noted below, not
the 1990 census counts.
~rsons of Spanish origin may be of any race.
Source: Population values for 1950 are from the 1950 decennial
census from the U.S. Bureau of the Census. Other values by
race and ethniclty are from Spencer (1986; 1989).
253

254
Table 6.3: Projections of the Percent of the U.S. Population
by Age and Race/Ethnicity for Selected Years,
1990-2ffi0 (total population in thousands)
Percent Population by Age
and Race/Ethnicity
Age Total
Year Group White Black Other Hispanica Percent
1990 18 24.4 32.1 30.3 35.7 25.6
18-24 10.1 12.2 11.5 12.0 10.4
25-44 32.7 31. 8 34.7 32.3 32.7
45-64 19.3 15.5 16.5 14.3 18.7
65+ 13.5 8.4 7.0 5.7 12.6
Total Population 210,616 31,148 8,646 19,887 250,410
2000 18 23.4 30.3 27.3 34.5 24 .5
18-24 9.0 11. 2 11.3 11. 0 9.4
25-44 30.1 30.6 32.3 30.1 30.2
45-64 23.6 19.0 20.7 17.6 22.9
65+ 13.9 8.9 8.4 6.8 13.0
Total Population 221,514 35,129 11, 623 25,233 268,266
2030 18 19.9 24.2 22.3 27.6 20.7
18-24 8.1 9.6 9.8 11.8 8.4
25-44 24 .7 26.0 28.0 27.5 25.1
45-64 24.2 22.7 23.8 20.0 24.0
65+ 23.1 17.5 16.1 13. 1 21. 8
Total Population 235,167 44. 596 20,866 42,514 300,629
2050 18 19.5 21. 9 20.3 24.3 19.9
18-24 7.9 8.9 9.1 11. 2 8. 1
25-44 24.3 25.1 27.4 27.7 24.7
45-64 24.6 23.8 23.9 22.6 24.4
65+ 23.8 20.3 19.3 15.6 22.9
Total Population 226,611 47' 146 26,093 50,790 299,849
aHispanlcs may be of any race.
Source: Computed from Spencer (1986; 1989).

255
6.6 uses United States population projections and 1986 eth-
nicity-specific college enrollment rates to examine the implications of
future population change for college enrollment in the United States
(Murdock et al., 1989a). Since the projected values reflect different
assumptions among population and service projection series, the
absolute values vary slightly from one table to another.
An examination of the data in Tables 6.4 and 6.5 suggests that
rates of growth in the labor force will slow substantially in the
coming years with rates of growth among middle-aged workers
exceeding those for younger workers, increases among women
exceeding those among men, and increases for minorities exceeding
those for other ethnic groups both in the immediate future (Table
6.4) and in the longterm (Table 6.5). Thus the labor force increased
by nearly 3 percent per year during the 1970s and by roughly 2
percent per year from 1980 to 1988. From 1988 to 2000, only the
highest rate of growth projected would equal that of the 1980s, and
from 2000 to 2025 (see Table 6.5) the growth would be very slow.
The labor force would increase by about 5 percent during the decade
from 2000 to 2010 (by only about 0.5% per year), but would decline
between 2010 and 2025 such that the total percentage increase for
the 25-year period would be only about two percent, an annual rate
of growth of less than one-tenth of one percent per year.
The data in Table 6.5 also suggest that patterns of change in the
labor force would vary widely among racial/ethnic groups. The
white labor force would decline by more than 3.0 million from 2000
to 2025, the black labor force would increase by about 3.0 million,
the number of persons in the labor force from other racial and ethnic
groups would increase by 3.5 million by 2025. The number of
Hispanics in the labor force would increase by more than 6.2 mil-
lion. Oearly the coming years will witness a substantial increase in
the minority labor force.
The data in Table 6.6 show the impacts of projected population
change in the United States on enrollment in higher education. The
data in this table indicate that there is likely to be little increase in
the total number of persons enrolled in college in the coming dec-
ades. Total enrollment is projected to decrease from 1990 to 2000,
increase from 2000 to 2010, and then decline to only 200,000 more
than in 1990 by 2025. The total growth from 2000 to 2025 would be
only two percent, and although the fastest period of growth from
2000 to 2010 would result in a 6.2 percent increase in enrollment for
the decade, the overall rate of growth is substantially slower than
the more than 100-percent increase in enrollment during the 1970s

Table
6.4:
Three
Alternative
Projections
of
the
U.S.
Ovilian
Labor
Force
by
Selected
Characteristics
for
2000
Projected
in
2000
by
Scenario
Percent
Change
(in
thousands)
1988
-
2000
Number
in
1988
Characteristic
(in
thousands)
High
Moderate
Low
High
Moderate
Low
Total
Labor
Force
121,669
146,
770
141,
134
137
,684
20.6
16.0
13.2
Age:
16-24
22,535
23,581
22,456
21,788
4.6
-.004
-0.3
25-54
84,042
104,471
101,267
100,686
24.3
20.5
19.8
55+
15,092
18,718
17,411
15,210
24.0
15.4
0.8
Sex:
Men
66,927
77,323
74,324
72,519
15.5
11.1
8.4
Women
54,742
69,447
66,810
65,165
26.9
22.0
19.0
Race/Ethnicity:
White
104,756
123,392
118,981
116,041
17.8
13.6
10.8
Black
13,205
17,074
16,465
16,103
29.3
24.7
21.
9
Asian
ang
Other
a
3,709
6,304
5,688
5,540
70.0
53.4
49.4
Hispanic
8,982
14,
696
14,
321
13,971
63.6
59.4
55.5
aThe
•Asian
and
Other
group
Includes
American
Indians,
AJaskan
Natives,
Asians,
and
Pacific
Islanders.
The
historic
data
are
derived
by
subtracting
Black
from
the
Black
and
Other
group.
11-ersons
of
Hispanic
origin
may
be
of
any
race.
Source:
United
States
Department
of
Labor,
Bureau
of
Labor
Statistics.
Monthly
IAbor
Review
WasNngton,
DC:
U.S.
Government
Printing
Office,
November,
1989.
~

Table
6.5:
Projections
of
the
Number
of
Persons
Jn
the
Labor
Force
Jn
the
United
States
by
Race/Ethnicity,
1986-2025
White
Black
Other
Spanish
Origin•
Total
Labor
Force
Year
Number
Percent
Number
Percent
Number
Percent
Number
1986
104,372,692
1990
108,874,115
2000
118,970,169
2010
122,482,481
2020
118,154,550
2025
115,329,677
86.00
13,278,913
10.94
3,718,256
3.06
7,854,495
85.46
14,418,615
11.32
4,110,047
3.23
8,880,561
83.92
16,945,391
11.95
5,851,990
4.13
11,702,066
82.31
18,904,561
12.70
7,420,189
4.99
14,490,615
80.59
19,719,732
13.45
8,740,350
5.96
16,848,400
79.73
19,934,836
13.78
9,378,810
6.48
17,982,304
aSpanish-orlgln
persons
may
be
of
any
race.
Percent
6.47
121,369,861
6.97
127,402,777
8.25
141,767,550
9.74
148,807,231
11.49
146,614,632
12.43
144,643,323
Souru:
Projected
by
the
authors
using
U.S.
Bureau
of
Labor
Statistics
(1987)
projections
of
rates
of
labor
force
partldpation
(United
States
Department
of
Labor,
Bureau
of
Labor
Statistics.
Projections
of
the
Economy,
Labor
Force
and
Occupational
Change
to
the
Year
2000,
•
Monthly
Ulbor
Review
110,(9)
November
1987)
and
population
data
from
Spencer
(1986;
1989).
~
'
I

258
and roughly the same as the relatively slow 6-percent increase
between 1980 and 1988 (U.S. Bureau of the Census, 1990a).
As with the labor force, however, the rate of growth in the
number of persons enrolled in college will vary widely by racial and
ethnic group. As is evident from an examination of Table 6.6,
whereas the number of whites enrolled would decline by more than
550,000 from 1990 to 2025, the number of blacks would increase by
nearly 400,000, the number of persons in other racial groups would
increase by more than 450,000 and the number of Hispanics would
increase by more than 650,000 from 1990 to 2025. College enroll-
ment will increasingly depend on minority involvement in higher
education. Other data (not shown here) indicate that the college
population will also become older with persons 35 years of age
increasing to more than 26 percent of all college students by 2025
compared to 16 percent in 1988. The college population will be both
more ethnically diverse and older in the coming decades.
Patterns of Continuing Importance. In addition to the three
trends described above, several other factors seem likely to show
slower rates of change than in the past. However, they are likely to
continue to display trends and patterns that depart from those of the
past sufficiently to impact major dimensions of life in the United
States. These patterns include further reductions in the levels of
mortality, particularly at older ages; continued low rates of fertility;
a continuation of relatively high rates of immigration; continued
patterns of population redistribution, but at slower rates than in the
past; a continuing diversity of household types; and continuing
disparity in socioeconomic resources, especially between minorities
and other groups. Although most of the trends anticipated for these
factors assume a continuation of the patterns noted in Chapter 2,
they are nevertheless important to recognize in analyses for future
time periods. ·
Mortality declined markedly in the 1970s and 1980s, particularly
at older ages, and the existing trends in medical research suggest
that it is at older ages that the impacts of mortality reduction are
expected to be most substantial (Stoto and Durch, 1990). This
lengthening of life at the upper end of the age structure will have
implications for the health care system and for services oriented to
serving the elderly (Siegel and Taeuber, 1986), but it is unclear
whether the extension of life will lead to increased demands for
services required by healthy elderly persons or largely increase the
need for services required to meet the needs of an increasingly
large, but frail, elderly population (Brody et al., 1987).

