Nvasa data project guide v0907

GGuuiiddeebbooookk ffoorr
SSttuuddeenntt PPrroojjeeccttss iinn
DDaattaa AAnnaallyyssiiss
produced by
Southern Nevada Regional Professional
Development Program
and
Nevada Chapter of the
American Statistical Association
Principal Authors:
Maria Dufek, Jeffers ES, Las Vegas
David Thiel, SNRPDP
September 2006
(rev. 09/07)

Guidebook for Student Projects in Data Analysis 1
by SNRPDP and NV-ASA September 2006 (rev. 09/07)
Guidebook for Student Projects in Data Analysis
Produced by: Southern Nevada Regional Professional Development Program
and Nevada Chapter of the American Statistical Association
Principal authors: Maria Dufek, Jeffers ES, Las Vegas and David Thiel, SNRPDP
Table of Contents
Chapter I: Asking the Question............................................1
Chapter II: Collecting the Data............................................4
Chapter III: Organizing and Analyzing the Data.................9
Chapter IV: Completing the Project ..................................12
Chapter V: Assessing the Project.......................................14
Resources and References..................................................19
Chapter I: Asking the Question
What is Asking the Question?
Probably the most difficult aspect of doing a project in data analysis is determining its
subject. Who is to be studied? What about them interests us? Why? All of these are
important in doing a study. One statistics textbook calls these the “W’s.” It is what the
authors of this document refer to as Asking the Question.
Asking the Question means determining what we are interested in studying about a
population and why. It may be that we want to know the preferences of music by students in
our school because we are going to hold a dance. We might like to garden and are curious
about how plant growth responds to different amounts of sunlight. Asking the Question does
not refer to the specific question that will be asked of respondents in a survey. That is,
“What is your favorite type of music?” is not what is meant by Asking the Question.
Note: When we write “who is to be studied,” the who are not necessarily people. The who
could be bean sprouts or vehicles. What refers to the information of interest. In the case of
bean sprouts, what could be how tall they are after several weeks. If we’re studying vehicles,
what could be their type, color, gas mileage, or all three.
Deterministic Answers vs. Data that Vary
Students must recognize that a data analysis question is based on data that vary from
individual to individual. “How tall is Maria?” and “What is Dave’s favorite style of music?”
are not data analysis questions. The answers are a single height and a single style. However,
“How tall are students in my class?” and “What are my classmates’ favorite styles of music?”
are data analysis questions. Height varies from student to student, as does music preference.
The anticipation of variation is the key to asking a question in data analysis. After all,
statistics is the study of variation. Heights vary from student to student, and their tastes in
music are not all identical. This is variation.
So what do we want to know? About whom? Why? These are often difficult for students to
generate. The process can be made easier by teaching students to “notice” and “wonder.”

Noticing and Wondering
Students need to be given opportunities to observe their world using all of their senses (or as
many as possible!) Use notebooks to have students record what they notice about their
classroom. Have them work independently writing, illustrating and describing what they
notice. This may take the form of an open-ended noticing or may be focused on specific
criteria. These can be charted and shared.
From these noticings the wonderings will come. It may need to begin with Teacher Talk—
“Dave, you noticed that the desks are arranged in 6 rows of 4 desks in each. You also noticed
that we are really crowded in here. I’m wondering if there is another way to arrange the
desks to give us more room?”
“Sara, you noticed that we have 3 boys and 19 girls in our class. I’m wondering if the other
3rd
grade classes have more girls than boys.”
“Manuel noticed that the teacher has pictures of dogs and horses on her desk. I’m wondering
if those are her pets and what pets the others in the class might have.”
Encourage the students to record their wonderings. Some may not be data analysis questions.
You will need to guide the work from here. A focused noticing and wondering—“What do
you notice that can be measured or counted?” can bring up the awareness level that some
things have variability and some do not. If a student wonders how many chairs and desks are
in each classroom, that is a curiosity question that may lead to some counting experiences. If
the student wonders if all classrooms have the same area of floor space that may lead to some
measuring and comparing. The question is, “Why do you want to know about chairs?” Is
there a reason to know that? Does the number of chairs tell you anything else about the
class? What is it about floor space that interests you?
Having students do many forms of noticing and wondering generates interest in, and
curiosity about, a variety of topics.
Formulating Grade-level Appropriate Questions
When students take on a project, it’s important that they consider a question that is
appropriate for their grade level. For instance, let’s take the case of favorite music. Simply
surveying students in a class and asking each what is his or her favorite style of music is a
fairly simple, low-level study. This would be appropriate for students in the primary grades.
Intermediate students may wish to not only know what students’ favorites are, but also if
students like more than one style of music. They may also compare the music tastes of
students of different ages, or classes, or even schools. This turns up the complexity a bit by
making comparisons and looking for relationships.
Older students—middle and high school—should be exploring several (or many) aspects of
music preference. Do people who prefer rock music also like rap? Do they tend to dislike
country? Do they like jazz, but not classical? Is there a connection between music
preference and the rhythm of the beat? The possibilities are nearly endless, and are limited
only by the students’ imaginations, noticings, and wonderings.
Let’s look at another example where one might do a study that does not involve surveys or
people—plant growth. Primary students might be interested in how light affects plant
growth. They could plant two flowers, beans, or some other flora in separate pots, place one

