Soc 2155 SPSS 6: Elaboration (partial tables)

Cross tabulation is a form of bivariate statistics, a way of looking for causal relationships between two variables at a time. Elaboration is a form of multivariate statistics, a way to bring in a third control variable, and see how it affects the original bivariate relationship.

1. Open GSS2000Selection 2, and using the utilities function, look at the variables beginning with "nat"; notice that we can't be sure what these questions are asking about, so you need to consult the mnemonic index found at the General Social Survey home page. (Go to my home page and click on "General Social Survey." Under "codebook indexes," click on "mnemic," click on "n," and scroll down till you find the variables beginning with "nat." On a separate sheet of paper, tell me what this series of questions is about?

2. Let's do a cross tabulation to look at the relationship between "class" and "natcrime." Be sure to put the independent variable into columns and to ask that the table be percentaged within columns. Don't print the output, but on that separate sheet of paper, describe and interpret the pattern you see.

3. Let's do the same cross tabulation again, but this time, include a control variable, "race," in the bottom of the three boxes (the top one says "Row(s)," the next one "Column(s)," and the bottom one isn't labeled but it's for control variables. Be sure the column percentaging is activated (click on "Cells" to be sure) and then click on "OK." Describe and interpret the pattern, in terms of our categories of elaboration (direct, spurious, interventing, interaction) and also in your own words. Don't print yet.

4. Do the same cross tabulation again, but this time control for "sex." Just double-click on "race" to get it out of the control box and put "sex" in. Click ok. Describe and interpret the pattern Don't print yet.

5. Let's start over with another of the "nat" variables, this time "natfare," and use a cross tab to look at the relationship between "race" and "natfare." Describe and interpret the pattern. Don't print yet.

6. As you saw above, minorities were more likely than whites to support having the government spend more for welfare. The obvious question is whether this is just a function of a higher poverty rate. Does the higher rate of poverty function as the intervening variable here. You need to do another recode of income98, but this time recode it into five approximately equal portions, and call your new variable incomer5. Be sure to change the decimals, add values, and change the level of measurement. Now use your new variable as a control variable to re-examine the original cross tab between "race" and "natfare." Does the original relationship between race and natfare disappear within categories of the control variable (which would support the "intervening variable" interpretation) and if not, how would you describe and interpret the pattern you see. Don't print.

7. Finally go through the same process for the relationship "class" and "polindex." Then control for "race," and interpret the results. Now you can print the output you're created and attach it to your sheet of interpretations. Be sure to print in "landscape" orientation, so that those big tables will fit on one page (Click "file," "properties," "basics," landscape," "okay," "okay." If you want it still smaller, you could go into "properties" and "effects" and print it at 60 or 70% normal size.)