Table
6.6:
Projections
of
the
Number
of
Residents
Enrolled
in
Higher
Education
in
the
United
States
by
Race/Ethnicity,
1986-2025
White
Black
Other
Spanish
Origina
Total
Year
Number
Percent
Number
Percent
Number
Percent
Number
Percent
Enro
11
men
t
1986
10,605,708
85.10
1,454,952
11.
67
401,795
3.22
733,869
5.89
12,462,455
1990
10,331,092
84.31
1,468,539
11.99
453,457
3.70
774,065
6.32
12,253,088
2000
9,912,997
82.17
1,553,888
12.88
597,478
4.95
914,085
7.58
12,064,363
2010
10,369,818
80.94
1,717,640
13.U
724,864
5.66
1,162,737
9.08
12,812,322
2020
9,905,754
79.10
1,
773
I
071
14.16
844,468
6.74
1,324,170
10.57
12,523,293
2025
9,778,767
78.25
1,805,290
14.45
913,
181
7.31
1,410,541
11.29
12,497,238
aSpanish-orlgtn
Pft'9C'lS
may
be
of
any
race.
Source:
Projected
by
the
authers
using
U.S.
Bureau
of
the
Census
enrollment
rates
for
1986
and
Spencer
(1986;
1989).
t8

260
As noted in Chapter 4, the number of births and the birth rate
have increased in the last several years, largely as a result of in-
creased fertility among older women. Despite this recent pattern,
no resurgence of substantially increased fertility is expected. The
increasing involvement of women in the labor force and the contin-
ued economic needs of American households are expected to contin-
ue to keep fertility relatively low (Ryder, 1990).
The Immigration Reform and Control Act of 1986 was intended,
in part, to curtail the level of illegal immigration into the United
States in light of an already high level of legal immigration. Wheth-
er it will have a long-term impact is still unclear (Bean et al., 1989).
It seems apparent, however, that the world demographic situation
(Menken, 1986) and the labor force supply compared to the demand
for labor in developing nations (Espenshade, 1989) will lead to
continued relatively substantial immigration into the United States
in the coming years.
As indicated in Chapter 2, population redistribution in the
United States in recent decades has involved patterns that have
redistributed the population from the northeastern and midwestern
parts of the country to the south and west (Long, 1988), from the
central cities to the suburbs (Frey and Speare, 1988), and from
nonmetropolitan to metropolitan areas (Fuguitt et al., 1989; Johnson,
1989). Although predicting patterns of population redistribution is
the most difficult of all forms of projection, we anticipate that these
patterns will continue but decrease in magnitude over time. This
expectation is based on the fact that the populations in the southern
and western regions and in the suburbs and metropolitan areas are
large. Therefore, future rates of growth in these areas will likely be
smaller simply because increasing volumes of redistribution will be
necessary to maintain the rates of the past which were based on
much smaller numerical population bases. In addition, however,
despite the extensive volume of immigration noted above, the aging
of the population should lead to a population that is less mobile.
Although mobility and associated population redistribution will con-
tinue, rates are likely to become slower over time.
It is evident that the composition of households in the United
States has become increasingly diverse (Sweet and Bumpass, 1987).
The aging of the population slowed the overall rate of growth in the
number of households in the 1980s (U.S. Bureau of the Census,
1991e), but the proportion of married-couple households has not
increased, and there is little indication that the traditional family is
being restored. We expect that the rate of household formation will
slow because of the aging population base, but that a diversity of

261
household types will continue to characterize the population of the
United States because of the social and economic forces that contin-
ue to produce high rates of divorce and family disruption and which
lead to diverse forms of unions (Bumpass and Sweet, 1989).
The disparities in socioeconomic resources for different popula-
tion groups in the United States, especially minorities, have existed
for decades affecting income and poverty patterns, educational at-
tainment, occupational mobility (Farley and Allen, 1987), and the
physical segregation of population groups (White, 1987). Although
there have been signs of improvement in some factors, such as
increased rates of high school graduation for blacks and increased
college graduation rates among Hispanics (National Center for
Education Statistics, 1989), the trends still point to continuing and
large socioeconomic differences between minorities and other
groups. We anticipate some, but limited, closure in the differences
in socioeconomic resources between minorities and other groups in
the years to come, but the large differences in levels of and access to
resources between minorities and others are, unfortunately, expect-
ed to remain.
The Implications of Future Demographic Change
The major and continuing demographic trends noted above have
numerous implications for applied analyses, only some of which can
be reviewed here. This discussion is not intended to eliminate the
need for the reader to examine more exhaustive analyses of the
implications of such change (see for example, Teitelbaum and
Winter, 1985; Robey, 1985; Fosler et al., 1990). It is simply an
attempt to indicate the relevance of the expected patterns for some
of the factors that are likely to be of concern in applied analyses.
The slower growth of the population is likely to impact markets
and the demand for public and private goods and services in several
ways. Slower population growth will result in slower increases in
the markets for many goods and services. Slower growth will mean
that if the market for a product is to be increased, it will likely re-
quire identifying new market segments, new persons with different
characteristics than those who have traditionally used a product or
service. Marlet segmentation is likely to be increasingly required in
marleting analysis and in product promotion and advertising.
In addition, reduced growth is likely to increase the need for
more careful planning in many industries (such as real estate).
During the 1970s and, to some extent, the 1980s, population growth
could often be counted on to offset small errors in site location and

262
other forms of feasibility assessments. Slower growth will likely
lead to smaller allowable margins of error and increase the impor-
tance of planning and analysis.
Slower population growth may also require that the performance
of public- and private-sector managers be evaluated on an increas-
ingly diverse array of indicators. Growth in service populations or
in the number of customers may be less useful in differentiating
levels of performance. Quality indicators will likely continue to
increase in importance in performance appraisal compared to the
quantitative indicators used historically.
Both the increasingly middle-aged population of the next two
decades and the increase in the population in elderly ages thereafter
may have substantial impacts on goods and services. Table 6.7
shows the median household income levels for persons with differ-
ent age and other characteristics. The data in this table show that
middle-aged persons (those with a middle-aged householder) are
likely to be in their peak earning years. A middle-aged population
is one that is a relatively wealthy population, while younger and
older populations are likely to have lower income levels. The next
two decades should bring increased demand for goods and services
oriented to middle-aged, relatively affluent, households.
The longterm trend toward an elderly population has been
widely discussed by managers in the public and private sectors
(Siegel and Taeuber, 1986). Older populations will require increased
health-related products and other assistance-oriented services and
will generally demand different forms and types of services than a
middle-aged population. Given the siz.e of the baby-boom popula-
tion, and the magnitude of its impacts on both the growth of the
middle-aged population in the coming two decades and of the
elderly population after 2010, it is evident that political and socioec-
onomic policies will likely shift toward increasing concerns with the
problems of the elderly when the baby-boom population begins to
reach the elderly ages. Those in the public sector will need to be
alert to such shifts in order to effectively serve these clientele (and
perhaps to survive politically), and those in the private sector will
likely need to shift their products and services toward those oriented
to being purchased and monitored by public-service entities as the
baby-boom generation ages.
The growth of minority populations represents either a substan-
tial opportunity for the Nation or a potential problem, depending on
how access to resources is altered for minorities over the coming
years. Because of the young age structure of minority populations,
they offer the potential to partially offset the effects of the aging

263
workforce that will characterize the population as a whole, particu-
larly the white or Anglo population. A young minority workforce, if
properly educated, could give the United States a competitive
advantage relative to other developed nations.
If minority populations do not experience increased access to
socioeconomic resources, including increased levels of education, the
future labor force could be characterized by increased levels of
unemployment and the overall per capita purchasing power of the
population could decline. The data in Table 6.7 show, as did several
tables in Chapter 2, that the socioeconomic resources of blacks and
Hispanics are substantially more limited than those for whites (e.g.,
income is only 60 to 70% of that for whites). Unless, the socioeco-
nomic resources of minorities are increased, the projected growth of
the United States population could lead to decreased relative pur-
chasing power and related reductions in public-sector revenues.
Several of the continuing demographic trends also seem likely to
have effects that will be of interest to applied demographers and
other analysts. Increased longevity may further increase the
demand for goods and services oriented to the elderly and to those
with various physical disabilities. Small families resulting from
continued relatively low levels of fertility are likely to lead to con-
tinuing low levels of growth in the demand for educational services
for persons in traditional school ages, but to relatively large invest-
ments per individual child.
Continuing high levels of legal and illegal immigration will
continue to make immigration policy a topic of public concern.
Recurrent policy changes, aimed alternatively at welcoming the
disadvantaged or preventing the entrance of those with specific
characteristics, are likely to continue. Immigration will remain a
component of American life and will continue to increase the ethnic
diversity of the country. Immigration and the cultural differences
among immigrant groups will likewise remain a basis for product
segmentation.
Patterns of population redistribution will impact both the areas
of origin and of destination of migrants. Many central cities in the
Northeast and Midwest and many rural areas of the Nation seem
likely to continue to experience losses of population with accompa-
nying problems in the maintenance of their tax bases and in the
staffing of public and social service agencies. The continued subur-
banization of the population may additionally disadvantage central
cities relative to their more affluent suburbs.

