Principal steps in a Statistical Enquiry

Lecture Series on
Statistics
No. Bio-Stat_2
Date – 27.07.2008

“THE PRINCIPAL STEPS IN A
STATISTICAL ENQUIRY"
By
Dr. Bijaya Bhusan Nanda,
M. Sc (Gold Medalist) Ph. D. (Stat.)
Topper Orissa Statistics & Economics Services, 1988
bijayabnanda@yahoo.com

CONTENTS
Statistical Thinking in Empirical Analysis

Principal Steps in a Statistical Enquiry

OBJECTIVES
 The trainees will adopt Statistical Thinking to
.
problem solving in Medical Research.

 They will be able to plan Research in a scientific
way.

STATISTICAL THINKING
DIMENSION – 1: Investigative Cycle (PPDAC Model)
•Interpretation
•Conclusion •Grasping System dynamics
•New Ideas •Defining Problems
•Communication CONCLUSION PROBLEM

•Data Exploration
•Planned Analysis ANALYSIS PLAN Planning
•Unplanned Analysis •Measurement System
•Hypothesis generation •Sampling design
•Data Management
•Piloting & Analysis
DATA

•Data Collection
•Data Cleaning
•Data Management MacKay & Oldford, 1994

DIMENSION 2:Types of thinking
GENERAL TYPES
 Strategic
 Planning, anticipating problems
 Awareness of practical constraints
 Seeking Explanations
 Modeling
 Construction followed by use
 Applying Techniques
 following precedents
 recognition and use of archetypes
 Use of problem solving tools

TYPES FUNDAMENTALTO STATISTICAL
THINKING
 Recognition of need for data
 Transnumeration
(Changing representation to engender understanding)
 capturing “measures” from real system
 Changing data representation
 Communicating messages in data
 Consideration of variation
 Noticing and acknowledging
 Measuring and modeling for the purpose of prediction,
explanation, or control
 Explaining and dealing with
 Investigative strategies
 Reasoning with statistical model
 Integrating the statistical and contextual
 Information, knowledge, conceptions

DIMENSION – 3: The Interrogative Cycle

Decide what to:
•believe Imagine possibilities for:
•Continue to entertain •Plans of attack
•discard JUDGE GENERATE •Explanations/ modes
•Information requirement

Check against
Reference points: CRITICISE SEEK Information and ideas
•Internal •internally
•external •externally

INTERPRET
•Read/hear/see
•translate
•Internally summarise
•Compare
•Connect

DIMENSION 4: Disposition
(Arrangement / Person’s natural
Qualities of mind and Character)
 Scepticism
 Imagination
 Curiosity and awareness
• Observant, noticing
 Openness
• To ideas that challenge preconceptions
 A propensity to seek deeper meaning
 Being logical
 Engagement
 Perseverance

Pertinent methodological questions

1. What are the objectives?
2. What is the coverage?
3. What type of information required?
4. What are the methods to collect data?
5. Whether census or sampling?
6. If a sample survey- the design?

Some pertinent methodological
questions
7. What are the potential sources of errors
and possible precautions?
8. What are the steps to conduct the field
work efficiently?
9. What are the methods to process and
analyses the collected data?
10.What are the points to be kept in mind
while writing the report and presenting
the results of the survey?
11.What are the problems of accuracy,
errors and approximations?

A STASTICAL ENQUIRY
 It is with a Predefined purpose and dealing with
collection of data in a systematic manner.
 Data collected from such enquiry are called
statistical data.
 It may be Descriptive Field Surveys or Analytical
Experimental study under controlled conditions.
 Descriptive Field Survey:
 Cross Sectional Study/ Prevalence Study:
Simplest form of an observational Study.
 Longitudinal study: Observations are repeated
in the same population over a prolonged period
of time by means of follow up examinations.
 Analytical studies: Case Control Study and Cohort
Study

Planning a Statistical Enquiry
1. What are the objectives of the survey?
Objectives - clear and unambiguous.

