In the
previous chapter we discussed data analysis. In this chapter you will have an
opportunity to analyze data. The exercises in this chapter start at a fairly
simple level and become more complex. The final exercise asks you to define
your own problem and carry out the analysis. (Your instructor may choose to
supplement these exercises with some of his or her own.) You will be using your
computer facilities to do these exercises, but there is no assumption that you
know anything about computers before you start. Every effort has been made to
minimize what you will have to learn about computers. Your instructor will provide
you with the necessary information.EXERCISE ONE
One of the questions
in the data set asks respondents whether there are "more advantages in being
a man, more advantages in being a woman, or ... no more advantages in being
one or the other." We want to find out some of the factors that are related
to this opinion. Specifically, we want to know if men and women differ in
their opinions, and whether education is related to how one feels about this
matter.Question 1.
Find this question in the codebook (see Appendix A). Locate
the variable name for this question. This is the name the computer recognizes.
You will have to use this name when asking the computer to do something for
you. What is the variable name?Question
2. The codebook gives both the frequency distribution and the percentage
distribution for this question. What percent feel that there is more of
an advantage in being a man? What percent feel there is more of an advantage
in being a woman? What percent think it doesn't make any difference if one
is a man or a woman? Show how these percentages were calculated (i.e., do
the simple arithmetic to compute this percentage).Question
3. We would like to know if men and women hold different opinions? To
answer this, we must crosstabulate sex (V34) and advantages in being a man
or a woman (V1). We're interested in comparing the opinions of men and women.
So sex (V34) is the independent variable and advantages in being a man or
a woman (V1) is the dependent variable. (You probably will want to review
Chapter Three on independent and dependent variables.) Be sure to ask for
the appropriate percentages (either the row or the column percents) and
chi square. Ask the computer to give you the appropriate measure of association
(either Cramer's V or Gamma). After you get the table, write a short paragraph
interpreting your results. Be sure to refer to the percentages from the
table, chi square, and the measure of association. Remember that you are
trying to determine if men and women differ in their opinions, and, if so,
how they differ.Question
4. Would people with more education be more or less likely to see advantages
to being a man? You will want to crosstabulate education (V24) and advantages
in being a man or a woman (V1). First, let's recode education by dividing
it into three categories--those who have a high school degree or trade school
or less, those with some college but no college degree, and those who have
a college degree (bachelor's, master's, or post-master's). Now crosstabulate
the recoded education variable (V24) and advantages in being a man or a
woman (V1). Be sure to get the appropriate percentages, chi square, and
the appropriate measure of association. Write a short paragraph interpreting
your results. Use all the statistics (i.e., percentages, chi square, measure
of association) in your answer. Remember that the question is whether there
is a relationship between education and whether one thinks there is an advantage
in being a man or a woman, and, if so, what the relationship is.Question
5. In the previous questions, we considered sex and education separately.
In other words, we first crosstabulated sex and advantages and then crosstabulated
education and advantages, producing two tables, each with two variables.
Now we want to crosstabulate advantages and sex using education as the control
variable. In other words, we want to compare the opinions of men and women
holding education constant (review Chapter Three). Use the recoded version
of education that you created for question four. After you have obtained
the tables, start by looking at the relationship between sex and advantages
for those with a high school degree or less. Do men and women with a high
school or less education differ in their opinions? Now look at those with
some college and then finally those with a college degree. Do these men
and women differ in their opinions? Be sure to use the percentages, chi
square, and the measure of association to help you answer these questions.
Write a paragraph or two summarizing your findings.Question
6. This time let's crosstabulate advantages and education using sex
as the control variable. In other words, we want to compare the opinions
of those with different educational levels holding sex constant. Use the
recoded version of education that you used for questions four and five.
After you have obtained the tables, start by looking at the relationship
between education and advantages for the males. Is there a relationship?