264
Table 6.7: Median U.S. Household Income in 1989
by Selected Characteristics
Type of Household
All Households
Age of Householder:
15 to 24
25 to 34
35 to 44
45 to 54
55 to 64
65 and older
White
BI3ck
Hispanica
Family Households:
Married couples
Other family, female head
Other family, male head
aHlspanlcs may be of any race.
Median Income
$28,906
18,663
29,823
37,635
41,523
30,819
15,771
30,406
18,083
21,921
34,633
38,664
17,383
30,336
Source: Money Income and Poverty Status In the United
States, 1989. Current Population Report P-60, No. 168.
U.S. Department of Commerce. U.S. Bureau of the
Census. Washington, DC: U.S. Government Printing
Office, 1990.

265
The diversity of American households will require that the
evolution of goods and services towards meeting the needs of non-
traditional households continue. Products and services for non-
traditional households, such as legal services to address new means
of merging the interests of persons in non-traditional unions, are
likely to increase as are additional services to assist single parents.
Employee benefits may more frequently include child-care and elder-
care options and public and charitable organizations may find it
necessary to increasingly recognize the reality of non-traditional
households and families.
The disparity in socioeconomic conditions may mean that many
of the new growth markets will be composed of consumers with
relatively modest resources. Products and services appropriate to
the resources available to such consumers will be demanded. For
example, if the market for new single-family housing is going to be
maintained, concerted plans to finance and construct modestly
priced units may be required. Public policy also seems increasingly
likely to need to address the issues related to this disparity. This
attention is likely to occur because it is those groups that are most
disadvantaged that are the fastest growing segments of the popula-
tion and who are thereby playing an increasingly larger role in
public dedsionmaking. Minority issues and minority rights will
continue to be concerns in the coming decades.
The implications of future demographic change noted above are
speculative, and the record suggests that any attempt to project the
future should be viewed cautiously. The implications noted above,
however, are ones that reflect change in population growth and in
the characteristics of populations. Whether they occur to the extent
and in the form noted is uncertain, but the fact that demographic
change will substantially impact future events in the United States is
evident.
The Future of Applied Demography
Interest in the application of demographic perspectives and
methods to the analysis of applied problems is increasing. This is
evident in the growth in the reporting of demographic events in the
popular press, as well as in increasingly frequent inclusions of
demographic data and analyses in public- and private-sector plan-
ning and marketing analyses. It seems likely that the field of ap-
plied demography will continue to grow, becoming increasingly
important as data products and services become more accessible to
individual analysts, and that applied demographers will form a

266
larger component of the demographic profession. In this final sec-
tion, we examine the likely areas of growth and development of the
field in the coming years and delineate some of the potential oppor-
tunities and limitations that may affect that development. Although
we make no attempt to predict the exact time frames related to the
occurrence of these developments, we concentrate on those we
believe will emerge and become substantively important in the next
two decades.
Areas of Future Growth and Development
The concepts and methods of applied demography will be ap-
plied to additional areas of analyses in the coming years and grow
in their spheres of application in yet other areas. The traditional
role of applied demography and its practitioners in the analyses of
demographic variables for private-sector marketing, strategic plan-
ning, site-selection, market segmentation, and similar forms of
private-sector analysis will certainly continue. The use of applied
analyses in state and local demography for the production of esti-
mates, projections, and other information for governmental and
public-service planning, budgeting, and related forms of analysis
will remain areas of importance for applied demographers. Similar-
ly, applied demographers will continue to play instrumental roles in
the provision and analysis of demographic and other information
through data vending and consulting firms. Together, these areas
have been and will continue to form the base of applied demogra-
phy. Growth in these areas, however, may be limited. Federal,
state, and local governmental budget problems seem unlikely to be
quickly resolved, the areas in which many private-sector demogra-
phers are employed may show only modest growth, and traditional
forms of data provision and related consulting activities are increas-
ingly mature fields.
Demographic and socioeconomic events in the United States and
throughout the world, however, seem likely to create areas where
growth in demand for applied demographic services will be particu-
larly pronounced. Among those areas in which the most substantial
growth seems likely to occur are:
1. human resources planning;
2. health care service planning and marketing;
3. long-term care planning;
4. pension fund investment and utilization planning;
5. legislative redistricting;

267
6. environmental impact assessment and mitigation;
7. social, political, and socioeconomic policy analysis;
8. international business and marketing analysis; and
9. individualized data use and analysis.
Human resources are likely to become more difficult to manage
in the coming decades as population growth slows and the charac-
teristics of the population change. The slower growth of the labor
force discussed above clearly suggests that employers will need to be
more competitive in attracting new employees. At the same time,
the faster growth of minority populations could result in either a
younger and more viable work force or a less competitive one,
depending on whether or not minority and other disadvantaged
populations are provided with the access to educational and other
training opportunities necessary to develop competitive labor market
skills. Finally, it is evident that the growth of minority and other
low-income populations are likely to require an initial infusion of
expanded services to address their presently inadequate levels of
education and income so that they can develop competitive skills.
As a result, the level of demand for education, employment, and
other types of human services is likely to increase. These patterns
suggest that the need for applied demographic expertise in the
planning for human resources and human services is likely to in-
crease in both the public- and private-sectors in the coming years.
It is obvious that the aging of the United States population will
lead to increased demand for health care and long-term care for the
elderly. Applied demography can be used to discern much about
the patterns of population change likely to impact health and
morbidity in the coming years. Analyses of the demand for differ-
ent types of health care services, the feasibility of advanced technol-
ogy acquisition relative to the population base in specific service
areas, the demand for specialty services such as drug and alcohol
counseling, and other areas will increasingly require demographic
expertise. Similarly, long-term health and care facilities require
careful analysis of such factors as the likely mix of Medicaid and
private paying clients and careful siting to ensure that there is a
sufficient population within close proximity of the site and that the
site is sufficiently close to clients' relatives and friends (Murdock
and Hamm, 1991). Health and long-term care are likely to be
important areas of growth for applied demographic analysts.
Another area likely to receive increased emphasis because of the
aging of the United States population base is that of pension fund
investment and utilization planning. Managers in both public- and

268
private-sector entities are beginning to recognize that pension funds
may be extensively impacted by the coming increase in elderly reti-
rees. How many persons are likely to retire and at what periods is
critical information for establishing what types of investments
should be pursued by the managers of retirement funds. Demo-
graphic analysis is important as well in determining where invest-
ments in real estate and other developments and in various types of
corporations should be made. As the likely impacts of the aging
population base become apparent, demands for demographic exper-
tise to assist in the selection of pension fund investments is likely to
increase substantially.
The rapid growth of minority populations lies at the base of a
substantial increase in demand for applied demographic analysis
related to legislative and local area redistricting (see Hill and Kent,
1988). Particularly in the Voting Rights Act's designated (southern)
states that have had rapidly increasing minority populations, the
demand for persons who have expertise in population analysis and
knowledge of redistricting is growing rapidly. The expertise of
applied demographers is essential to assist in the development of
voting districts for state and local governments, for school and
junior college districts, to assist in the analysis of racial block voting
patterns, and to serve as expert witnesses regarding the demograph-
ic bases of alternative redistricting plans. As additional 1990 Census
data become available and the number of additional areas involved
in redistricting grows, the services of applied demographers are
likely to be in increased demand.
During the 1970s and the first few years of the 1980s, environ-
mental impact assessment firms employed demographers to deter-
mine the number of persons likely to be directly and indirectly
impacted by large-scale developments. This activity declined sub-
stantially by the last part of the 1980s as energy prices fell and large-
scale energy and other projects were canceled or delayed (Murdock
et al., 1986). There is renewed concern about environmental issues,
however, about the storage and/or disposal of industrial and nuclear
wastes (Murdock et al., 1983), and the need to address such issues
as the impacts of global climate change on human populations
(Murdock and Leistritz, 1991). Increased demand for environmental-
ly related impact assessments seems likely to occur with emphasis
being placed on estimating the populations impacted by past waste
and/or other noxious products and on the determination of the
populations that should be compensated for past impacts. Applied
demographers will play a key role in such analyses.