near a window, and place the other on the side of the room opposite the window. After some
period of time, they could compare growth.
Students in intermediate grades might do the same, but with multiple plants. They would
average the growth, consider its variability, and even compare it at multiple points in time.
As students get into middle and high school, they could consider multiple settings with
various degrees of “lightness,” using multiple plants in each case. They might also consider
other factors that could come into play, like soil moisture, fertilization, etc.
Questions Based on the Type of Data
Data comes in two flavors: categorical (qualitative) and numerical (quantitative).
Categorical data describes a quality about the individuals or things being studied. Favorite
style of music is categorical data: rock, jazz, country, etc. The type or color of car a teacher
drives is categorical information.
Numerical data describes a quantity about the individuals or things being studied. How
many hours each day students listen to their favorite music is numerical. The height of a
bean sprout plant after three weeks is numerical, as is the number of cars a teacher owns.
Generally, if one had to count or measure to gather the data, it is numerical.
The level of sophistication of a project is often a function of whether the data collected is
categorical or numerical. It can be difficult, but not impossible, to formulate erudite
questions if the data to be collected is only categorical. By extending a categorical-inspired
project to include numerical data adds a level of complexity to the project and can make it
more interesting. Consider the following situations that begin with a question that generates
categorical data and are broadened to include numerical data.
• What activities do students like to do in their free time? How much time do they
spend each week on them? How much time is spent doing the activity in a single
session?
• What are students’ favorite restaurants? How much do they spend each time they eat
at those restaurants?
• What kinds of pets do students own? How many pets do students own? How large
are the pets?
• What varieties of fruit do students like to eat? Which do they dislike? What are the
levels of sugar, acid, etc. in various fruits?
• Do students wash their hands before/after ________? How often do students wash
their hands each day? How long do students spend washing their hands?
Even categorical questions can be extended by using other categorical questions, to get more
detailed information, or look for relationships among the different categories.
• What styles of potatoes do people like? If they like French fries, which fast food
chain is the best? If they like baked, what toppings do they like?
• What varieties of fruit do students like to eat? Which do they dislike? Does it
depend on their texture? On their peel/rind?

• What kinds of pets do students own? Do the students live in apartments or houses?
Helping students learn to formulate a “question” takes some time, but having them notice and
wonder makes things go easier. As students try to determine a subject for their project, have
them consider the following question stems. These are just a start. They generally become
more sophisticated as one moves down the list.
• What are…favorite …?
• How often/how many/how much…?
• Does…differ between…?
• What are the similarities and differences between…?
• What is the relationship between…?
• Is there a trend…?
• How would one predict…?
Chapter II: Collecting the Data
What is Collecting the Data?
Once students have decided upon a question, the next step is Collecting the Data to answer
the question. What method will be used to collect the data? How much data will be
collected? Are there any concerns that will require certain techniques? Each of these must
be considered in the data collection process. Usually, the question will dictate what method
of data collection is required.
Data Collection Methods
Data collection methods can be divided into three basic categories: surveys, observational
studies, and experiments.
Surveys. A survey is a method where information is gathered from individuals about
habits, preferences, opinions, attitudes, etc. The survey may be a brief series of orally-
posed questions, like a poll, or may last longer and take the form of an interview. Paper
or electronic questionnaires may be used to allow more time for respondents to answer,
access more potential respondents, or simplify the data collection process. (To be
precise, a survey is actually a specific form of the next category, the observational study.)
Observational Studies. An observational study is the method by which a researcher
tries to establish relationships between variables by making observations of the target
population. Data may come from direct observations of the individuals in the population,
or may come from examinations of records in books, almanacs, databases, or the Internet.
In an observational study, the researcher does nothing to influence the individuals being
studied. When doing an observational study, the data already exists; it simply needs to be
gathered.
Experiments. An experiment is a study where the researcher deliberately influences
certain events and observes the reaction to them. Its goal is to establish a cause-effect