Possible uses of final results to be
expected and the desired degree of
accuracy be discussed at the beginning.

Lack of clear cut objectives might
undermine the value of the final results.

and the manipulation of data at the end
would not solve the issue in questions
and eradicate the inherent defects.

2. Population to be covered:

 What is the population to be studied?

• Target population
• Sampled population

 What are the units from which data is
to be collected?

3. Data to be Collected

Should be decided keeping in view the
objectives of the study.
Relevant data should not be omitted.
Irrelevant data should not be
collected.

Primary & Secondary data
 Primary data: Data Collected by the investigator
through direct observation, mailed
questionnaire or interview method from the field
or conducting experiments.

 Secondary data: Ready made data colleted from
published or unpublished form in Hospital
Records
 Involves less time and money
 It may not always be adequate and representative
enough to serve the required purposes and may also
be outdated.
 Further, secondary data may have unknown degree
of accuracy

The primary data are usually in a raw and bulky
form and do not posses the simple look of the
secondary data, which are already in processed
form. This when processed and put in an appropriate
classified and tabulated form by a recognized body,
becomes secondary data for future users.

4. Designing questionnaire / schedule

 Questionnaire: To be filled in by
the respondent.

 Schedule: - To be filled in by the
interviewer

Designing questionnaire / schedule
 This requires:-
 Skill,
 special techniques,
 Familiarity with the subject matter
 Question content:
 Respondents are likely to have knowledge
to answer them.
 Large number of questions to the
respondents should be avoided.
 The questions should be such that one is
able to extract brief answers.
 The questionnaire should have some cross-
questions to check the authenticity of the
information supplied

Question wording
 The question should be specific, simple and
unambiguous. Vague, presuming and
hypothetical questions should be avoided. If
necessary, the questionnaire should contain some
leading questions. The questions should be such
that it is capable of being answered without
prejudice.
Open and pre-coded questions
 In an open question the respondent has the
freedom in choice of answers. But in case of pre-
coded or multiple choice questions the
respondent is asked to choose one from the
limited number of given answers. Open questions
are, often, found difficult to be coded. Pre-coded
questions restrict the freedom of thought of the
respondents.

Question order

 Ordering of the questions is an important
part of the construction of questionnaire.
The ordering systematizes the thought
process of the interviewer and as a result
reduces the refusal rate of the survey
based on interviewing.

 Suitable instruction for filling-up the
same is a must.

5. Approach to data collection

 Complete enumeration (Census)

 or Sample study?

6. Selection of a proper sampling
design
 Sampling units
 Reference unit
 Sampling frame:-List of all the
sampling units in the population
under study?
 Minimum sample size required
 Sampling design

Minimum Sample size
 This will depend upon
• Aim, nature, scope of the study,
variability in the population and on the
expected results.
 Standard procedure for different
situations
1. One sample situation
 Estimating a population proportion with
specified absolute/relative precision

2. Two sample situation
 Estimating the difference between two
population proportions
 Hypothesis tests for two population
proportions Case-control studies
3. Case-control studies
 Estimating an odds ratio
 Hypothesis tests for an odds ratio
4. Cohort studies
 Estimating a relative risk
 Hypothesis test for a relative risk

5. Lot quality assurance sampling
 Accepting a population prevalence as
not exceeding a specified value
 Decision rule for “rejecting a lot”

6. Incidence-rate studies
 Estimating an incidence rate

 Hypothesis tests for an incidence rate

 Hypothesis tests for two incidence
rates in follow-up (cohort) studies

Sampling Design
 Sampling design will depend upon
• Objectives, scope and coverage of the
study, nature and size of the population,
hypothesis to be tested
 Different scheme of sampling
• Probability sampling
• Non-probability sampling

7. Different methods of data
collection
Personal observations,
Mailed questionnaires,
Interviewing,
Documental sources, etc.
A choice to be made depending
upon: the type of data to be
collected and type of population to
be covered.