269
Policy analyses have always employed demographic data.
However, there is a new found awareness of the role of demograph-
ic factors in such changes as those projected to occur in the labor
force (U.S. Bureau of Labor Statistics, 1989), in health (Brody et al.,
1987), and in other areas (Robey, 1985). We expect that governmen-
tal agencies providing health and human services will increasingly
include demographic analyses in their planning efforts and, as a
result, increasingly require the services of applied demographers.
Although most applied demographers are presently employed to
complete domestic analysis (and this work has emphasized such
analyses), it appears likely that the global economy will require
expansion of American firms' marketing efforts in other nations.
Demographers have long been involved in international develop-
ment and family planning throughout the world as part of academic,
United Nations, U.S. Agency for International Development, and/or
foundation-sponsored activities. The new demand, however, will
likely be for applied demographers to assist firms in determining the
markets for different types of products and services. The present
involvement of applied demographers in the private-sector makes
this emerging area one that may expand the demand for applied
private-sector demographers.
The explosion of such data-handling capabilities as high-speed,
large-storage capacity microcomputers coupled with such data media
as CD-ROM makes demographic data more accessible to analysts
who are not affiliated with large public-sector organizations or large
corporations. Use of these and related forms of technology will
produce an expanding market for the products and services of the
individual data entrepreneur and analyst. Applied demographic
products, such as specialized software for demographic analysis,
additional on-line data bases, and numerous other products to assist
the individual data user and analyst, should produce increased
demand for the services of applied demographers.
Overall, then, these and many other areas should expand the
demand for applied demographic data and analyses and for applied
analysts. For those interested in careers in applied demography, the
future appears likely to provide substantial opportunities.
Potential Opportunities and Problems
for Applied Demography
Although several specific areas where opportunities may develop
for applied demographic analysts were described in the last section,
here we wish to discuss several generic opportunities and problems

270
that are not related to a specific area or subfield of analysis. We
first discuss emerging opportunities and then discuss potential prob-
lems affecting the future of applied demography.
Emerging Opportunities. We believe that there are substantial
opportunities for the field of applied demography to improve its
general level of conceptual and analytical capabilities by:
1. developing concepts and methods to integrate its knowl-
edge base with that in related fields such as regional
economics, sociology, geography, urban planning, and
applied psychology and social psychology;
2. ensuring that it employs the most sophisticated appropri-
ate methodologies available; and by
3. developing procedures to integrate its methods with
computer-based geographic methods such as Geographic
Information Systems (GIS).
Many of the concepts of demography have been developed or
are at least widely used in other fields of analysis. In this regard,
such areas as regional economics, which examines the role of
changes in local economies on employment and income growth, is
clearly of importance for such applied demographic activities as
determining local levels of migration and related population esti-
mates. Such areas in sociology as human ecology and urban and
rural sociology tend to view social change in terms of population-
based concepts that may be of use to applied demographers. Many
geographers are also demographers, but applied demography has
tended to ignore geographic-based concepts of geographers interest-
ed in demographic patterns. In fact, many demographers with a
sociological or economic background are largely unaware of the
parallel literature in geography on such topics as population redistri-
bution and suburbanization, or minority population growth. Simi-
larly, applicable knowledge from urban planning, such as the effects
of physical and transportation planning on population growth, is
often not part of the base of knowledge of applied demographers.
Finally, although applied demographers are often employed in
interdisciplinary research units within marketing or other depart-
ments and work on a regular basis with psychologists and social
psychologists, there is relatively little indication that such work has
been appropriately incorporated in applied demographic analyses.
These are only some examples; others could have been noted. The
point, however, is that applied, real-world problems are seldom

271
resolved by analysts from a single field; they are inherently multi-
dimensional and require multidisciplinary approaches. Applied
demographers should utilize existing opportunities to incorporate
relevant knowledge from other disciplines into their knowledge
base.
Applied demography should also take advantage of the expand-
ing range of increasingly sophisticated forms of analysis employed in
formal demography and other statistical and social science disci-
plines. _Although the applied analyst must always remain sensitive
to the need to effectively communicate results to clientele groups,
the analysis underlying the results reported should be as appropri-
ately sophisticated as possible. This is not meant to imply that
applied demographers should employ complex methods simply to
show their academic prowess, but rather to emphasize that the need
for a constant updating of knowledge is as important in applied as
in other areas of demography. The fact that such techniques as
hazard models, multi-dimensional and multi-state projection models,
and even log-linear models are seldom used in applied demographic
analysis suggests that applied demographers should take steps to
develop such methods for use in their more complex forms of analy-
sis. Applied demography, like the overall field of demography,
requires substantial and continuing methodological and conceptual
development.
The development of geographic-based products such as Geo-
graphic Information Systems (GIS) and the TIGER census maps
suggest the need for applied demographers (whose work tends to be
centered on specific local areas) to become more heavily involved in
the opportunities to develop geographic-based demographic analy-
ses. For example, such areas as population estimation may well
benefit from such an approach (Tayman, 1991); the applicability of
GIS techniques for estimating the population in areas within speci-
fied distances of given locations is clearly relevant to site location
analyses and is widely used (Merrick and Tordella, 1988). The
simple mapping capabilities of GIS systems allow one to complete
examinations of patterns of population change within urban areas at
an unprecedented level of detail. Although we would not presume
at this early stage of their utilization to suggest how GIS and other
such systems might be integrated with applied demographic con-
cepts and techniques, it seems evident that this is an area with
substantial opportunities for applied demographers.

272
Potential Problems. As in nearly all fields of study, past devel-
opments and future demands may also create problems which could
limit the developing field of applied demography. Among the
potential problems that we believe must be carefully avoided are:
1. the extension of demographic concepts and methods
beyond their empirical and conceptual bases; and
2. the popularization of demographic concepts and the
demographic approach such that they are inappropriately
applied and their limitations unrecognized.
Demography has a base of knowledge and methods that have
been developed over several centuries. It is a mature field
with much to offer the applied analyst. Its base of knowledge is
limited, however, and some tendency has arisen to examine the
demographic concomitants of nearly any phenomena as a means of
demonstrating the relevance of demographic factors. The old adages
that correlation is not causation and that demography is not destiny
should always be kept in mind when applying demographic con-
cepts. The simple fact that some behavior or phenomenon varies
with a demographic characteristic is not a sufficient reason to
assume the importance of the demographic correlate. The conceptu-
al, substantive, empirical, and statistical bases of such relationships
must be examined before the relevance of a demographic character-
istic can be assumed. It is essential that applied demography not
extend its application too far beyond the conceptual and empirical
base of knowledge in demography and related fields. This is not
intended to imply that applied demographers should limit their
attempts to discern how demographic factors relate to as many
previously unexamined factors as possible, but is only an admoni-
tion that such relationships should be thoroughly analyzed before
they are assumed to be substantively significant.
We are also concerned that the popularization of demography be
pursued with care. This is not because we believe that familiarity
breeds contempt, but rather because of the concern, noted above,
that we not claim importance for demographic variables prior to
finding empirical support for such importance. In addition, unless
carefully qualified and contextualized, popularization can lead to the
widespread misuse of the methods and concepts of applied demog-
raphy by persons who lack adequate training in their use. This is
not a statement based on a pandering after academic elitism, be-
cause we strongly believe that applied demography should be
extended to nondemographers for their use in resolving applied

273
problems. It is, however, a call for caution in implementing that
extension to be sure that what is conveyed is also understood.
Although popularized demographic analyses done by entities such
as American Demographics have been competently completed and
properly qualified, the same cannot be said for some of the newly
emerging coverage appearing in the print and broadcast media.
There is a clear need to ensure that applied demographic knowledge
is not trivialized or its credibility eroded by careless popularization
of its concepts and methods.
Condusions
In sum, this work has attempted to provide an introduction to
the field of applied demography. As the discussion in this chapter
suggests, we believe the field is one with significant opportunities
and that it is likely to expand substantially in the coming years if it
remains firmly grounded in a solid conceptual and empirical base of
knowledge. We trust that this effort will be but one of many that
will further expand and systematize the field of applied demogra-
phy. Even more important, we hope this work, and those to follow,
provide concepts, methods for analysis, and understanding of
human populations that are useful to those who are attempting to
arrive at solutions to real-world problems.

References
Alonso, W.
1968 Predicting best with imperfect data.• Journal of the American Insti-
tute ofPlanners 4:248-254.
American Demographics
1989 The Best 100 Sources for Marketing Information: Who's Who from
American Demographics. New York: American Demographics.
American Statistical Association
1977 Report to the Conference on Economic and Demographic Methods
for Projecting Population. Washington, DC: American Statistical
Association.
Ascher, W.
1978 Forecasting: An Appraisal for Policy Makers and Planners. Baltimore:
Johns Hopkins University Press.
Backman, K.
1989 An empirical analysis of the utility of a human ecological approach
for explaining the level of urban economic development. Doctoral
Dissertation. College Station: Department of Urban and Regional
Planning, Texas A:M University.
Barclay, G.W.
1958 Techniques of Population Analysis. New York: John Wiley and Sons,
Inc.
Bean, F.D., A.G. King, R. D. Benford, and L.B. Perkinson
1982 Estimates of the Number of Illegal Migrants in the Stllfe of Texas.
Austin, Texas: University of Texas Population Research Center.
Bean, F.D., A.G. King, and J.S. Passel
1983 The number of illegal migrants of Mexican origin in the United
States: sex ratio-based estimators for 1980.• Demography 20:99-109.
Bean, F.D., G. Vernez, and C.B. Keely
1989 Opening and Closing the Doors: Evaluating Immigration Reform and
Control. Rand Corporation and the Urban Institute. Washington,
DC: The Urban Institute Press.
Berry, B.J.L., and J.D. Kasarda
1977 Contemporary Urban Ecology. New York: Macmillan Publishing
Company, Inc.
Bianchi, S., and E. McArthur
1991 Family disruption and economic hardship: the short-run picture
for children.• Current Population Reports. Series P-70, No. 23.
Washington, DC: U.S. Government Printing Office.