What kind of relationship exists (i.e., are those with more education more
or less likely to think that there is an advantage in being a man)? Now
look at the females. Is there a relationship? Write a paragraph or two summarizing
your findings. Be sure to use the statistics to help you. Why do you think
these findings occurred? In other words, what do you think accounts for
these relationships?EXERCISE TWO
The Field Poll
included two very interesting questions about housework. One of the questions
asks who should clean the house when both the husband and the wife work full
time outside the home; the other asks who actually does most of the housework
in the home. These questions open a number of intriguing opportunities for
analysis.Question 1.
Let's start by looking at the percentage distributions. Find these two questions
in the codebook and locate the variable names. Who do most respondents think
should clean the house when both partners work full time? What percent of
the respondents gave this answer? Who do the respondents think is actually
doing the housework in their homes? Write a complete sentence summarizing
the answers of the respondents in the sample. Use example percents to illustrate
your description.Question
2. It would be interesting to know how the married females with spouses
who work full time feel about housework. The computer can help us by selecting
only the married females who have spouses who work full time. This means
you will have to tell the computer to select out the married (value 1 on
V19) females (value 2 on V34) whose spouse works full time (value 1 on V20).
(Your instructor will show you how to do this.) What percent of these married
females think that both partners should share the housework equally when
both partners work? What percent of these married females feel that the
housework is being shared equally in their homes? (Hint: First, you will
have to select out the married females with spouses who work full time.
Then, you will have to ask for frequency distributions for the two questions
dealing with housework.) Write a complete sentence summarizing these results.Question
3. What about the married women who work full time themselves? Perhaps
housework is shared more equally in the homes of these women than in the
homes of the married women who don't work full time. To simplify this, have
the computer recode the respondent's work status (V18) to separate those
working full time from all others. Do this by combining part-time, temporarily
unemployed, and not employed into one category. This will leave two categories
-- employed full time and not employed full time. Obtain a frequency distribution
for this recoded variable for these married females with spouses who work
full time. What percent of these married females worked full time?Question
4. By now it is clear that we want to compare the married women who
work full time with the married women who don't and find out if housework
is more likely to be shared in the homes of the married women who work full
time than in the homes of the married women who don't work full time. Before
you start, we had better go through the steps involved. First, select out
the married females with spouses who work full time. Second, make sure that
you have recoded employment status (V18). Third, crosstabulate who actually
cleans the house (V3) and the recoded employment status (V18). We're really
interested in comparing who does the housework in families in which women
work full time and in families in which women do not work full time. We
suspect that employment status influences who they think should do the housework.
So employment status (V18) is the independent variable and who actually
cleans the house (V3) is the dependent variable. (You probably will want
to review Chapter Three on independent and dependent
variables.) Fourth, be sure to ask the computer to give you the appropriate
percentages, chi square, and the appropriate measure of association. After
you get the table, write a short paragraph interpreting your table using
the percentages, chi square, and the measure of association from your computer
output. Summarize the relationship in words. Are the married women who work
full time more or less likely than the married women who don't work full
time to say that housework is shared equally in theirEXERCISE THREE
Another question
asks respondents if they "favor or oppose efforts to strengthen and change
women's status in society today." As in exercise one, we want to find out
how some of the factors are related to one's opinion. We will focus on the
relationship of sex, education, and age to opinions regarding women's status.Question 1.
Find this question in the codebook and locate the variable name. What percent
favor efforts to strengthen and change women's status? What percent oppose
such efforts? How many respondents had no opinion? What percent is this of
the total number of respondents in the sample?Question
2. Do men and women differ in their opinion? What two variables do we
have to crosstabulate to answer this question? Be sure to obtain the percentages,
chi square, and the appropriate measure of association. Write a short paragraph
interpreting your table, using the statistics you obtained.Question
3. Is education related to whether one favors or opposes efforts to
strengthen or change women's status? Before you obtain the crosstabulation,
be sure to recode education by dividing it into three categories--those
who have a high school degree or trade school or less, those with some college
but no college degree, and those who have a college degree. Obtain the crosstabulation
you need to answer this question, along with the appropriate statistics.
