Exercise 8: Abortion attitude, presidential vote, and perception of candidate positions
Another example of a conditional relationship involves the same initial relationship that we used in exercise 7, the relationship between attitude on abortion and presidential vote. Since you should already have this two-variable table from the previous exercise, we can begin immediately with a consideration of some possible conditional variables.
The initial hypothesis that pro-abortion voters would be more likely to vote for Obama (and anti-abortion voters more likely to vote for McCain) assumes that voters would perceive the two presidential candidates as differing in the expected way on this issue. To examine how voters perceived the positions of the two candidates on this issue, examine the frequency distributions for V106 and V107.
As you can see, not every voter saw Obama as pro-abortion and McCain as anti-abortion. This might lead us to hypothesize that the relationship between abortion attitude and the vote will be much stronger for voters who correctly saw the differences between the candidates on this issue. To test that hypothesis, you should generate a table with abortion attitude as the independent variable, presidential vote as the dependent variable, and two control variables (V106 and V107). To simplify the tables, recode V106 and V1071 so that they have just two categories, pro-abortion and anti-abortion.
1Recoding V106 and V107: Since you have recoded V104 (the respondent's attitude on abortion) into two categories, V106 and V107 should be recoded in the same manner. The first two categories (never allowing abortion and allowing abortion only in very limited circumstances) should be combined into one anti-abortion category, and the last two categories should be combined into one pro-abortion category.