Marriage polls have infidelity problem
by Paul Demko
Published: October 12,2012
Time posted: 3:16 pm
Tags: Chuck Darrell, Greg Lewis, marriage amendment, Minnesotans for Marriage, minnesotans united for all families, Patrick Egan, Richard Carlbom, same-sex marriage, Stephen Frank
Recent polling has suggested that the vote on the proposed constitutional amendment prohibiting gay marriage is going to be extremely close. A survey released Monday by Public Policy Polling showed that 49 percent of respondents were opposed to the marriage amendment, while 46 percent supported it. That follows a Star Tribune poll released last month that likewise showed support for the measure falling short of the passage, with 49 percent in favor and 47 percent opposed.
Those figures look pretty encouraging for opponents of the proposed amendment to define marriage as the union of one man and one woman. That’s because, in Minnesota, ballot initiatives must win support on more than 50 percent of all the ballots cast in the state. Voters who don’t indicate a preference are effectively casting “no” votes.
Because of this wrinkle in Minnesota law — wrought in the late 19th century at the urging of Minnesota distilling interests fearful of a temperance amendment — St. Cloud State University political science professor Stephen Frank, who oversees the biennial SCSU Survey, estimates that proponents of the marriage amendment will actually need to capture 55 to 57 percent support from those who do vote on the ballot question in order to prevail.
But before marriage amendment opponents pop the champagne, they should consider another wrinkle unique to the issue: In states across the country, pre-election polls have consistently underestimated support for gay marriage bans. A 2010 study by New York University political science professor Patrick Egan looked at a total of 167 surveys conducted in the six months prior to Election Day on 32 different gay marriage ballot measures around the United States. He determined that, on average, the polls underestimated support for same-sex marriage bans by 7 percentage points.
In North Carolina, for instance, the most recent state to adopt a constitutional amendment prohibiting gay marriage, the final pre-Election Day poll conducted by Public Policy Polling determined that the measure was backed by 55 percent of likely voters. It ultimately prevailed with 61 percent of the vote. “North Carolina had a result that was almost perfectly predicted by that paper in the first place,” Egan noted. “For all intents and purposes, our best guess is that that dynamic is still in place.”
No consensus on cause
There is no widely accepted explanation for why this phenomenon has persisted. The most commonly expressed view is that voters don’t respond truthfully to polltakers because they don’t want to be seen as prejudiced.
“It has become increasingly unacceptable to be seen as homophobic,” said Jeremy Kennedy, who served as campaign manager for the campaign opposing the marriage amendment in North Carolina. “So I think a lot of voters, in polling, have been hesitant to give an honest answer about where they are on the issues.”
Chuck Darrell, communications director for Minnesota for Marriage, the main group supporting the marriage amendment, similarly argues that poll respondents provide what they believe to be the “politically correct” answer when surveyed. “When people get into the voting booth, they vote to affirm marriage as between one man and one woman,” Darrell wrote in an email.
But Egan’s study did not find that such an explanation, known as “desirability bias,” held up under close scrutiny. For instance, he looked at six elections in which surveys were conducted by both human beings and automated robo-calls. Research has shown that respondents are more willing to be forthright about socially sensitive topics if the human interaction is taken out of the equation. But Egan found that the electronic surveys proved no more accurate than ones conducted by human interviewers.
Another possible explanation — that voters misunderstand the ballot questions when being polled — also failed to hold up under Egan’s scrutiny. He theorized that if such an explanation were pertinent, the polling gap would be larger in states where citizens are less knowledgeable about politics, but that didn’t turn out to be the case.
“Evidence is scant for the explanation offered so far by scholars and advocates for the gap between polling and election results,” Egan concluded in the study. “It does not appear that voters are particularly reluctant to express their support for banning same-sex marriage, and no support is found for the hypothesis that the gap is due to confusion among survey respondents about the meaning of a ‘yes’ and ‘no’ vote.’”
SCSU’s Frank suggests a different explanation for the polling discrepancy: voter intensity. “Certain groups — for example, anti-abortion people and, I suspect, the anti-gay marriage people — are more likely to follow through,” Frank noted. “I still think there might be a little tiny bit of social desirability … but I think it’s more in terms of salience, intensity and follow-through on the part of those who are opposed to gay marriage.”
Effect mitigated by presidential year?
If that’s the case, however, the polling gap may not be as pronounced in Minnesota this year. That’s because in a presidential year voters across the spectrum are already highly motivated to turn up at the polls. Turnout in Minnesota for presidential contests has averaged 73 percent over the past two decades.
Richard Carlbom, executive director of Minnesotans United for All Families, the main group opposing the ballot initiative, points out that the two most recent referendums on gay marriage, in North Carolina and Maine, were very different from this year’s contest. The 2009 Maine ballot initiative, which was to repeal a gay-marriage bill that had been passed by the Legislature, drew less than 50 percent of eligible voters to the polls. Voter turnout for this year’s referendum on a gay marriage amendment in North Carolina, which was held in May, ran below 35 percent.
“Voter turnout in Minnesota in 2012 is much more predictable than in other states that have voted on this recently,” Carlbom said. “The way people view this question, the way people are actually discerning their position on the freedom to marry for gay and lesbian couples, has been changing rapidly. People are evolving on this question.”
Greg Lewis, a professor at the Andrew Young School of Policy Studies at Georgia State University, has studied public perceptions of gay marriage extensively. He points out that polling tends to be more accurate when voters are asked directly about whether they support same-sex marriage, as opposed to their stance on a looming ballot question. Based on results from 102 different surveys over the past two decades, Lewis has charted what he believes to be the support level for gay marriage in each state.
In Minnesota he calculates the approval rate at 49.9 percent presently, with acceptance growing by 2.7 percent each year. There are gay marriage referendums on the ballot this year in four states: Maine, Maryland, Washington and Minnesota. Lewis is predicting that this will be the year that gay-marriage supporters finally break their nationwide losing streak in such contests, with voters in Maine and Washington the most likely to back same-sex marriage. “I don’t know if Minnesota’s going to be one of them; it’s plausible,” Lewis said. “It’s probably going to be really close this year. And if it were held two years later, it probably wouldn’t be all that close, because it’s changing fast.”