Write a short paragraph interpreting your table, using the statistics you
obtained.Question
4. For the rest of this exercise we want to consider only those respondents
who work full time. You will have to tell the computer to select out these
respondents. Do men and women who work full time differ in their opinions?
Obtain the appropriate crosstabulation and statistics. Compare your results
here with those in question two. Are they similar or different? Write a
short paragraph interpreting your table and comparing it to the table from
question two. Be sure to use the appropriate statistics to help you interpret
these tables.Question
5. For this question you should compare the opinions of men and women
who work full time controlling for education. Remember to first select out
those who work full time. Then obtain the appropriate three-variable table,
along with the statistics. Has education affected the relationship between
sex and opinion regarding women's status? Compare the partial tables obtained
in this question with the two-variable table from question four. Write a
short paragraph interpreting these results.Question
6. There is a problem with the crosstabulations you obtained in question
5. Notice that the expected frequencies are less than five and that some
of these expected frequencies are quite small. Chi square assumes that these
expected frequencies are five or larger. Statisticians tell us that as long
as 80 percent of the expected frequencies are five or larger and no single
expected frequency is very small, we don't have to worry. However, in this
case some of the expected frequencies are quite small. (You will want to
review the section in Chapter Three on chi square.) We can solve this problem
by recoding V2. How are we going to do this? If we combine favor strongly
and favor somewhat we will have over 80 percent of the cases in this one
category. It might be better to leave favor strongly as one category and
combine favor somewhat, oppose somewhat, and oppose strongly as a second
category. This would give us those who are strongly in favor of efforts
to strengthen women's status in one category and all those less committed
in another category. Recode V2 in this manner and repeat the analysis in
question 5. Notice that the expected frequencies are larger now. You no
longer violate one of the assumptions for chi square.EXERCISE FOUR
- Choose one
of the questions about women's status and roles (V5 to V17) as a dependent
variable. What is the variable name? Using the information in the codebook,
write a sentence describing the responses to that question.- What variables
do you think might be related to your dependent variable? Write down the
ways in which these variables (i.e., your independent variables) might be
related to the dependent variable. Be sure to explain the rationale underlying
each hypothesis. (Review the discussion of hypotheses in Chapter Three.)
In other words, explain why you think this hypothesis should be true.
For example, you might think that age would be related to the dependent
variable such that, as age increases, your dependent variable decreases.
Explain why age should be related to your dependent variable in this manner.
Include at least three hypotheses.One hypothesis
should consider the relationship between sex and your dependent variable.
(Would you expect men and women to be similar or different? Why? The rationale
for your hypothesis should be organized as an argument leading us to conclude
that your hypothesis is plausible.)A second hypothesis
should consider the relationship between your dependent variable and another
social-status variable (e.g., age, race, education).A third hypothesis
should consider the relationship between one of the opinion variables (e.g.,
favor or oppose efforts to strengthen or change women's status) or behavior
variables (e.g., who cleans house when both spouses work) and the dependent
variable. You may consider more than three hypotheses that can be tested
with these data.- Obtain the
two-variable tables needed to check on your hypotheses. One crosstab should
consider possible sex differences. The other tables should include one other
social-status characteristic from the background data (V18 to V33) and one
opinion or behavior variable (V1 to V17) as independent variables. Present
these in the form of individual tables with a short written summary of the
results of each table that links it back to the hypotheses considered.- Then choose
one of your two-variable crosstabs and explore the relationship (or lack
of relationship) more fully using another variable as a control. For example,
you might want to consider the relationship between sex and your dependent
variable controlling for such variables as age, income, or education. Restate
your hypothesis as clearly as possible. Indicate why you selected this control
variable. Present the three-variable table with a written summary that considers
this hypothesis.- Write a brief
summary of your findings. Be sure to discuss your hypotheses and whether
the data support the hypotheses. What were the most important findings in
your analysis?
Last Modified 15 August 1998