relationship between two or more variables. When doing an experiment, the data does
not exist and needs to be created.
Experiments also include simulations, a study where actual conditions are imitated using
chance outcomes based on predetermined models. An example of a simulation would be
to explore variation in the percentage of male newborns in a hospital by using a coin toss
to determine gender.
To understand the difference between observational studies and experiments, consider the
following scenario. Students may notice that the bean sprouts in each of several classrooms
are of different heights. They also notice that the amount of sunlight reaching the plants is
different in each room. The students could gather data about plant height and sunlight, then
attempt to find a relationship between the two. This is an observational study. In an
experiment, the students would deliberately expose plants to various degrees of sunlight, and
then record how plant growth was influence by it. In the first case, the data was there, the
students just had to record it. In the second case, they actually created the data by
influencing certain conditions.
The Quantity of Data to be Collected
When collecting data for a survey or observational study, there are two terms that apply to
how much data is collected: census and sample. A census is a study where data is collected
from all individuals in a population. A sample studies only part of the population.
If a sample is done, because it is impossible or impractical to do a census, the issue becomes
how large of a sample is needed. Larger samples are preferable to small ones, as the
researcher is more likely to get a clear picture of the population’s characteristics. However,
there is no magic number as to how many individuals should be in a sample. It depends on
the complexity of the study and how accessible the individuals in the population are.
Considerations in Sampling
If a census cannot be done, then a sample must be taken. A sample must be as representative
of the population as possible. The group of individuals that has the potential to be in a
sample must reflect the population. If we were interested in the music preferences of
students in our school, it would not make sense to survey only those in our classroom, only
our grade, or only the boys.
Those chosen to be in the sample must be selected as impartially as possible. People have
natural biases that may guide them to include or exclude certain individuals from a sample,
even unconsciously. To overcome this, and to increase the chance of a representative
sample, individuals should be selected randomly. The use of chance devices—slips of paper
in a hat, dice, spinners, and random number tables—allows for selection of individuals in an
unbiased manner.
Considerations in Surveys
Often when surveys are done, unexpected things occur that hamper exploring the data or
drawing conclusions from it. A datum may be missing, unreadable, or so unusual that it
makes one wonder if it were recorded correctly. Results may be much different from what
was anticipated. An individual chosen to be in the sample or census may have been missed
or surveyed multiple times. Variables that were not considered before a study was done may

now be of interest to us when it is over. These issues can usually be addressed by
maintaining detailed records about who provided what information, in the event a follow-up
survey is needed.
Another option is to do pilot studies. Pilot studies are usually small surveys, observational
studies, or experiments done to establish what types of results would be obtained in the main
study, to become aware of other variables of interest, and to work out problems in the study’s
design.
When conducting surveys, how questions are asked is a very important consideration.
Should the question be open-ended or closed-ended? Is the question understandable? Is the
question being posed in a manner that does not tend to elicit a particular response? If a
question is not comprehensible, if it tends to sway respondents to a certain answer or
answers, or if it is so open-ended as to result in a different response from every respondent,
then drawing conclusions from the data may be difficult or impossible.
Considerations in Experiments
Just as random selection is important in samples, randomization is important in experiments.
In a study of how sunlight affects plant growth, the plants available for the experiment must
be randomly assigned to the various levels of sunlight. The chance assignment filters out
problems that may be hidden from us, the plants’ health for example.
The number of subjects in the experiment is a key consideration. If one has a single plant in
each of the different levels of sunlight, or even two or three, it becomes more difficult to
determine if differences in growth are actually due to the differences in light or some other
variable. The more subjects involved in the experiment, the more we can attribute
differences between groups to the variables under study.
The final consideration is the control of other variables. Even though our plants may be
exposed to different amounts of sunlight, other variables should be controlled as much as
possible. That is, experimental conditions should be as identical as possible for all plants,
except for the sunlight. All plants should receive the same amount of water and food at the
same time, artificial lighting should be identical, etc. Anything that could potentially affect
growth should be controlled.
Designing for Differences
Asking the Question is about anticipating variation. Collecting the Data is about
acknowledging that variation. As students progress in their understanding of data analysis,
their studies should be designed to expose those differences. The principles described above
assist in doing that.
An Example of a Survey
Question: Do students at our middle school who prefer one style of music tend to like/dislike
certain other styles of music?
The population of interest is all 1800± students at the school. It would be nearly impossible
to do a census, so a survey using random sampling will be done. A sample size of 300 is
desired, with 100 from each of the 6th
, 7th
, and 8th
grades. The teachers of the Math
Department have agreed to help in accessing the respondents.