276
Bianchi, S.M., and N. Rytina
1986 The decline in occupational sex segregation during the 1970s:
census and CPS comparisons. Demography 23:79-86.
Bogue, D.J.
1950 A technique for making population estimates, Journal of the
American Statistical Association 45:149-163.
1974 Techniques for Making Population Projections: How to Make Age-Sex
Projections by Electronic Computer. Chicago: University of Chicago
Community and Family Study Center.
1985 The Population of the United States: Historical Trends and Future Pro-
jections. New York: The Free Press.
Bouvier, L.F., and R. Gardner
1986 Immigration to the U.S.: the unfinished story. Population Bulle-
tin 41:4. Washington, DC: Population Reference Bureau, Inc.
Brody, J.A., D.B. Brock, and T.F. Williams
1987 Trends in the Health of the Elderly Population. Annual Review of
Public Health. Palo Alto, CA: Annual Reviews, Inc.
Bumpass, L.L., and J.A. Sweet
1989 National estimates of cohabitation. Demography 26:61~25.
Oogg, C.C., and S.R. Eliason
1988 A flexible procedure for adjusting rates and proportions, includ-
ing statistical methods for group comparisons. American Sociologi-
cal Review 2:267-283.
Cohen, L.E., and M. Felson
1979 Social change and crime rate trends: a routine activity approach.
American Sociological Review 4:588-608.
Das Gupta, P.
1978 A general method of decomposing a difference between two rates
into several components. Demography 15:99-112.
1990 Decomposition of the difference between two rates when the fac-
tors are nonmultiplicative with applications to U.S. Life Tables.
Paper presented at the annual meetings of the Population Associa-
tion of America. Toronto, Canada.
Duncan, O.D., D.L. Featherman, and B. Duncan
1972 Socioeconomic Background and Achievement. New York: Seminar
Press.
Ericksen, E.P.
1973 A method for combining sample survey data and symptomatic
indicators to obtain estimates for local areas. Demography 4:137-
160.

277
Espenshade, T.J.
1989 Growing imbalances between labor supply and labor demand in
the Caribbean basin.• In F.D. Bean et al. (eds.), Mexican and Central
American Population and U.S. Immigration Policy. Austin: University
of Texas Center for Mexican American Studies.
Farley, R., and W.R. Allen
1987 The Color Une and the Quality of Ufe in America. New York: Russell
Sage Foundation.
Fosler, R., W. Alonso, J. Meyer, and R. Kem
1990 Demographic Change and the American Future. Pittsburgh: Univer-
sity of Pittsburg Press.
Fossett, M.A., and K.J. Kiecolt
1990 Mate availability, family formation, and family structure among
Black Americans in nonmetropolitan Louisiana 1970-1980. Rural
Sociology 55:305-327.
Frey, W.H., and A. Speare, Jr.
1988 Regional Metropolitan Growth and Decline in the United States. New
York: Russell Sage Foundation.
Fuguitt, G.V., D.L. Brown, and C.L. Beale
1989 Rural and Small Town America. New York: Russell Sage Founda-
tion.
Glenn, N.D.
1977 Cohort Analysis. Sage University Paper Series on Quantitative
Applications in the Social Sciences. Beverly Hills, CA: Sage Publi-
cations.
Greenburg, M.R., D.A. Krueckeburg, and C.O. Michaelson
1978 Local Population and Employment Projection Techniques. New Bruns-
wick, NJ: The Center for Urban Policy Research.
Greenwood, M.J.
198.5 Research on internal migration in the United States.• Journal of
Economic Literature 2:397-433.
Haub, C.
1987 Understanding population projections.• Population Bulletin 42:4.
Washington, DC: Population Reference Bureau, Inc.
Hauser, P.M., and O.D. Duncan (eds.)
1959 The Study of Population. Chicago: University of Chicago Press.
Hertsgaard, T.S., S. Murdock, N. Toman, M. Henry, and R. Ludtke
1978 REAP Economic-Demographic Model: Technical Description. Bismarck:
North Dakota Regional Environmental Assessment Program.
Hill, D.B., and M.M. Kent
1988 Election Demographics. Population and Public Policy Series. Wash-
ington, DC: Population Reference Bureau, Inc.

278
Hollman, F.
1989 U.S. population estimates by age, sex, race, and Hispanic origin:
1989. Current Population Reports. Series P-25, No. 1057. U.S.
Bureau of the Census. Washington, DC: U.S. Government Print-
ing Office.
Irwin, R.
1977 Guide for Local Area Population Projections. U.S. Bureau of the
Census. Washington, DC: U.S. Government Printing Office.
Isserman, A.
1977 The accuracy of population projections for subcounty areas.
Journal of the American Institute of Planners 43:247-259.
1984 Projection, forecast, and plan: on the future of population fore-
casting. Journal of the American Planning Association 50:208-221.
Johnson, K.M.
1989 Recent population redistribution trends in nonmetropolitan
America. Rural Sociology 3:301-326.
Kendall, M.C.
1977 Labor-market models. Pp. 49-58 in Population Forecasting for Small
Areas. Oak Ridge, 1N: Oak Ridge Associated Universities.
Kerlinger, F.N., and E.J. Pedhazur
1973 Multiple Regression in Behavioral Research. New York: Holt, Rinehart
and Winston.
Keyfitz, N.
1972 On future population. Journal of the American Statistical Association
67:347-363.
1982 Can knowledge improve forecasts? Population and Development
Review 8:729-752.
Kitagawa, E.M.
1955 Components of a difference between two rates. Journal of the
American Statistical Association 50:1168-1194.
Land, K.C., and A. Rogers (eds.)
1982 Multidimensional Mathematical Demography. New York: Academic
Press.
Leistritz, F.L., and S.H. Murdock
1981 The Socioeconomic Impact of Resource Development: Methods of Assess-
ment. Boulder, CO: Westview Press.
Liao, T.F.
1989 •A flexible approach for the decomposition of rate differences.
Demography 4:717-726.
Lichter, D.T.
1989 Race, employment hardship, and inequality in the American
nonmetropolitan South. American Sociological Review 3:436-446.

279
Lichter, D.T., and J.A. Constanzo
1987 Nonmetropolitan underemployment and labor-force composition.•
Rural Sociology 3:329-344.
Lichter, D.T., and G.F. De Jong
1990 The United States. In C.B. Nam, W.J. Serow, and D.F. Sly (eds.),
International Handbook on International Migration, New York:
Greenwood Press.
Long, L.
1988 Migration and Residential Mobility in the United States. New York:
Russell Sage Foundation.
McGirr, N.J., and S.O. Rutstein
1987 Comparison of microcomputer population projection programs.
Paper presented at the annual meetings of the Population Associa-
tion of America. Chicago.
Macunovich, D., and R.A. Easterlin
1990 How parents have coped: the effect of life cycle demographic
decisions on the economic status of preschool age children, 1964-
87.•Population and Development Review 2:301-325.
Mangalam, J.J., and H.K. Schwarzweller
1968 General theory in the study of migration: current needs and
difficulties.• The International Migration Review 1:1-18.
Mason, K.0., W.M. Mason, H.H. Winsborough, and W.K. Koole.
1973 Some methodological issues in cohort analysis of archival data.•
American Sociological Review 38:242-58.
Massey, D.S., and N.A. Denton
1988 The dimensions of residential segregation.• Social Forces 2:281-315.
1989 Hypersegregation in U.S. metropolitan areas: black and Hispanic
segregation along five dimensions.• Demography 3:373-391.
Menken, J. (ed.)
1986 World Population and U.S. Policy: The Choices Ahead. New York:
W.W. Norton and Company.
Merrick, T.W., and S.J. Tordella
1988 Demographics: people and markets.• Population Bulletin 43:1-47.
Washington, DC: Population Reference Bureau, Inc.
Messner, S.F., and R.J. Sampson
1991 The sex ratio, family disruption, and rates of violent crime: the
paradox of demographic structure.• Social Forces 3:693-713.
Morrison, P.
1971 Demographic Infonnation for Cities: A Manual for Estimating and Pro-
jecting Local Popullltion Characteristics. Washington, DC: Depart-
ment of Housing and Urban Development.

280
Murdock, S.H.
1982 A Primer on Basic Demographic Methods for Small-Area Analysis.
Austin, TX: Texas State Data Center.
Murdock, S.H., and R.R. Hamm
1988a Demographic Factors Affecting Real Estate. Technical Report. College
Station, TX: Real Estate Center, Texas A:M University.
1988b Secondary Sources of Research Data. Applied Community Research
Monograph A3. Alexandria, VA: American Chamber of Commerce
Researchers Association.
1991 •A demographic analysis of the market for a long-term care facility:
a case study in applied demography. In T. Merrick, P. Morrison,
P. Voss, and H. Kinter (eds.), Case Studies in Applied Demography
(forthcoming).
Murdock, S.H., and F.L. Leistritz
1991 The socioeconomic dimensions of global climate change: search-
ing for a theoretical and empirical base.• Sodety and Natural Re-
sources (forthcoming).
Murdock, S.H., and T.K. Ostenson
1976 Population Projections by Age and Sex, 1975-2000. Fargo: North
Dakota Agricultural Experiment Station.
Murdock, S.H., F.L. Leistritz, L.L. Jones, D. Andrews, B. Wilson, D.
Fannin, and J. deMontel
1979a The Texas Assessment Modeling System: Users Manual. College Sta-
tion, TX: Texas Agricultural Experiment Station, Texas AM
University System.
1979b The Texas Assessment Modeling System: Technical Description. College
Station, TX: Texas Agricultural Experiment Station, Texas AM
University System.
Murdock, S.H., F.L. Leistritz, and R.R. Hamm (eds.)
1983 Nuclear Waste: Socioeconomic Dimensions of Long-Term Storage.
Boulder, CO: Westview Press.
Murdock, S.H., F.L. Leistritz, R.R. Hamm, B. Parpia, and S. Hwang.
1984 •Assessment of the accuracy of a regional economic-demographic
projection model.• Demography 21:383-404.
Murdock, S.H., F.L. Leistritz, and R.R. Hamm.
1986 The state of socioeconomic impact analysis in the United States:
limitations and opportunities for alternative futures.• Journal of
Environmental Management 23:99-117.
Murdock, S.H., LL. Jones, R.R. Hamm, and F.L. Leistritz
1987a The Texas Assessment Modeling System (TAMS): User's Guide. Col-
lege Station, TX: Texas Agricultural Experiment Station, Texas
A:M University System.

