suggestions apply to the three variations of this exercise: SPSS with the CSU
student survey, SDA with GSS data and SDA with ANES data
are other data sets and web sites
focused on SDA analysis. You might want to check these out for data more
appropriate for your teaching topic
Purpose of Exercise:
Social Science classes talk a lot about being a science but the general classroom
is little different from a history class with lectures, discussions and possible
group work but no "labs" or science exercises as one would find in Chemistry,
Biology, etc. This exercise is designed to let students learn the scientific
process and research report writing by practice. Obviously students cannot perform
the scientific process in depth since defining a topic, performing library research
on the topic, specifying a hypothesis and performing data collection and analysis
are time consuming and not possible in the typical time frame of a social science
class. What this assignment does is provide high quality data sets -- a survey
of CSU students, the General Social Survey (GSS) or the American National Election
(ANES or NES) -- to eliminate costly and time consuming data collection. The
theoretical and library portion of the research process is limited to information
readily available to the student from class text and notes. The focus for this
exercise is to develop students' knowledge and skills in the scientific process
and in data analysis.
In general I have
found that presenting parts of the exercise over several days for lower division
classes works better than covering it in one class session. (For upper division
and majors, it "may" be possible to assign this in one or two sessions or as
an extra credit assignment. Upper division students and majors will have more
questions which are more varied than the questions of beginning students.) For
beginning social science classes you may want to have one short session in a
computer lab to run SPSS or SDA to generate the frequencies and tables. A "smart
classroom" or overheads for presenting the tutorial and assignment are useful.
I have students access the web page, print the survey and assignment and bring
them to class for discussion. Following is a rough outline of how I go through
this exercise in a 3 day per week Introductory Sociology class in a quarter
system. This is the last assignment of the quarter and is given about week 7.
I usually let the paper from this exercise count as 10-20% of the final exam.
The following is an approximate schedule.
- Day 1 (10-15
minutes): give a broad overview of the exercise and assign the class the
task of taking the sample survey. Tell the class members to think about questions
that interest them from the sample survey. Questions might include how other
students would answer the question or how types of students (older vs young,
republicans vs democrats, different religious groups, etc.) might answer the
questions differently. Have students bring back the survey and assignment
- Day 2 (15-20
minutes): discuss with the class as a whole (and then have them discuss
in smaller groups), possible hypotheses about the survey data (relationships
between different questions). Discuss the assignment example A person's
Religious identification affects their attitude about abortion. Share
the groups efforts with the class as a whole. Students should bring at least
one original hypothesis to the next class. They cannot use an example discussed
in class unless the example was theirs.
- Day 3 (10
minutes): discuss topics and how a topic is a broad abstraction while
the hypothesis is a specific testable relationship between variables. For
example discuss the topic and hypotheses used in the assignment A person's
Religious identification affects their attitude about abortion is part
of the more general topic the importance of religious beliefs in relationship
to other beliefs and behaviors. Ask the class members to take their hypotheses
and expand to a general topic that they would like to examine with the survey
data. I usually tell them their hypothesis and topic should be their own.
The examples used in the assignment and discussed in class belong to those
who created them. Class members should bring to the next class a short typed
statement of their topic and hypothesis with the survey questions for independent
and dependent variables specified. You will have to discuss briefly dependent
and independent variables for those students unclear on the meaning.
- Day 4 (15-20
minutes): have the class share their topic and hypothesis statements in
groups and improve or change them if necessary. You will need to make clear
that step 2 must be done now so that the hypothesis truly is an "educated
guess," not a "common sense" position. Explain that research starts from both
specific interests (hypothesis) as well as broad interests (topics) and that
seeing the broader implications of findings (topic) is an important part of
the scientific enterprise. Have the class discuss this in their groups and
possibly have summary examples from the groups presented in class. Clarify
questions that will be related to independence and dependence. Have the class
members hand in their statements with pencilled corrections so that you can
read and respond to any problems. NOTE: in this exercise students have not
done an extensive look at the literature. Our goal for this exercise is developing
skills in following the scientific process and analysis of data so we have
chosen to minimize the literature review. Have the class go to the web site
and print Obtaining data from the ... and bring to the next class.
- Day 5 (10-15
minutes): Demonstrate the SPSS or SDA web based analysis with their data
(CSU student survey, GSS or ANES) using the tutorial example specifying and
using the independent and dependent variables to appropriately create a table
with percentages. Assign the class the task of creating a table of their independent
and dependent variables using SDA and ANES. This is not as time consuming
as it may seem and should go pretty rapidly. If possible you might arrange
for the class to meet in a computer lab to do this as a group. Have the class
go to the web site and print Reading a Table and bring to the next
- Day 6 (10-15
minutes) Discuss "how to read a table," using the example in the assignment,
and have the class members share their understanding of their table with other
class members. The class is then assigned the completion of the exercise by
writing up steps 1-6.
If you want to take the analysis further:
- Day 7 (10-15
minutes): Use the tutorial example to explain significance and Chi Square.
I usually use a coin toss as the example since it is pretty simple to understand.
I can also discuss the difference between statistical significance (the likelihood
of obtaining 90 heads out of 100 tosses) versus personal significance (the
probability that the person getting the 90 heads is cheating with a two headed
coin). Don't forget to say that the .05 significance is a convention, not
a hard and fast rule. Here you will have to discuss characteristics of the
sample and the population; keep it simple since a beginning class student
may not have had a college statistics class. Have the class discuss their
table and ChiSq with their group. The class is then assigned the completion
of the exercise steps 1-6. If interpreting ChiSq is not of concern, you can
skip this part and have students leave out the ChiSq in their report.
- Day X,Y
and Z: Individuals or the whole class can test other hypotheses derived
from their topic. I usually restrict the first part of the exercise to two
variables for simplicity, but many students will see that social variables
have multiple causes and affect many other variables.
- Day X, Y
and Z: Discuss further possible analysis with the students. Choosing variables
not included in the sample questions, using a control variable, recoding variables
are all possibilities. You could show how to obtain data from the yearly surveys
to see changes over time. I would likely do this only in an upper level course
or as an independent studies or extra credit assignment.
The range of hypotheses/topics testable with this data set is enormous. They range
from the obvious to those that only the knowledgeable and creative social scientist
might examine. For example, one basic goal might be to predict voting behavior
from other variables. Some possibilities:
- Race and ethnicity
related to political preferences
- Social status
(income, education, etc.) related to political preferences
- Race, ethnicity,
age, gender related to attitudes
Some not so obvious
- Studies have
found that those who get their information from TV change little in their
knowledge of the candidates over time while those who get their knowledge
of candidates from newspapers, magazines, etc. gain in their knowledge of
candidates. Our hypothesis would be that TV source people would be less likely
to change their voting preferences.
- Studies have
found that those who watch situation comedies are more anarchistic then those
who watch crime shows. What hypothesis could we come up with from this?
Some possible questions
- Who votes?
- Who doesn't
Sunday, 6 August 2000