DO-YOUR-OWN
EXERCISE
THE EFFECTS OF CLASS AND RACE ON ATTITUDES AND BEHAVIOR
Introduction
to Sociology
Professor Robert Wood
Overview: This exercise asks you to use the survey variables in the MicroCase program to 1) formulate a hypothesis about the relationship between class and some other variable; 2) operationalize the variable, social class; 3) run a cross tabulation to test the hypothesis; 4) introduce a control variable; 5) run a new set of cross tabulations to see what happens to the original relationship when the control variable is introduced; 6) interpret the results in a clear and accurate way.
Requirements: The exercise should be written up as a mini-research report. It must be typed, with the various tables "pasted" into the text. It is important that it perform all the tasks listed above (which are elaborated upon in more detail below). There is a link to a model exercise at the bottom of this assignment.
Step 1: Formulate a hypothesis
I want you to start by formulating a hypothesis about the relationship between social class background and one of the behavioral or attitudinal variables from the General Social Survey (Data File: GSS). These variables are listed in the Appendix in the back of your workbook, Discovering Sociology; but you will also want to access the questions on which they are based by clicking on the variable in the MicroCase program. The reason why I want you to start with social class as your independent variable is that social class has proven to be a powerful predictor of an extremely broad array of attitudes, behaviors, and life chances. Choose a dependent variable that you think is likely to be associated with social class, and formulate a hypothesis about the relationship between it and social class, being specific about the direction of the relationship. Be sure that your hypothesis is stated in roughly the form:
The higher/lower (the independent variable) .... the higher/lower (the dependent variable)
Explain briefly why you think this relationship is likely to exist. Your dependent variable must be either an attitude or a behavior. Write down the question on which it is based and include it in your discussion.
Step 2: Operationalize the concept, social class
You now need to figure out how to measure social class. We call this process of linking a concept to data "operationalization." We will discuss in class the variables that would be most appropriate for doing this. Choose one of them and explain briefly why you think it is a good measure of class. (Note: it is not acceptable to justify your choice of measure by its likely relationship to the dependent variable; the issue here is having a good measure of social class.) Now restate your hypothesis, substituting for "social class" the variable you've chosen to operationalize it with.
Be sure to use one of the three variables specified in class: these are #24 (education), #43 (family income) and #235 (Parent's occupational prestige). Do not run one of the measures of class against another (e.g. income and education). Remember, your dependent variable must be either an attitude or a behavior and must be defined independently of the independent variable.
Step 3: Test your hypothesis
Now run a cross-tabulation for these two variables, making your independent variable the column variable and your dependent variable the row variable. Click on Column % to see the column percentages. Examine the Cramer's V statistic to see if the differences in the table are statistically significant.
| Note: Be sure that you have made your independent variable your column variable and your dependent variable your row variable. Failure to do this will produce an incorrect table and will result in substantial loss of points. For a review of correct table layout, as well as guidelines for interpreting Cramer's V, see Table and Graph Format at the departmental website. |
Print out your table or copy and paste it electronically as indicated below. Either way, your table should be integrated into the text, not appended at the end.
| [Note: Using the Edit->Copy feature, you can copy the table into your computer's memory and then paste it into a word-processing document, but you will have to reformat it subsequently. It is also possible to use the print-screen function and then crop the image in Start->Accessories->Paint; for instructions on how to do that, click here.] |
Examine the level of significance and the data in the table to determine whether your hypothesis was supported or not. If the table is not statistically significant, you must reject your hypothesis. If the table is statistically significant, you still must examine the table to make sure that the differences are in the direction that you predicted. You should also examine Cramer's V for a rough indication of the strength of the relationship between the two variables. Explain whether the data support your hypothesis or not and why, and what the strength of the relationship is if the hypothesis is supported. Check the department's Table and Graph Format webpage to interpret the strength of the relationship indicated by the value of Cramer's V.
| If the data are statistically significant but do not support your hypothesis, see if you can reformulate your hypothesis to fit your data. Is this revised hypothesis supported by the data? Explain. |
As we have seen in class on several different occasions, sometimes a relationship that is statistically significant "disappears" or is otherwise modified when we introduce a third, control variable. In other cases, the relationship may hold for some sub-groups but not for others. In the next part of this exercise, I want you to introduce a control variable and see if it modifies the results you have found so far.
For your control variable, choose Race/Ethnicity (variable 29) to see if your original findings are altered when you control for it.
| Note: if your hypothesis was rejected because the table was statistically insignificant, you may treat this as a finding in its own right and proceed to choose a control variable as indicated above. It is possible that controlling for race could produce statistically significant results, despite the fact that the original relationship was statistically insignificant. Since this is somewhat unlikely, however, you may want to restart the exercise and play around with different possible dependent variables until you find one that is significantly associated with the variable you used to operationalize social class, and then retrace the steps above before adding the control variable. You will probably get more interesting findings this way, and I recommend doing this. |
In this step you are seeing if the results you just obtained about the relationship between social class and the dependent variable you chose hold for whites and blacks separately (note: because the small number of Hispanic respondents tends to cause significance problems, we will focus only on whites and blacks in this exercise) . Run your cross tabulation again, entering your row and column variables as before but adding Race/Ethnicity (variable 29) as the control variable. As before, get the column percentages and then print out the tables for whites and blacks. Be sure your printouts include Cramer's V and the level of significance. As before, the two tables should be integrated into the text.
Step 6: Interpreting the Results
In a paragraph or two, discuss the results of controlling for race. Did your original finding hold or did you get different results for whites and blacks? Even if the relationship holds, are there noteworthy differences in the response range, or are the distributions basically the same? Discuss.
Checklist For Your Final Report
Be sure that your report contains the following (review the text above for more complete detail):
- Your original hypothesis and a brief explanation of why you think the hypothesized relationship is likely to exist. Include the actual question in your discussion.
- An explanation of how you operationalized social class and why you think the measure you chose is a good one.
- Your restated hypothesis
- The cross-tabulation of your two variables, including column percentages, Cramer's V, and the level of significance. This and the tables below should be pasted into your paper at the proper point, not appended at the end.
- An explanation of whether the data support your hypothesis and why you conclude this. If your hypothesis is supported, note the strength of the relationship as explained on the department's Table and Graph Format webpage.
- A brief hypothesis about the impact of the control variable (race) on the original relationship.
- The two tables produced when you added your control variable, including column percentages, Cramer's V, and the level of significance
- An interpretation of these results, including a discussion of the what happened to the original relationship when you controlled for race. The strength of the relationship should be properly indicated.
|
Further Resources for this Assignment Click here for a sample exercise using the variable "happy" that was used to demonstrate this exercise in class. Do not choose this variable for your exercise; I will not accept any paper that does so. Click here to see the checklist that will be used in grading your exercise. Click here to access the MicroCase Resources page at our departmental website, which includes two tutorials on testing a hypothesis using MicroCase. |
Return to Introduction to Sociology syllabus
November 14, 2006