Wednesday, July 19, 2006

How important were ‘moral values’ issues in the 2004 election? (continued)

This is the second in a report of research on reasons for the 2004 Bush victory, with emphasis on the effect of ‘moral values.’

G B Lewis analyzes data on individuals from a March 2004 LA Time opinion survey. The poll was not designed to be nationally representative, and does not include peoples’ votes, obviously, but the author references a study that finds that the data from this LA Time survey strongly resemble those of exit polls on election day. The study analyzes individual preference for Bush as a function of the following issue variables, which are expressed as

--Iraq war worth it
--economy doing well
--support of civil unions
--support of same-sex marriage

The study includes the following variables in the analysis, so that the effect of positions on the issue variables can be estimated while holding other characteristics constant: ideology, regular church-going, Born-again Christian conservatism, Party ID, religious affiliation, age, education sex, income, race/ethnicity, self-identified LGBT. Opinion on abortion was not included in the analysis.

The way the author modeled the preference for Bush vs. Kerry as a function of the issues is complex, so the results are presented in terms of a simulation. The results are presented as the net change in probability of voting for Bush, for the ‘average' voter. The results are

--belief that war in Iraq was worth it: +49 percent points
--belief that economy was doing well: +27 points
--support of same-sex marriage: -25 points
--belief in Bush’s approach to terrorism: + 21 points.

A variation of the model was estimated by adding information from the survey on support for a constitutional amendment was added, then same-sex marriage supporters were more likely to prefer Kerry, but opponents were not more likely to support Bush. Civil unions still had no effect.

The main difference between this study and Hillygus and Shields is that terrorism is much less important here, and same-sex marriage almost as important as the economy. However, this study still finds that civil unions and same-sex marriage is less important than the Iraq war or the economy.

Votinglinks’ comments: Obvious weakness of the study is the use of data from about seven months before the election. The time lag may explain the results on terrorism since the full effect of the terror alerts and campaign strategies would not have had an effect. Lack of abortion is also a weakness. A strength is that the author does do a simple simulation to estimate the impact on the probability of voting for Bush or Kerry, which is more reliable for these kinds of statistical models than simply reading off the results the computer spits out.

Just so you know where this all is going: Votinglinks has found about fifteen studies on this topic. Two or three more that seem the most reliable and present their results in the most easily interpretable way will be summarized in detail . Then will briefly summarize the rest, and make best attempt at overall conclusion in a concluding post.

G B Lewis. Same-sex marriage and the 2004 presidential election. PSOnline (PS: Political Science & Politics) April 2005; 195-199.
http://www.apsanet.org/imgtest/PSApr05Lewis.pdf
(may need subscription depending on from where you access)

previous post in this series