It turns out that you can build a very effective model by including just three variables:
1. The year in which the amendment was voted upon;
2. The percentage of adults in 2008 Gallup tracking surveys who said that religion was an important part of their daily lives;
3. The percentage of white evangelicals in the state.
Race, education, political party, and every other variable he looked at either didn’t have an effect or duplicated the effect of #2.
Marriage bans, however, are losing ground at a rate of slightly less than 2 points per year. So, for example, we’d project that a state in which a marriage ban passed with 60 percent of the vote last year would only have 58 percent of its voters approve the ban this year. …
So what does this mean for Iowa? … [T]he model predicts that if Iowans voted on a marriage ban today, it would pass with 56.0 percent of the vote. By 2012, however, the model projects a toss-up: 50.4 percent of Iowans voting to approve the ban, and 49.6 percent opposed. In 2013 and all subsequent years, the model thinks the marriage ban would fail.