Homework 5: Index Construction & Use Comments
This assignment continues a series of labs and homeworks in which you utilize statistical skills for basic
research. For this assignment, you will again manipulate variables and construct a basic index, as you have
done in several earlier assignments. However, for this assignment, you will take an additional step, using the
index that you create for simple bivariate descriptions of the sample. The index will be the “dependent variable”:
Specifically, you will use ordinal measures to compare support for possible explanations of variation in the index.
You will be using the data file hw5.sav to examine variation in respondents’ satisfaction with four areas of their
lives (family, friends, finance, and job). You will then create a summary measure of overall satisfaction, and will
explore how (and whether) that summary measure varies in two ways: across educational levels and with
frequency of sexual activity. Finally, you will briefly explore interactions among these possible influences on
satisfaction. (Note that most of the recoding has been done for you – this is not always the case.)
Requirements & Questions
You must submit your output file (complete but cleaned) and typed answers to these questions. Typed. Probably
with a computer, maybe with some other device, possibly a typewriter. But not a pen, pencil, or crayon. Typed.
1. Univariate analyses of component and independent variables:
• Perform a univariate analysis of SATFAM, SATFIN, SATJOB, and SATFRND – For each, you should
look at and briefly summarize the frequency distribution, as well as basic summary statistics for central
tendency and dispersion. Go beyond just reporting the data and say something interesting (here and
below). For example, about which issues are the respondents the most/least happy?
• Look briefly at the distributions of EDUC and SEXFREQ. (Note, in particular, the percent of the sample
who refused to answer or otherwise did not have an answer for SEXFREQ.)
2. Construct and assess index:
• Construct an index (including variable labels and value labels, at least for the extremes), called
HAPPY, as the summation of values for the four components listed above.
• Perform a univariate analysis of HAPPY – look at and briefly summarize the frequency distribution, as
well as basic summary statistics for central tendency and dispersion..
• What is this variable conceptually? What does it measure, and what does it mean? What does it tell us
that the individual components do not?
• Interpret the “alpha” for your index – is the index reliable? is it a good one? why or why not?
3. Bivariate analyses – what makes people happy?
• Using correlations and chi-square, what can you say about the relationship between educational
attainment and overall satisfaction (i.e. between HAPPY and EDUC)? (You will need to request a
crosstab to get chisquare, but ignore the table itself, for now.) Is it strong? statistically significant?
• Using correlations and chi-square, what can you say about the relationship between frequency of
sexual activity and overall satisfaction (i.e. between HAPPY and SEXFREQ)? (You will need to request
a crosstab to get chisquare, but ignore the table itself, for now.) Is it strong? statistically significant?
• What can you infer from these findings about what makes people happy? (Hint: Did either of the two
independent variables (EDUC and SEXFREQ) have a statistically significant effect on the dependent
• Bonus: Put that at a conceptual level, thinking about what broader concepts these variables might
operationalize. Of what larger concept might education be a specific instance, indicator, or aspect?
What about sexual frequency?