Saturday, August 12, 2006

How important were ‘moral values’ issues in the 2004 election? (continued) Effect of same-sex marriage ban initiatives

How the world has changes in two weeks. With all the fur flying following more terror alerts, elderly rejected Democrats converting to reactionary Republicanism (excuse me, I meant Independent Unityism that echoes reactionary talking points), does anyone care about the moral values vote anymore? Whatever, we will continue with two more studies, which look at a slightly different question: could moral values issues have made the difference in the close election of 2004? I was planning on wrapping up this series with this post, but have found a few more interesting papers, so there will be one or two more posts on this topic before the wrap-up.

The first study is DE Campbell and JQ Monson, "The case of Bush's re-election: did gay marriage do it?" The focus is on whether state ballot initiatives that would ban same sex marriage had an effect on turnout. The study looks at county and individual data from the eleven states that had a ‘Gay Marriage Ban’ (GMB) initiative.

Campbell and Monson's county level data are from the secretaries of state offices for the eleven GMB states. The authors look at each state and estimate the relationship between the proportion of voters in each county who voted for the state's GMB and the change in proportion of those voting for Bush between 2000 and 2004. There was a statistically significant relationship in five of the eleven GMB states: AR, GA, MS, OH and OK. So, the conclusion is that GMB helped Bush in some GMB states.

Campbell and Monson also look at individual level data from the 2004 Election Panel Study, a nationally representative survey that followed individual’s opinions from early summer to post-election. They estimated the probability that an individual would vote for Bush as a function of GMB being on the ballot. The individual's age sex and other demographic information, ideological and political identification, feelings of identification with Bush and rate of political contact were also included in the analysis, so the effect of GMB on the state ballot could be analyzed separately from all those characteristics. A GMB on the state ballot appeared to have increased turnout by 5.6% among White Evangelical Protestants, and 5.2% among Catholics, and it increased the probability of Bush vote by 3.3% for Catholics, given their party identification.

Votinglinks’ comments: the study showed that a GMB on a state ballot appeared to have been a significant factor in increasing the vote for Bush, but mainly among White Evangelical Protestants who would have voted for Bush anyway. The most significant effect in turning vote from Kerry to Bush appeared to be among the Catholic vote. Campbell and Monson do not present an analysis of other issues, so the relative importance of moral values cannot be measured.

The second study is by E Donovon, C Tolbert, DA Smith and J Parry, Did gay marriage elect George W Bush. We'll call the authors DTSP for short. DTSP used state level data [correction Aug 14: oops -I meant indificual level data from several states] from PEW Research Center for Press and Opinion Survey with data from just before and after the election in Arkansas, Ohio and Michigan. DTSP estimate several individual regressions for each state that analyze relationships between importance of GMB as an issue, support for GMB, intention to vote for Bush, and actual vote. The analysis took party ID, age, sex, race, education, income and residence and pre vs. post election opinion into account. Individuals' opinions on Iraq war, war on terrorism, and economy were also included in the analysis.

The authors test a complex series of hypotheses, but the important bottom line conclusions are as follows. DTSP estimate that GMB on state ballot increased likelihood that same-sex marriage would be considered in decision between candidates, other factors held constant. DTSP estimated that the effect of GMB on state ballots increased the probability of voting for Bush by 9%. The authors note that in many GMB states, the margin of victory for Bush was so large it was unlikely to have made a substantial difference. In Ohio, GMB did seem to increase Bush's support among African-Americans (+4% over increase nationally in this group), over age 60 (+3% over national increase), HS level education (+2% over national increase) and white evangelical voters (+16% over national increase). The authors conclude that the Ohio GMB could have made the difference for Bush in that state

Votinglinks’ comments: DTSP do not do calculations to compare the issue of same-sex marriage to Iraq war, terrorism and economy. However, the statistical results shown in the paper indicate that these other issues were at least as, or more important than, GMB. So there is certainly evidence that GMB increased turnout and the vote for Bush, but no evidence that this effect was more important than others. Unfortunately, the DTSP do not do an analysis of the relative impact of all the issues.

DE Campbell and JQ Monson, The case of Bush's re-election, did gay marriage do it?
http://americandemocracy.nd.edu/working_papers/files/case_of_bushs_reelection.pdf

T Donovan, C Tolbert, DS Smith, J Parry, Did Gay Marriage Elect George W Bush? http://polisci.msu.edu/sppc2005/papers/fripm/dtsp_sppc05.pdf

previous post in this series