281
Murdock, S.H., R.R. Hamm, S. Hwang, and K. Backman
1987b PopulaHon ProjecHons: A Review of Basic Principles, PracHces, and
Methods. College Station, TX: Texas Agricultural Experiment Sta-
tion, Texas A:M University System.
Murdock, S.H., Md. N. Hoque, and R.R. Hamm.
1989a State and Regional Projections for Texas to 2025 of Population-Based
Change in the: Labor Force, Enrollment in Elementary and Secondary
Schools, Enrollment in College, Households, and Incidences of
Diseases/Disorders. College Station, TX: Texas Agricultural Experi-
ment Station, Texas A:M University System.
Murdock, S.H., R.R. Hamm, D. Fannin, B. Pecotte, and P. Voss
1989b EvaluaHng Small-Area PopulaHon EsHmates and ProjecHons. Applied
Community Research Monograph E3. Alexandria, VA: American
Chamber of Commerce Researchers Association.
Murdock, S.H., K. Backman, E. Colberg, Md. N. Hoque, and R.R. Hamm.
1990 Modeling demographic change and characteristics in the analysis
of future demand for leisure services.• Leisure Sciences 12:79-102.
Murdock, S.H., K. Backman, Md. N. Hoque, and D. Ellis.
1991a The implications of change in population size and composition on
future participation in outdoor recreational activities.• Journal of
Leisure Research 3:238-259.
Murdock, S.H., K. Backman, R.B. Ditton, Md. N. Hoque, and D. Ellis.
1991b The implications of future demographic change for participation in
fishing in Texas.• North American Journal of Fisheries (forthcoming).
Murdock, S.H., R.R. Hamm, P.R. Voss, D. Fannin, and B. Pecotte.
1991c Evaluating small-area population projections.• Journal of the Ameri-
can Planning Association (forthcoming).
Murdock, S.H., S. Hwang, and R. Hamm
1992 Component methods for the estimation of total population.• In W.
Reeves, W. Serow, A. Lee, P.R. Voss, and H. Gol$mith (eds.). Basic
Methods for Preparing Small-Area PopulaHon EsHmates. Madison,
Wisconsin: Applied Population Laboratory.
Nam, C.B., and S.G. Philliber
1984 PopulaHon: A Basis OrientaHon. Englewood Cliffs, NJ: Prentice-
Hall, Inc.
Nam, C.B., and M.G. Powers
1983 The Socioeconomic Approach to Status Measurement. Houston, TX:
Cap and Gown Press. ·
Namboodiri, K., and N.M. Lalu
1971 The average of several simple regression estimates as an alterna-
tive to the multiple regression estimate in postcensal and intercen-
sal population estimation: a case study.• Rural Sociology 36:187-
194.

282
Namboodiri, K., and C.M. Suchindran
1987 Life Table TechnU,ues and Their Applications. Orlando: Harcourt,
Brace, and Jovanovich.
Nathanson, C.A., and Y.J. Kim
1989 Components of change in adolescent fertility. Demography 26:85-
98.
National Academy of Sciences
1980 Estimating Population and Income of Small Areas. Washington, DC:
National Academy Press.
National Center for Education Statistics
1989 Race/Ethnicity Trends in Degrees Conferred by Institutions of Higher
Education: 1978-79 through 1988-89. Washington, DC: U.S. Gov-
ernment Printing Office.
1975 U.S. Decennial Life Tables, State Life Tables: 1969-1971. Rockville,
MD: U.S. Department of Health, Education and Welfare.
1986 U.S. Decennial Life Tables for 1979-1981. Hyattsville, MD: U.S.
Department of Health and Human Services.
O'Hare, W.
1976 Report on a multiple regression method for making population
estimates. Demography 13:369-380.
1980 ·A note on the use of regression methods in population estimates.•
Ott, L.
1988 An Introduction to Statistical Methods and Data Analysis. Boston:
PWS-Kent Publishing Company.
Palmore, E.
1978 When can age, period and cohort effects be separated? Social
Forces 1:282-295.
Pearl, R., and L.J. Reed
1920 On the rate of growth of the population of the United States since
1790 and its mathematical representation. Washington, DC: Pro-
ceedings of the National Academy ofSciences 6:275-288.
Pedhazur, E.J.
1982 Multiple Regression in Behavioral Research: Explanation and Prediction
(2nd edition). New York: Holt, Rinehart and Winston.
Pittenger, D.
1974 •A typology of age-specific net migration rate distributions.• Journal
of the American Institute of Planners 40 (4):278-283.
1976 Projecting State and Local Population. Cambridge, MA: Ballinger
Publishing Company.

283
Pol, L.G.
1987 Business Demography: A Guide and Reference for Business Planners
and Marketers. New York: Quorum Books.
Pollard, A.H., F. Yusuf, and G.N. Pollard
1990 Demographic Techniques. New York: Pergamon Press.
Portland State University
1975 Population Projections to the Year 2000. Oregon Administration Dis-
trict. Center for Population Research and Census. Portland, OR:
Comprehensive Health Planning Association for the Metropolitan
Portland Area.
Preston, S.H.
1976 Mortality Patterns in National Populations: With Specific Reference to
Recorded Causes. New York: Academic Press.
Pursell, D.E.
1970 Improving population estimates with the use of dummy
variables.• Demography 17:341-343.
Ritchey, P.N.
1976 Explanations of migration.• In The Annual Review of Sociology II.
Palo, Alto, CA: Annual Reviews, Inc.
Rives, N.W., and W.J. Serow
1984 Introduction to Applkd Demography: Data Sources and Estimation
Techniques. Sage University series on Quantitative Applications in
the Social Sciences. Beverly Hills: Sage Publications.
Robey, B.
1985 The American People: A Timely Explonition of a Changing America and
the Impact of New Demographic Trends Around Us. New York:
Truman, Tally Books.
1989 Two hundred years and counting: the 1990 Census.• Population
Bulletin 44:1. Washington, DC: Population Reference Bureau, Inc.
Rodgers, W.L.
1982 Estimable functions of age, period, and cohort effects.• American
Sociological Review 47:77~787.
Rosenberg, H.
1968 Improving current population estimates through stratification.•
Land Economics 44:331-338.
Russell, C.
1984 The business of demographics.• Population Bulletin 39:3. Washing-
ton, DC: Population Reference Bureau, Inc.
Ryder, N.B.
1964 Notes on the concept of a population.• American Journal of Sociol-
ogy 64:447-464.
1965 The cohort as a concept in the study of social change.• American
Socological Review 30:843-861.

284
1990 What is going to happen to American fertility? Population and
Development Review 3:433-454.
Saunders, J.
1988 Basic Demographic Measures: A Practical Guide for Users. Lanham,
MD: University Press of America.
Shryock, H., and J. Siegel
1980 The Methods and Materials of Demography. Washington, DC: U.S.
Government Printing Office.
Siegel, J.S., and C.M. Taeuber
1986 Demographic perspectives on the long-lived society. Daedalus
115:77-177.
Sjaastad, L.A.
1962 The costs and returns of human migration. Journal of Political
Economy 70:580-593.
Smith, S.K.
1986 •A review and evaluation of the housing unit method of popula-
tion estimation. Journal of the American Statistical Association
394:287-296.
Smith, S.K., and T. Sincich
1991 An empirical analysis of the effect of the length of the forecast
horizon on population forecast errors. Demography 2:261-274.
Snedecor, G.W., and W.G. Cochran
1967 Statistical Methods. Sixth Edition. Ames, Iowa: Iowa State Universi-
ty Press.
Spencer, G.
1984 Projections of the population of the United States, by age, sex,
and race: 1983 to 2080. Current Population Reports. Series P-25,
No. 952. U.S. Bureau of the Census. Washington, DC: U.S.
1986 Projections of the Hispanic population: 1983 to 2080. Current
Population Reports. Series P-25, No. 796. U.S. Bureau of the Census.
1989 Projections of the population of the United States, by age, sex,
and race: 1983 to 2080. Current Population Reports. Series P-25,
Sternlieb, G., and J.W. Hughes
1986 Demographics and housing in America. Population Bulletin 41:2-
34. Washington, DC: Population Reference Bureau, Inc.
Stahura, J.M., and J.J. Sloan, m
1988 Urban stratification of places, routine activities and suburban
crime rates. Social Forces 66:1102-1118.