There are 20 math classes in each grade, so five students will be sampled in each class to
obtain the desired 100 students from each. Each teacher has provided a count of the number
of students in each class. Five students are chosen at random using the following as an
example:
If the class count is 30, the numbers 1 to 30 are written on slips of paper and placed in
a hat. Five of the slips are drawn, at random, and the numbers on those slips
recorded. Those numbers identify the potential respondents by the order they appear
in the teacher’s roll book. If the number drawn is 5, then one of the potential
respondents will be the fifth student listed on the teacher’s class list.
The teacher will administer the survey questionnaire to the five students during the last
minute of class on November 1. If the potential respondent is absent, the teacher will attempt
it only once again on November 4. The questionnaires will be collected at the end of the day,
November 4.
The questionnaire appears as follows:
ID: 823605
(1) What is your favorite style of music? (Check only ONE)
___ country ___ classical ___ hip-hop ___ rap ___ rock ___ metal
___ jazz ___ Gregorian chants ___ easy listening ___ blues
________________ other (write-in)
(2) For each of the following styles of music, write a “D” in the space if you dislike that style
of music or write an “L” in the space if you like that style of music. Please complete all
spaces with “D” or “L.”
___ country ___ classical ___ hip-hop ___ rap ___ rock ___ metal
___ jazz ___ Gregorian chants ___ easy listening ___ blues
The first three digits of the ID is the teacher’s room number, the fourth is the class period,
and the last two identify the randomly selected student by position on the class list. If a
follow-up is needed, the student can be identified.
An example of the raw data from the survey, before analysis, is given below:
ID
favorite
country
classical
hip-hop
rap
rock
metal
jazz
G.chant
easylist.
blues
823605 rap D D L L L D L D D L
823612 jazz D L D D L D L D L L
823613 rock L L L D L L L D D D
823620 rap D L L L L D D L L L
823647 reggae D D D D D D D D D L

An Example of an Experiment
Question: Does the amount of sunlight a bean plant receives affect its growth rate?
The process begins by planting several dozen bean seeds at equal depth in identical pots and
placing them under a grow light in the multi-purpose room. During this time, all containers
will be covered, and receive an equal amount of water each morning. The pots will be placed
to receive equal amounts of light. After several days, we will “harvest” the 16 plants that
have heights closest to 5 cm in height.
The experiment will take place in the school’s quad area. It is well-shielded from wind, but
is exposed to open air and sunlight. Plants will be placed next to the walls of the quad,
halfway down each wall. The area next to the south wall receives no direct sunlight during
the day. The north wall receives several hours of direct sunlight. The east and west walls
both receive some direct sunlight, but not as much as the north wall.
The 16 chosen plants will be numbered from one to 16 and their heights to tenths of
centimeters recorded. Slips of paper numbered from 1–16 will be placed in a hat and mixed.
Four slips will be drawn and the plants with the same numbers as those slips will be placed
on the north wall. Four more will be drawn for the west wall and the east wall. The
remaining four plants will be placed on the south wall.
The plants will receive equal amounts of water (about 1 oz.) each morning before school.
The teacher will water them on weekends. Every Monday, Wednesday, and Friday, the
plants’ heights will be measured to the nearest tenth of a centimeter. The experiment will run
for three weeks, at which time the plants’ heights will be compared.
An example of the raw data from the experiment, before analysis, is given below:
Plant Wall Apr1 Apr3 Apr5 Apr8 Apr10 Apr12 Apr15 Apr17 Apr19 Apr22
1 E 4.7 6.5 8.3 11.0 12.8 14.4 17.4 20.0 22.8 26.1
2 S 5.2 6.2 7.6 9.1 9.5 10.5 11.7 13.3 14.3 15.8
3 S 5.0 5.8 6.8 8.0 9.2 10.2 11.7 12.7 14.1 16.5
4 W 5.6 7.4 8.4 10.5 12.3 14.1 16.2 18.2 19.6 22.0
5 N 4.9 7.1 9.7 12.7 15.5 18.5 21.8 24.4 26.8 30.7
6 W 4.9 6.1 7.9 10.9 12.3 14.1 16.5 18.1 18.9 21.3
7 N 5.1 7.7 9.7 13.6 16.0 18.4 22.0 24.6 26.8 30.1
8 W 4.8 6.4 8.0 9.8 11.8 14.2 16.9 18.3 19.7 22.1
9 E 5.1 7.3 9.7 13.0 15.4 17.4 20.1 22.3 24.3 27.6
10 N 5.0 7.4 9.6 13.2 15.4 17.6 20.6 22.8 25.4 29.0
11 S 5.1 6.3 7.5 9.9 11.5 12.7 14.8 15.6 17.0 18.2
12 S 4.7 6.1 7.1 8.3 9.1 10.3 11.8 13.2 14.6 16.7
13 W 5.1 6.1 7.9 10.3 12.1 13.7 16.4 17.8 19.6 22.9
14 N 5.1 6.9 9.3 12.9 15.1 17.7 21.0 23.6 26.0 29.6
15 E 5.1 6.9 8.5 11.8 13.4 15.0 17.4 19.4 21.0 24.3
16 E 4.7 6.9 8.9 12.5 14.7 16.9 19.9 21.5 23.5 26.5

