Exercise 4. Attitude on Obamacare and the Presidential Vote
Another issue that might have influenced how people voted in the 2012 presidential election is the issue of the Affordable Care Act, commonly referred to as Obamacare. To examine whether attitudes on this issue affected voting, we can look at a table that relates attitudes toward Obamacare to the presidential vote. Respondents in the 2012 ANES were asked about their support for Obamacare (J10). We would hypothesize that having a more favorable view of Obamacare would would make one more likely to vote for Obama. To examine that possibility, we can look at a table that relates attitude toward Obamacare (J10) to the presidential vote. For the reasons suggested in Exercise 1, you should use the recoded version of A02 that you created for that exercise, so that you examine only the major-party vote (i.e., only the Obama and Romney voters).
After examining Table 4A, you should conclude that those who approved of Obamacare had a far greater propensity to vote for Obama than those who disapproved. However, as we learned from Exercise 3, this does not necessarily mean that attitudes on this issue really had a significant effect on the vote, meaning that people voted on the basis of this issue. The relationship in Table 4A could be the result of the actions of a confounding variable.
As in Exercise 3, party identification is a possible confounding variable, which should be examined to better understand why our independent and dependent variables are related. To do so, you need to construct a three-variable table that shows the relationship between attitude toward Obamacare, presidential vote, and party identification. To ensure that you have a sufficient N for each column, you should recode J10 so that it has just three categories (favor, neutral, and oppose) and use the recoded version of party identification that you created for Exercise 1 (Democrats, independents, and Republicans).
In this example, the relationship between attitude toward Obamacare and the presidential vote persists even after we control for party identification. This indicates that attitudes on this issue did influence how people voted. The association between the independent and dependent variables is not as strong in Table 4B as it was in Table 4A, so we would conclude that party identification did inflate the bivariate association between the two variables — but the important point is that the two variables still are clearly and substantially associated even when party identification is controlled for.