The "Moral Values" Illusion
Pollsters and pundits continue to debate the importance of "moral values" in the 2004 presidential election. In the November 22, 2004, issue of Newsweek, Joe Klein dubbed the phenomenon "The Values Gap" for the Democrats. Commenting on Gary Langer's November 6 New York Times op-ed piece on the poorly worded exit poll question on "moral values," senior Gallup poll editor, David Moore, expressed the view in a December 7 POLLTALK column that, despite all the fuss about the fuzziness of the meaning of the phrase and the closed-ended form in which the question was asked, the conclusion that "moral values" affected the election's outcome seemed justified.
After all, Moore noted (as Howard Schuman reminded us in his recent op-ed in Public Opinion Pros), the phrase "moral values" must have been sufficiently meaningful and motivating to Bush voters because it differentiated them so sharply from Kerry voters in both the National Election Pool (NEP) and Los Angeles Times exit polls. Furthermore, many critics still believe that asking the question in a closed-ended form probably exaggerated the significance of the "moral values" issue to the American voter. But as Schuman told us, this belief rests on a rather iffy set of assumptions about the supposedly greater validity and virtues of asking the question in an open-ended form. So maybe, after the exit poll smoke has cleared, "moral values" did matter after all. Or so it appears.
As I contended, however, in an essay in the January issue of POP, the whole idea about "moral values" being a driving force in the 2004 election represents no more than an illusion, because it rests on the naïve assumption that exit-poll respondents can actually tell us why they voted the way they did. Instead, when exit-poll respondents tell us that "moral values" or some other issue mattered the most in deciding how to vote, they are not reporting causal influences. Such self-reports of why people voted the way they did represent not causal reasons, but plausible reasons, justifications, or after-the-fact rationalizations for their behavior. Scientifically speaking, respondents cannot be expected to give interviewers univariate, bivariate, or multivariate, causal accounts of the forces driving their behavior. Such forces involve mechanisms and processes beyond their conscious awareness.
Consider as Exhibit A the evidence from exit polls themselves on how much voters can actually tell us about the causes of their behavior. Take, for example the recent debate among pollsters about the influence of the closed-ended form in which the now infamous question about "moral values" was asked. Was it true, as Gary Langer and other polling critics contended, that the sizable percentage of voters selecting "moral values" in the 2004 national exit polls was an artifact of a multiple-choice format that legitimated an otherwise minor factor in the election?
A postelection experiment conducted in November by the Pew Center for the People and the Press suggested that the importance of "moral values" to respondents indeed depended heavily on whether it was asked in an open- or closed-ended form. More than a fourth (27 percent) of the respondents who were asked the question in a multiple-choice form similar to that used in the NEP national exit poll-"Which ONE issue mattered most to you in deciding how you voted for president?"-selected "moral values" as the decisive issue. But only about half as many mentioned "moral values" (14 percent) when the question was presented in an open-ended form: "What ONE issue mattered most to you in deciding how you voted for president?" Furthermore, the 14 percent who volunteered "moral values" on the open form may have done so, in large part, because of the postelection buzz about the supposed importance of the values factor in Bush's victory.
The most revealing thing about this timely experiment, however, is what it tells us about a respondent's inability to tell us about the causal influences on his or her behavior. The results of the Pew experiment clearly demonstrate that the form in which the question was asked-open or closed-had a significant causal influence on what a given respondent believed was the one issue that mattered most in deciding how he or she voted. And yet not a single respondent in the Pew survey volunteered, for example, that "it was the form in which you asked me that question, which legitimized 'moral values' as an option, and which made me more likely to select it as the decisive issue in my vote decision." Nor could they be expected to do so, as such a causal influence was beyond their conscious awareness.
|