Chapter III: Organizing and Analyzing the Data
What is Organizing the Data?
Once data has been collected, it must be organized into a form from which patterns can be
recognized, conclusions made, and inferences drawn. This is the heart of the poster
competition—creating a graphical display that leads the viewer to these goals. The structure
of such displays depends in large part on the question asked and the type of data collected,
and in some measure the creativity of the author. Just as the study should be designed to
expose differences (variation) within and between groups, graphical displays should reveal
those differences.
Forms of Data Display—Categorical Data
The simplest form of categorical data display is a frequency table.
A frequency table simply counts how many times each category
has occurred and displays it in tabular form. For example, a
student may ask respondents to identify their favorite type of
music. The responses range from classical to rap. A frequency
table may look like the one shown.
If an association between two categorical variables
is being investigated, then a two-way frequency
table can be made. Each cell in the table shows
the frequencies of combinations of categories. For
instance, perhaps students are wondering if people
who like plain vanilla ice cream prefer other plain
foods, like cheese pizza. The table shows how
preferred pizza and ice cream styles are related.
Often, a table is insufficient to get a clear picture of the distribution of data, or see patterns or
associations. What is called for is a “picture” of the data. The most common pictures of
categorical data are circle graphs (or pie graphs) and bar
graphs.
Circle graphs can show how a whole group is broken into
smaller parts. The circle graph shown represents the music
data from above.
Bar graphs show similar information to circle graphs. While
comparing part to whole is harder, sometimes seeing
differences between parts is
easier. Multiple-bar graphs
can be used to compare
groups by breaking data into
sub-groups. The music data
is shown as a whole group
and split by gender.
Type of Music Frequency
Classical 3
Country 6
Rap 4
Rock 9
Ice Cream
Vanilla Chocolate Total
Cheese 8 2 10
Pepperoni 3 6 9
Pizza
Total 11 8 19
Country
Rap
Rock
Classical
0
2
4
6
8
10
Classical Country Rap Rock
Type of Music
Frequency
0% 20% 40% 60%
Classical
Country
Rap
Rock
TypeofMusic
Percentage
Girls
Boys

Note: when sub-groups are different sizes—there were 14
boys and 9 girls in the favorite music survey—it is often
better to graph the data as percentages than frequencies.
One can compare how categories are distributed between two
groups using circle graphs. As an alternative, a stacked bar graph
is equivalent to comparing multiple circle graphs. Such a graph is
shown using the ice cream/pizza data from the previous page.
Forms of Data Display—Numerical Data
Displaying numerical data usually takes the form of a picture
based on a number line. The simplest form of numerical
display is the line plot (or dotplot). Each observation is
represented by a dot (or some other symbol) and like
observations are stacked on a number line. For example, a
dotplot of the heights of plants measured to the nearest
millimeter is shown.
When data is much more spread out, or takes on many different
values, it may be done as a stem-and-leaf plot or as a histogram.
A stem-and-leaf plot shows the actual values of the data using
consecutive place values. In the stem-and-leaf plot shown, the
heights of fifteen plants range from 33 mm to 69 mm. It is
important that the leaves on each stem (of the plot, not the actual
plant) have equal spacing.
A histogram shows the distribution of data in intervals, but
the individual values cannot be determined from the display,
like in a stem-and-leaf plot. It is preferable to a stem-and-
leaf plot for very large sets of data. A histogram is shown
that displays the same data as the stem-and-leaf plot.
A box-and-whisker plot (or boxplot) is a form of graph that
divides a set of ordered data into four equally-sized groups,
then visually shows the ranges of those groups. The spread
of the lowest and highest fourths are shown with whiskers;
the 2nd
and 3rd
fourths are shown with boxes. It provides a
quick visual representation of the center, spread, and to some
extent, the shape of the data. The boxplot shown is of the
same data as the dotplot above.
The power of the boxplot is in comparing the center and
spread of multiple groups. The parallel boxplot shown
displays a comparison between the heights of plants on two
consecutive measuring days.
When numerical data is gathered over a period of time, it can
best be displayed on a time plot. In the time plot shown, the
height of a particular plant has been recorded regularly over
several weeks. The graph shows the trend of its growth.
3|349
4|01137
5|244
6|0139
Key: 3|4 represents
34 mm
0
10
20
30
3/29
4/5
4/12
4/19
4/26
Date
Height(cm)
10 15 20 25 30 35
Height (cm)
4/194/22
0%
20%
40%
60%
80%
100%
Vanilla
Chocolate
Ice Cream
Percentage
Pepperoni
Cheese
46 47 48 49 50 51 52 53 54 55 56 57
Height (mm)
46 47 48 49 50 51 52 53 54 55 56 57
Height (mm)

