Referendums in Minnesota and North Carolina both look likely to ban gay marriage. But if New Jersey votes, the story changes
The ninth US circuit court of appeals' overturning of California's Proposition 8 ban on same-sex marriage was greeted by many in the LGBT movement as a turning point. Until the moment same-sex marriage may be ruled a constitutional right by the US supreme court, however, proposals for same-sex marriage bans like Proposition 8 will continue to appear on ballots throughout the United States.
Minnesota and North Carolina already have constitutional amendments seeking to ban same-sex marriage on the ballot this year. New Jersey may also have a ballot referendum on whether to legalise same-sex marriage in the state.
First, let's examine polling for Minnesota's marriage amendment. The last three pollsters to survey Minnesotan voters have found that more people support a ban than oppose one.
Support for the amendment is consistent across the three surveys, which differ in methodology, at 47%-48%. Public Policy Polling and SurveyUSA use automated (or interactive voice response) technology, while the Star Tribune poll employs live telephone interviews. This consistency gives us confidence that the pro-ban side very nearly commands a majority of support, but is not quite there.
Opposition to the ban differs between the three pollsters. Public Policy Polling and the Star Tribune find that 43%-44% do not want to a constitutional amendment to outlaw same-sex marriage, but SurveyUSA finds opposition only at 39%. The trickiest part of trying to determine if the amendment will pass is that SurveyUSA, like actual Minnesota ballot initiatives, provides a "not voting" option. SurveyUSA finds that "not voting" receives 10% support. Unlike most other states, Minnesota initiatives need a majority of those who cast a ballot in a given election, not those who cast a vote on the initiative. More people can vote yes than no, but if the proposition does not get a majority of ballots cast, then the ban will fail. According to SurveyUSA, the initiative is actually quite close to failing (39% oppose +10% not voting = 49%).
The question is whether the 10% of voters who chose "not voting" will actually not vote. I believe half of them probably will end up voting on the ban. As Nate Silver pointed out, only about 5% of people usually choose not to vote on ballot initiatives in Minnesota. If we added the 5% not voting to the 39% saying they would vote no in the SurveyUSA poll, we end up with 44% opposing the ban (identical to the Public Policy Polling and Star Tribune survey). In fact, the percentage not voting could be even lower for a controversial and highly publicized issue such as same-sex marriage.
History tell us, though, that we cannot take the polling at face value. As I have previously noted, Patrick Egan and I found same-sex marriage ballot questions tend to do worse on election day than pre-election polls predict. In fact, Egan found the bans ("yes" side) picked up, on average, 7% support from the final polls, while the "no" side picked up no appreciable support. In Minnesota, 7% additional support to the "ban" side would push it to a majority (with 55% support).
My belief in the relative strength of support of the ban is bolstered by the demographics of the state. Nate Silver has developed a model based on religiosity and year of the amendment to predict support for same-sex marriage bans. I have modified the model to include a variable for the social conservatism of a state. Both Silver's and my model show support for the Minnesota amendment in the mid-50s; the same as the polls plus the average 7% "bonus" Egan demonstrated. The error on both of these models is relatively large, to the extent that the amendment failing would be within the margin of error.
Silver also argues persuasively that a rapid rise in the percentage of people supporting same-sex marriage in polls makes the model potentially inaccurate. His "accelerated" model finds that the amendment would actually be slated to receive less than 50% support and would thus fail. I'm somewhat suspicious, though, of a model that uses polls, which usually understate opposition to same-sex marriage, to argue that opposition is decreasing. To my mind, the original model predicting the amendment would pass, which matches the poll data pretty well, is probably closer to the truth.
North Carolina's ban is more straightforward, but still somewhat tricky, to predict. The only pollster, Public Policy Polling, to consistently survey with the correct amendment wording has the ban passing with 56%, to 34% against. The percentage supporting the amendment, however, has decreased in each of the last three polls, from a high of 61%.
Many North Carolinians are apparently unaware that this particular amendment bans both same-sex marriage and civil unions (unlike Minnesota's), whereas 56% of North Carolinians support some form of recognition for same-sex couples. It seems possible that as voters become more well-versed in what the actual amendment will do, opposition to the broader ban will rise. Still, those in favor of the ban have a lot of leeway, especially considering the average 7% "bonus" for the ban side, before the amendment has any chance of failing.
The demographic models agree with the polls. My model estimates 57% supporting the ban, compared with Silver's original 59.5%, and his "accelerated" 53.5%. While it's possible that there could be a large enough modelling error to allow the result of the ban failing, it is unlikely.
If North Carolina's polling suggests a ban is a strong likelihood, there is one state that could prove a real showdown for an initiative on same-sex marriage: New Jersey. There have been recent efforts in the Garden State to put a same-sex marriage initiative on the ballot. Polling shows a majority (52%) of New Jerseyans support same-sex marriage. Silver's unaccelerated model, and mine, both estimate that a referendum to pass same-sex marriage would barely fail; but in 2013, it would barely pass. Silver's accelerated model projects an easy pass for same-sex marriage. Considering the polling and modeling, New Jersey could be the first state where same-sex marriage was approved by the voters – if they get a referendum, as proposed by Governor Chris Christie, who actually opposes a same-sex marriage law.
Indeed, New Jersey would probably be the first in a long line of states where voters lift bans or vote in favor of same-sex marriage. Time is on the side of those supporting same-sex marriage. Most young voters are in favor of same-sex marriage, and those against it are dying out. The ballots in 2012 will probably provide a short-term boost to those against same-sex marriage, but it will be fleeting; the historical trend is not on their side.