285
Stoto, M.A.
1983 The accuracy of population projections.• Journal of the American
Statistical Association 78:13-20.
Stoto, M.A.,and J.S. Durch
1990 National health objectives for the year 2000: the demographic
impact of health promotion and disease prevention. Paper present-
ed at the annual meeting of the Population Association of America.
Toronto, Canada.
Strong, M.A.
1987 Software for demographic research.• Population Index 2:183-199.
Swanson, D.A.
1978 An evaluation of 'ratio' and 'difference' regression methods for
estimating small, highly concentrated populations: the case of
ethnic groups.• Review of Public Data Use 6:18-27.
1980 Improving accuracy in multiple regression estimates of population
using principles from causal modeling.• Demography 17:413-427.
Swanson, D.A., and L.M. Tedrow
1984 Improving the measurement of temporal change in regression
models used for county population estimates.• Demography 21:373-
382.
Sweet, J.A.
1984 Components of change in migration and destination-propensity
rates for metropolitan and nonmetropolitan areas: 1935-1980. •
Sweet, J.A., and L.L. Bumpass
1987 American Families and Households. New York: Russell Sage Founda-
tion.
Tarver, J.D., and T.R. Black
1966 Making County Population Projections: A Detailed Explanation of a
Three-Component Method, Illustrated by Reference to Utah Counties.
Logan: Utah Agricultural Experiment Station.
Tayman, J.
1991 Population and housing estimates for microgeographic areas: a
blend of demographic and geographic information system tech-
niques. Paper presented at the annual meeting of the Population
Association of America. Washington, DC.
Teitelbaum, M.S., and J.M. Winter
1985 The Fear of Population Decline. New York: Academic Press.
U.S. Bureau of the Census
1976 Computer Programs for Demographic Analysis. Bureau of the Census.

286
1977 Gross migration by county: 1965 to 1970.• Current Population
Reports. Series P-25, No. 701. U.S. Bureau of the Census. Wash-
ington, DC: U.S. Government Printing Office.
1979 lliustrative projections of state populations by age, race, and sex:
1975 to 2000.• Current Population Reports. Series P-25, No. 796. U.S.
Bureau of the Census. Washington, DC: U.S. Government Print-
ing Office.
1984 Gross migration for counties: 1975 to 1980.• Census of Population
Supplementary Report. PC80-Sl-17. U.S. Bureau of the Census.
1989 1990 Census of Population and Housing Tabulation and Publication
Program. U.S. Bureau of the Census. Washington, DC: U.S.
1990a School Enrollment-social and economic characteristics of students:
October 1988 and 1987.• Current Population Reports. Series P-20,
1990b Household and family characteristics: March 1990 and 1989.•
Current Population Reports. P-20, No. 447. U.S. Bureau of the
Census. Washington, DC: U.S. Government Printing Office.
1990c Household wealth and asset ownership: 1988.• Current Population
Reports. Series P-70, No. 22. Household Economic Studies,
Survey of Income and Program Participation. Washington, DC:
U.S. Government Printing Office.
1991a Census Bureau releases preliminary coverage estimates from the
post-enumeration survey demographic analysis.• U.S. Census
Bureau Press Release CB91-131. Washington, DC: U.S. Bureau of
the Census.
1991b Census Bureau releases refined estimates from post enumeration
survey of 1990 census coverage.• U.S. Census Bureau Press Re-
lease CB91-221. Washington, DC: U.S. Bureau of the Census.
1991c International programs.• CFF No. 21 (Rev.). Washington, DC:
1991d Guide To Foreign Trade Statistics. U.S. Department of Commerce.
1991e 1991 Census showed gain of 14 million housing units since 1980.•
U.S. Census Bureau Press Release CB91-155. Washington, DC:
United States Bureau of Economic Analysis
1991 1990 OBERS BEA Regional Projections: Economic Activity in the
United States, by State, Economic Area, and SMSA, 1979-2040. U.S.
Department, Bureau of Economic Analysis. Washington, DC: U.S.

United States Bureau of Labor Statistics
1987 Monthly Labor Review 110:9.
1989 Monthly Labor Review 112:11.
van der Vate, B.J.
287
1988 Methods used in estimating the population of substate areas in the
United States. Paper presented at the International Symposium on
Small Areas Statistics. New Orleans, LA.
Voss, P.R., C.D. Palet, B.D. Kale, and H.C. Krebs
1992 Censal-ratio methods used to estimate total population. W.
Reeves, W. Serow, A. Lee, P.R. Voss and H. Goldsmith (eds.). In
Basic Methods for Preparing Small-Area Population Estimates. Madison,
Wisconsin: Applied Population Laboratory.
Wallman, K.K.
1988 Losing Count: The Federal Statistical System. Washington, DC:
Population Reference Bureau, Inc.
White, M.J.
1987 American Neighborhoods and Residential Differentiation. New York:
Russell Sage Foundation.
Wilson, F.D.
1988 Components of change in migration and destination-propensity
rates for metropolitan and nonmetropolitan areas: 1935-1980.

Age,
data sources for, 75-96
defined, 18
measures of, 149-152
median, 41, 117, 149
Index
pyramid, 151-152
standardization for, 158-164
trends in, 18, 38, 43-46, 93, 251
American Demographics, 103, 273
American Statistical Index, 70-71
Annual Housing Survey, 92
Arithmetic change, 120-121
Baby boom, 12, 15, 18, 28-32, 184, 251
Baby bust, 13, 18
Bibliography of Agriculture, 73-74
Birth rate,
age-specific, 31, 117, 119, 124
crude; 115-117
data sources for, 75-78, 97-98, 100-102
general, 115-118
trends in, 32
Bookkeeping equation, 13, 134, 178, 228
CENDATA, 95
Censal ratio, 184-186, 188-191
Census, 81, 92
Census Bureau. See U.S. Bureau of the Census
Census and You, 95
Census survival rates, 134-138, 225
Child-woman ratio, 124-126
Cohabitation, 21-22
Cohort
analysis of, 13
definition of, 12
Cohort-component methods
estimates, 181, 203-210
projections, 211, 223-243
Commuting, 16

290
Compact Disk-Read-only Memory (CD-ROM), 92, 106
Concepts. See Demography-concepts,
Congressional District Data Book, 76, 78
Congressional Information Service Index, 70-72
Controlling to a total, 178, 180
County and Oty Data Book, 76-78, 110
County Business Patterns, 94
Current Business Survey, 93
Current Construction Survey, 92
Current Industrial Survey, 93
Current Population Survey, 92-94
Data
compilations of, 76-79
examples of uses, 110-112
general indices
American Statistical Index, 70-71
Congressional Information Service, 70-72
Index to U.S. Government Periodicals, 70, 72
Index to International Statistics, 71-72
Monthly Catalog of U.S. Government Publications, 71
Statistical Reference Index, 71-72
media for, 108-109
principles of use, 106-109
sources, Federal, 79-100
Agency indices, 73-74
Bibliography of Agriculture, 73-74
Censuses, 80-81
Census of Agriculture, 80
Census of Construction Industries, 80
Census of Governments, 80
Census of Housing, 82
Census of Manufacturing, 80
Census of Mineral Industries, 80
Census of Population and Housing, 80-81, 83-87
Census of Retail Trade, 80
Census of Service Industries, 80
Census of Transportation, 80
Census of Wholesale Trade, 80
1990 Census, 81-92
U.S. Bureau of the Census, 73
U.S. Bureau of Economic Analysis, 83, 96-97

U.S. Bureau of Labor Statistics, 83, 97
Federal compilations, 76-78
Congressional District Atlas, 78
Congressional District Data Book, 76, 78
County and Oty Data Book, 76-78
Historical Statistics of the United States, 76-77
State and Metropolitan Area Data Book, 76, 78
Statistical Abstract of the United States, 76-77
National Center for Education Statistics, 79, 98-99
National Center for Health Statistics, 79, 97-98, 124
P-Series, 93-94
Superintendent of Documents' Oassification System, 75
surveys, 92-94
U.S. Bureau of the Census, 80-96
sources, nongovernmental (private), 103-105
sources, state, 100-102
agricultural, 100-101
economic, 100-101
education, 100-101
employment, 100-101
health, 100-101
human services, 100-102
state data center, 100, 102
state libraries, 100, 102
Data Users News, 95-96
Death. See Mortality
Death rate,
age-specific, 124, 126, 128-129, 132-133
crude, 124, 126
data sources for, 80-82, 94-95, 97-99
infant, 126, 128-129
neonatal, 126, 128-129
post neonatal, 126, 128-129
Decomposition of rates. See Rate decomposition
Demographic processes. See Fertility; Migration; Mortality
Demography, 1-8, 272-273
applied
concepts and variables in, 6-7
definition of, 6
dimensions of, 6-7
limitations affecting, 272-273
opportunities in, 269-271
291

292
trends in, 265-269
concepts of, 6-7
definition of, 4
formal, 4
social, 4
Dependency ratio, 149-150, 152
Economic-based techniques, 217-223
Economic characteristics, 7
data sources for, 75-81, 92-97, 100-105
income, 7, 25
industry, 7, 24
occupation, 7,24, 26
trends in, 24-25
Economic development, 101
Education, 7, 21, 23-24, 26, 48, 75, 98, 259
attainment, 93
characteristics, 153, 156
data sources for, 75-81, 92-94, 98-101
Grade Graduation Rate, 139-140, 153, 155
Grade Retention Rate, 139-140, 153, 155
measures of, 153, 155-156
trends in, 24
Elderly, 18, 251
Employment, 7, 24-25, 156
data sources for, 75-81, 94, 97, 100-101
definition of, 24
industry, 24-25
occupation, 24-25
trends in, 24-25
underemployment, 24
unemployment, 24
Errors,
mean absolute percent, 244-246
mean percent absolute, 245-246
mean percent error, 245-246
of closure, 92
of estimation, 244, 246-247