When an association between two numerical variables is
being explored, the type of graph most commonly used is a
scatterplot. Each observation has two values, which are
plotted as an ordered pair on a coordinate plane. For
example, one may be wondering if there is a relationship
between the amount of sunlight a plant gets each day and its
growth over a three-week period.
Summary Measures
Summary measures are single numerical values that are used to describe the center or spread
of numerical data. The most commonly used measures of center are mean and median.
Frequently used measures of spread are the range, interquartile range, and standard
deviation. When examining bivariate numerical data, correlation and the slope and intercept
of a fitted line may be used. Which measures are used is dependent on the nature of the data,
its shape, and whether or not extreme values (outliers) are present.
Summary measures like the mean and median are NOT used with
categorical data. It is common to see students take data as shown
in the table, add the frequencies, and conclude that the mean is
5.5. The raw data is categorical: classical, country, rap, and rock.
One cannot do arithmetic with categories and the value 5.5 makes
no sense in this context. The same is true for the median, range, and all other measures of
numerical data.
Statistical Inference
High school students in honors or Advanced Placement statistics courses may formulate a
question that can be answered using inferential procedures (t-tests, chi-square tests, analysis
of variance, etc.). This should be encouraged and the results of inferential procedures should
be included in the presentation of the project. The timing of project deadlines may require
some out-of-class instruction on inferential procedures, as these tend to be taught in the
spring.
Final Notes about Graphical Displays and Summary Measures
Any graphs and measures students choose to provide in their analysis should be germane to
the question asked. Graphs should not be made just for the sake of making a graph.
Additionally, multiple representations of the same data need not be provided if the additional
graphics do not give any more information. That is, a stem-and-leaf plot and a histogram of
the same data are redundant, as potentially are a circle graph and a bar graph.
The displays and summaries outlined here are only some of the possible ways to organize
and analyze data. Additionally, only brief summaries of the different methods have been
provided. See the appendix for resources that provide more detail in graphical display and
data analysis.
Type of Music Frequency
Classical 3
Country 6
Rap 4
Rock 9
5
10
15
20
25
30
1 2 3 4 5 6 7
Sunlight (hr)
Growth(cm)

Chapter IV: Completing the Project
How Does One Complete the Project?
The purpose of a project in data analysis is to answer a question by the collection and
analysis of data. It is in a sense, the telling of a visual, statistical story with a beginning
(asking the question), a middle (collecting the data), and a climax at the end (displaying and
analyzing the data). The story must be told in such a way that the reader or viewer can
follow all parts of the story and be able to see or draw the correct conclusion at the end.
The mechanism for telling the story may take on various forms. Depending on the
complexity of the project and any external factors (e.g. competition requirements, teacher
directions), the project may be a paper, a poster, or an oral presentation.
Whatever the means of delivery, the question must be understandable, text must be readable,
and graphics must be clear, concise, and link back to the question. Any conclusions
presented must be sound and based on the data, or if not directly communicated, at least
easily drawn from the information presented. The use of color and other accouterments
should enhance the presentation without distracting from the overall message.
Is There Advice Specifically for Statistical Posters?
John W. Tukey (Statistical Science, 5(3): 327-339) said,
“The purpose of a statistical poster, then, is to visually tell a story from the data about some phenomena
revealing the conclusions that can be drawn. Because there is no narrator to tell the story, nor an
accompanying report to discuss the data, the poster must be able to stand alone; it should not have to be
explained.
For this reason, special care must be taken to present ideas clearly. Not only must the viewers understand
the individual graphics, but they must also understand the relationships among the graphics and how the
graphics address the question(s) being studied.”
Here are some guidelines from the Poster and Project Competition webpage at
www.amstat.org:
BASIC GUIDELINES
While constructing a poster, it is important to keep in mind that the central idea of the study should be the
prominent feature of the poster. To bring the main idea into focus, questions such as the following should
be asked.
◊ What is the purpose for displaying this information?
◊ What comparisons should be made?
◊ Which trends should be shown?
Questions should be asked until the central idea of the study becomes clear. This becomes the focal point of
the poster. The poster must reveal what the data have to say. It must allow the viewer to see the data; that
is, to see the variation in the data, the structure of the data, the important patterns in the data (or lack of a
pattern), the data points that do not fit the pattern, and the conclusions to be drawn from the data. Further,
each graphic on the poster should convey new information about the data-a pattern or structure, for
example, that cannot be seen in the other graphics.
The poster title should be informative in order to reduce the need for additional explanatory text. For
example, the title may indicate the questions addressed by the graphics or even convey the major
conclusion to be drawn from the data.