 Mailed Survey
• Less costly
• Non-response may be high
• Practicable for educated respondents
 Interview Method
• More cost
• Less non-response
• Practicable for educated and
uneducated both
• Concept and definition can be explained
to the respondents.

 Observation Method

• Specify the method of measurement

• Specify unit of measurement

• Provide measuring equipments and
instruments

8. Dealing with non-response

 Due to non-availability of the respondent
 Respondent may fail to give the data
when contacted.
 Respondent may refuse to give
information.
 Non response tends to change the results.
 Procedure to be devised to deal with it.

• Call-back procedures,

• Substitution

9. Errors
 Any statistical survey is not free from
errors:
 Errors may be on account of
• Sampling: Which can be kept under
control by following a suitable sampling
design
 Faulty selection of the sample
 Substitution
 Faulty demarcation of the sampling unit
 Improper choice of statistics for estimating the
population parameter.

9. Errors- Contd.
 Non-sampling errors
• Faulty Planning or Definitions
• Response Errors
 May be accidental, Prestige bias, Self-interest,
Bias due to interview, Failure of respondent’s
memory
• Non-response bias: Full information not
obtained of all sample units
• Error in coverage: Objectives not clearly
stated
• Compiling Errors: Various operations in
data processing
• Publication errors: Committed during
presentation and printing of tabulated
results

9. Organisation of field work

 Reliable field work is a precondition
for the success of the statistical
enquiry.
 Organizational set-up
• Selection of Team leader
• Selection of Investigators
• System of supervision
• Logistic arrangements: Stay,
Transport & Communication,
printing of schedules, supply of
appropriate stationeries

9. Organisation of field work

 Training
• Orientation of the Investigators
with the;
 instruments,
 Levelsand units of data collection
 concepts and definitions of the field of
information
• How to deal with non-response

10. Pre-testing and Pilot survey

Pre-testing:
Trying out the questionnaire on a small
scale.

The pre-testing of a questionnaire
brings out the drawbacks of the
questionnaire as to question order,
content and wording and enables the
survey specialist to reach a decision
whether to add certain more questions
or delete some already incorporated. It
studies possible reactions and
difficulties of the respondents to
answer certain question and remodel
accordingly.

Pilot Survey
 Pilot survey is like theatrical dress
rehearsal
 This is to try out the entire survey on a
limited scale to know:-
the nature of the population
 efficacy of the sampling design
 deficiency in the planning
 This enables the researchers to correct
deficiency before the main survey

11. Processing of the Data
 Data Validation or Editing of data
 Coding-To put the results in
quantitative form by assigning
numeric codes
 Classification and tabulation of data
 Presentation of data
 Analysis of data

12. Reporting and Conclusions
 Finally, the analysis should be interpreted
in the form of a report incorporating:-
 Detailed statement of the different stages
of the survey,
 technical aspects of the design, viz., the
types of the estimators used along with
the amount of error to be expected in the
most important estimates, and

 Presentation of the results.

13.Information gained for future surveys
 Any completed survey helps to understand the
nature of the data in terms of means
(average), standard deviations (variability),
cost involved in obtaining the data.

 Any completed survey provides lesson for
designing future surveys in recognizing and
rectifying the mistakes committed in the
execution of the survey.

 Thereby it is a learning point for improved
planning and execution of the future surveys.

REFERENCES
 Biostatistics – A foundation for Analysis in the Health
Sciences: Wayne W. Daniel, Seventh Edition, Wiley
Students Analysis.

 A First Course in Statistics with Application: A.K. P. C
Swain, Kalyani Publishers.

 Fundamentals of Applied Statistics, SC Gupta and V. K.
Kapoor
 Sample Size determination in Health Studies: A Practical
Manual, SK Lwanga and S Lemeshow, WHO,

 Parks Text book of Preventive and social medicines

Principal steps in a Statistical Enquiry

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Principal steps in a Statistical Enquiry (20)

More from Southern Range, Berhampur, Odisha (20)

Recently uploaded (20)

Principal steps in a Statistical Enquiry