Estimates
accuracy of, 244-247
adjustments in, 178-181
concepts, 176-177
definition of, 176
evaluation of, 241-247
limitations of, 177-178
population-based statuses and conditions, 234, 241
principles of, 177-178
techniques for
component methods, 204-210
administrative records method, 204
component method II, 204-205
cohort-component methods, 204-210, 251
extrapolative, 182-184
arithmetic rates, 182
geometric rates, 182
exponential rates, 182
Gompertz Curve, 182-183
logistic curve, 182-183
regression-based, 196-204
ratio-correlation, 198-204, 247-250
symptomatic, 184-196
censal-ratio methods, 184-196, 199-202, 238-250
housing unit method, 186-191, 197-200
other ratio-based methods, 191-1%, 243
Ethnicity, 7, 19-21, 47, 55, 152-153, 234-235, 252-254
data sources for, 75-81, 92-94, 100-103
definition of, 19-20
measures of, 153, 252-254
trends in, 20-21, 251
Exponential change, 120, 123, 124
Families, 7, 22-23, 25, 48, 92-93
data sources for, 75-81, 92-94, 100-102, 104
measures of, 153
trends in, 22-23, 48, 59
types of, 22
Fecundity, 15
293

294
Fertility, 15, 26
definition of, 15
measures of, 30, 124-126, 127
age-specific rates, 28, 119
child-woman ratio, 124-126
crude rate, 117, 124
data sources for, 75-81, 92-94, 97-98, 100-102
general rate, 117-118, 124
total fertility rate, 124, 126-127
trends in, 28-30
Forecasts. See Projections
Foreign Trade Survey, 93
Gender. See Sex
Geographic Information System(s), 82, 270-271
Geometric change, 120, 122
Gini Coefficient, 145-148
Gompertz Curve, 182-183
Health care planning, 79-98, 266-267, 279-280
Hispanic origin. See also Spanish origin, 19-21, 47, 49-54
Historical Statistics of the United States, 76-71
Households, 7-8, 21-25, 48, 53, ~, 153
average size of, 23, 48-60
data sources for, 75-81, 92-94, 100-102
definition of, 22
measures of, 153
trends in, 28, 48, 56, 58-60
types of, 22-23
Housing Unit Method, 186-191
Immigration, 16, 28, 32, 60, 168, 263
Immigration Reform and Control Act, 260
Income,
data sources for, 75-81, 92-94, 96-97, 100-103
definition of, 25
measures of, 156, 264
trends in, 25
Index of Dissimilarity, 145-149, 152-153
Index to International Statistics, 71-72
Index to U.S. Government Periodicals, 70, 72, 74, 76,
Infant mortality rate, 26, 30, 32, 126, 128

lnnrlgration, 16-17,32,38, 134, 139, 141
International marketing, 267
Labor force, 255-258
definition of, 24
measures of, 140, 156-157, 217-223
participation rate, 140, 156-157, 217-223
age-specific, 140, 156-157
crude, 140, 156-157
general, 140, 156-157
trends in, 25, 255-257
Life expectancy, 129, 132
Life table, 129-134, 169, 171-174
definition of, 129, 132
elements of, 129-134
multi-decrement, 169, 171-174
nuptiality, 169-171
school life, 169-171
working life, 169-174
Life table survival rates, 134-138
Logistic curve, 182
Long-term care planning, 266
Lorenz Curve, 144, 147-148
Marital status, 7, 21-22, 48, 55-57, 59, 153
data sources for, 75-81, 92-94, 97, 100-102
definition of, 21
age-specific marriage rate, 153-154
crude marriage rate, 153-154
general marriage rate, 153-154
trends in, 21-22, 48, 55-57
Mean, 117
Mean absolute percent error. See Estimates and Projections
Mean percent absolute difference. See Estimates and Projections
Mean percent error. See Estimates and Projections
Median, 117
Metropolitan Statistical Area (MSA), 18
Migration, 7, 13-14, 16-17, 26, 28-32, 134, 139-141, 227-229, 260
data sources for, 75-78, 93
defination of, 16-17
295

296
measures of, 31, 134, 139-141
gross, 134, 228
net, 134, 139, 228
residual, 134, 140-141, 228
trends in, 28-32, 260
Minority. See Ethnicity and Race
Mobility. See Migration
Mode, 125
Monthly Catalog of U.S. Government Publications, 70-71
Mortality, 7, 15-16, 28, 30, 32, 66, 98, 126, 128-129, 225, 230
age-curve of, 28
data sources for, 75-81, 97-98, 100-102
definition, 15-16
measures of,
age-specific mortality (death) rate, 129
cause-specific death rate, 129
crude death rate, 126
infant mortality rate, 126, 128-129
neonatal death rate, 126, 128
post-neonatal death rate, 126, 128-129
life table. See Life table
trends in, 28, 30-31, 258
Multi-state projection models, 9, 271
National Center for Education Statistics, 73, 79, 98-99
National Center for Health Statistics, 73, 79, 97-98
Natural decrease, 28, 60
Natural increase, 28-29, 60
Neonatal mortality, 126, 128
Old-age dependency ratio, 152
Outnligration, 16, 134, 139
Population
data sources for, 75-105
defined, 11-12
change, 7, 13-14, 114
arithmetic, 120-121
doubling rate, 124
exponential, 120, 123

trends in, 26-28, 250-254
geometric, 120, 122
characteristics, 7, 17-25, 38, 42, 48, 59~, 137, 152-153, 156,
158-159, 163-164, 167-169, 171,174, 250-261
controlling the effects of, 158-174
trends in, 38-67
composition. Su Characteristics
components of change. See Fertility; Migration; Mortality
data sources. Su Data
distribution, 7, 17, 32-38
definition of, 7
measures of, 141-142, 147-149
density, 141
Gini Coefficient, 145-146, 148
Index of Dissimilarity, 145-146, 148
Lorenz Curve, 144, 147-148
potential, 141-143
trends in, 32:-38, 263
equation, 13-14, 178
estimates. See Estimates
ethnicity. See Ethnicity
fertility. Su Fertility
growth rate. See Change
projections. Su Projections
mortality. See Mortality
potential. Su Population-distribution
pyramid. Su Age
race. See Race
special population, 179
subpopulation. See Cohort
Poverty,
definition of, 25
trends in, 66.07
Projections,
accuracy of, 244-247
adjustments in, 178-179, 181
concepts in, 76-177
definition of, 177
evaluation of, 241-247
297

298
limitations of, 177-178
population-based statuses and conditions, 234, 241
principles of, 177-178
techniques for, 210
cohort-component, 223-243
economic-based, 217-223
extrapolative, 211
land-use, 214, 216-217
ratio-based, 213-215
Public Use Microdata Sample (PUMS), 80
Pyramid, 151, 152
Race. See also Ethnicity
data sources for, 76-81, 92-94
trends in, 19-21, 250-256, 262-263
Rates, 28, 32, 113, 115
crude, 28, 30, 32, 115-116
general, 28, 32, 118
decomposition of, 158, 163-169
specific, 31, 32
Ratio-correlation, 198-203
Sex, 7, 19, 38, 42, 149, 152
definition of, 19
measures of, 149, 152
pyramid, 151-152
ratio, 150, 152
trends in, 38, 42
Socioeconomic status, 7, 25-26, 60
definition of, 25
measures of, 156
trends in, 60, 261
Spanish origin. See also Hispanic origin, 19-21
Standardization, 158-163
direct, 158-163
indirect, 158-163
State and Metropolitan Area Data Book, 76, 78
Statistical Abstract of the United States, 76

Statistical Reference Index, 71-72
Superintendent of Documents Classification System, 75
Survey of Income and Program Participation (SIPP), 93
Survey of Minority-Owned Businesses, 93
Texas Almanac, 78
The U.S. Government Manual, 74-75
Topologically Integrated Geographic Encoding and
Referencing System (IlGER), 81-82
Total Fertility Rate, 126-127
U.S. Bureau of Economic Analysis, 79, 96-97
U.S. Bureau of Labor Statistics, 79, 97
U.S. Bureau of the Census, 80-100
U.S. Department of Agriculture, 73
U.S. Government Manual, The, 74-75
Vital statistics data, 76-81, 97-98, 100-101, 124
Wealth. See Economics
Youth dependency ratio, 149-150, 152
299

Steve H. Murdock, David R. Ellis - Applied Demography_ An Introduction to Basic Concepts, Methods, and Data-Routledge (2020).pdf

More Related Content

What's hot (20)

Similar to Steve H. Murdock, David R. Ellis - Applied Demography_ An Introduction to Basic Concepts, Methods, and Data-Routledge (2020).pdf (20)

Recently uploaded (20)

Steve H. Murdock, David R. Ellis - Applied Demography_ An Introduction to Basic Concepts, Methods, and Data-Routledge (2020).pdf