Each graphic's legend should be positioned so there is no question which graphic and legend go together.
Further, each graphic and its legend should stand alone. If the graphics need to be viewed in a certain
sequence, however, then the viewer's eyes must be guided in that sequence.
Try to eliminate trivial and extraneous information, gridlines, or lettering. In particular, redundancy in titles
and legends should be omitted. Only explanations needed to make the conclusions clear should be included.
Data tables should not be shown on the poster, as reading off numbers is not the point of the display.
Choose a few harmonious colors that are easily visible. The key to using color effectively is restraint.
Colors should not distract the viewer, but should enhance recognition of the structure of the data and the
conclusions.
Edward Tufte (The Visual Display of Quantitative Information) said graphics may "…reveal the data at
several levels of detail, from a broad overview to the fine structure." In a similar sense, a poster may do the
same. At a distance, only a broad overview of the poster and the data is possible. Therefore, main titles
should be visible, and overall outlines of the data-as revealed by the graphics-should be seen. On closer
inspection, however, aspects such as individual data labels and legends should be more apparent.

Chapter V: Assessing the Project
How does one Assess the Project?
The assessment of a project will depend on its guidelines and requirements set forth by the
teacher or sponsoring organization. For example, the following is a project assessment
outline by teachers Amy Strube (Neal ES, Las Vegas) and Chris O’Neill (Lied MS, Las
Vegas):
GRAPH PROJECT
Name: ____________________
Project Due Date: ____________________
• Each student will create and present an original graph project to the class. The poster and presentation
must depict an understanding of the data and the graphs used.
• The graph project will be assembled and displayed on a poster board measuring between 18 and 24 inches
high and 24 and 30 inches wide.
• Computer graphics may be used to illustrate, decorate, or label the poster. The student must make all
graphs and tables. Neatness and organization of the poster is important and will factor into the final
grade.
• Tape this form to the back of the poster.
• Be prepared to present your poster, discuss why you chose this topic and explain how you collected your
data.
__________ 1. Formulate a survey question (10 pts.)
Think of something you can measure, or get an opinion about. Phrase it as a question.
You might have to offer a limited choice of responses. Form a secondary question that
further explores or examines the topic. This is also displayed on the poster.
Display a title for your investigation on the poster.
__________ 2. Gather data for your survey questions (20 pts)
You will collect data about your questions ether by observing and recording it over a
period of time, or by interviewing or observing at least 50 people. Display the results of
the survey question in appropriate tables on the poster.
Be sure to properly title and label the tables.
__________ 3. Present a statistical analysis of the data. (10 pts.)
Calculate the mean, median, mode, and range of the data. Remember, this cannot be done
with all types of data. Display a summary of the statistical analysis on the poster.
If your data does not have this, then write a statement explaining why.
__________ 4. Display the data using 2 different graphs. (50 pts.)
Create 2 different graphs that appropriately display the results of the survey questions.
Be sure to title and label the graphs. Also be sure to show the scale and intervals.
Display the 2 graphs on the poster. Neatness and accuracy are very important parts of
these graphs. Also, this is to be done in color.
__________ 5. Write a summary, conclusion and prediction based on the data (10 pts.)
Summarize the results of the survey data, draw a conclusion about the data, and make a
prediction using the data. Include the summary, conclusion, and prediction on the
poster.
__________ Total: 100 pts.

The Nevada K-12 Statistics Poster Competition uses the following guidelines in assessing
posters submitted to it:
♦ Overall Impact of the Display
Poster beckons the viewer.
• It is eye-catching and has visual attractiveness.
• Spelling and grammar are correct.
• Construction of the poster is neat and balanced.
• Poster is free of clutter.
• Color is pleasing and well-used (not over used or under used).
• Type is readable.
Points that will reduce ranking on a poster include:
• Errors in spelling and grammar.
• Errors in construction of the poster.
• Clutter.
• Inappropriate or overuse of color.
• Lack of readability of type.
♠ Thoughtfulness of the Purpose and Method of Collection
Poster conveys a purposeful question properly investigated.
• The question shows depth of thought for the grade level.
• The methods and volume of data collection are appropriate and sufficient to answer the
question.
• The raw data is included.
• Questions that lack depth of thought or are not grade level appropriate.
• Errors in the methods or volume of data collection.
♥ Clarity of the Message
Poster tells a story.
• The message demonstrates important relationships and patterns, evident conclusions and the
ability to stand alone, even without the explanatory paragraph on the back.
• The question is clear.
• The answer to the question is obvious from the components.
• All components of the poster lead to answering the question.
• The display may lead to further questions.
• The question is not clear and requires examination of the back of the poster, or is not clear at
all.
• Determining the answer to the question requires some effort such as: examination of the back
of the poster or it is indeterminable.
• One or more components are irrelevant to answering the question.
♣ Appropriateness of Graphs
Graphs (and measures) are correctly created.
• All graphs are correctly constructed.
• Any measures are correctly computed.
• All graphs and measures are appropriate for the data.
• One or more graphs have errors in construction.
• Measures are incorrectly computed.
• Graphs and/or measures are inappropriate for the data.
• Creativity

The following are some examples of well-designed and poorly-designed posters and graphs
from the Nevada K–12 Statistics Poster Competition sponsored by the Nevada Chapter of the
American Statistical Association.
Well-designed
Figure 1 Figure 2
Figure 3 Figure 4
Figure 5 Figure 6

Figure 1: A first-grade poster that is well-organized and the display clearly communicates
how animals move.
Figure 2: Very clean and nicely laid out third-grade poster. Graphs are easy to read and
interpret. Brief predictions and summary, while not necessary, communicate purpose and
add more information.
Figures 3 & 4: This 12th
-grade poster has a balanced layout and is easy to follow. Summary
paragraphs were a bit verbose, but did not detract from or contradict the information in the
graphs. Computer-produced graphs were easy to read and interpret, and clearly showed the
trends in rainfall. Each graph showed a different aspect of the rainfall data.
Figures 5 & 6: Well-produced poster with graphs displaying multiple aspects of the
information collected. Majority of the poster is dedicated to data display and graphs are easy
to read. Notice that no summary paragraphs are present, or needed, as the graphs effectively
stood alone in communicating how computers impact students’ lives.
Poorly-designed
Figure 7 Figure 8
Figure 9 Figure 10

Figures 7 & 8: Not only does this sixth-grade poster fail to include two graphical displays
relating different aspects of the data, but the included summary statistics are inappropriate.
Favorite sport is a categorical variable, from which quantitative measures like mean and
median cannot be computed. The mode here is actually the category “soccer.” This poster
has a lot of clutter and not much information.
Figure 9: More than three-fourths of this poster is dedicated to accouterments of neon paper,
glitter, and glue. The graphs are practically unreadable. Like the poster in Figure 7, it is
clear more time went into decorating this poster than went into collecting and analyzing the
data.
Figure 10: The topic of this 10th-grade poster was a comparison of areas of six New
England States. The line plot poorly displays the data as it is not drawn to scale. The box-
and-whisker plot also suffers from scaling problems, is inappropriate for such a small data
set, and generally conveys the same information as the line plot. For this reason, these were
not considered to meet the competition’s multiple graphs requirement. Finally, and most
importantly, simply listing the areas of six states and making a graph of the areas is a feeble
topic for a high school, or even middle school, poster.

Resources and References
• Nevada Chapter of ASA Poster Competition Web page, where one may view
previous winning posters, download registration forms, and find this document.
www.nevada.edu/~nvasa/
• American Statistical Association Poster Competition Web page, where one may view
previous winning posters.
www.amstat.org/education/index.cfm?fuseaction=poster1
• Bock, David E., Velleman, Paul F., and DeVeaux, Richard D. (2004). Stats:
Modeling the World. Boston: Pearson/Addison-Wesley.
• Curcio. F.R. (2001), Developing Data-Graph Comprehension in Grades K–8. The
National Council of Teachers of Mathematics. www.nctm.org
• Landwehr, J.M. and Watkins, A.E. (1987). Exploring Data, Palo Alto, CA: Dale
Seymour Publications.
www.pearsonlearning.com
• Tufte, E.R. (1983). The Visual Display of Quantitative Information, Cheshire, CT:
Graphics Press.
www.amazon.com
• A Curriculum Framework for PreK-12 Statistics Education (2005), Participants of the
Guidelines for Assessment and Instruction in Statistics Education (GAISE)
www.amstat.org/education/gaise
• Math on Call: A Mathematics Handbook (1998), Wilmington, MA. Great Source
Education Group.
www.greatsource.com
• Navigating through Data Analysis [and Probability] (2003), Reston, VA, National
Council of Teachers of Mathematics.
• Take It to the MAT Newsletters. Click on the Math Newsletters link.
www.rpdp.net
• Any good high school- or college-level statistics textbook.

Nvasa data project guide v0907

More Related Content

Similar to Nvasa data project guide v0907 (20)

More from Sarah Sue Calbio (20)

Recently uploaded (20)

Nvasa data project guide v0907