That’s the question of the week after Gallup’s national poll two days ago showing 50 percent support for SSM was abruptly followed by North Carolinians voting overwhelmingly to ban the practice. That makes 42 states that now define marriage as between a man and a woman. How do we get from that national point A to the state level’s point B? Can’t be that Gallup is wildly off; Pew got similar numbers when they polled this issue too.

Ross Douthat speculates:

In a 2010 paper, for instance, the New York University political scientist Patrick J. Egan compared polling in advance of state same-sex marriage referendums to the actual results, and found that

“the share of voters in pre-election surveys saying that they will vote to ban same-sex marriage is typically seven percentage points lower than the actual vote on election day.”

That seven-point gap between appearances and reality may help explain why same-sex marriage supporters lost referendums they expected to win in liberal states like Maine and California. And it explains why a savvy White House might take polls suggesting that the issue is a political winner with a very large helping of salt.

“Voters who say they support it when Gallup and other pollsters come calling can behave very differently in the privacy of the voting booth,” he writes, implying some sort of “Bradley effect” on this issue. Eh, I’m skeptical of that. One big reason why is that the Bradley effect itself is almost certainly a myth. Read this insightful LA Times op-ed from 2008 to see why — or better yet, consult the results of the 2008 presidential election. The final RCP poll average that year was Obama by 7.6 points; Obama ended up winning by 7.3 points. It’s hard to believe that any significant number of people would so fear being judged a racist/sexist/homophobe by some random pollster that they’d actually lie to do so. If you doubt me on that, go dig up one of those freak-show polls that PPP conducts from time to time asking people whether they think Obama is the antichrist or a werewolf. People aren’t shy about copping to “disfavored” responses. But wait, you say — after the blacklists in California of people who donated money to support Proposition 8, might not poll respondents be extra fearful of being found out as opposing gay marriage? Well, yes, if they know about those blacklists, which I’d guess maybe one in 100 do. You guys know about it because you follow politics every day. The average voter likely would have trouble picking Joe Biden out of a line-up. Just not enough knowledge on average to fear reprisals.

Rachel Weiner has another theory for the poll disparity:

Turnout is also a factor. Older voters tend to vote in higher numbers, and there’s a stark age divide on gay marriage.

As Columbia Political Science professor Jeffrey Lax wrote in 2009: “If policy were set by state-by-state majorities of those 65 or older, none would allow same-sex marriage. If policy were set by those under 30, only 12 states would not allow-same-sex marriage.”

Primaries, like the one in North Carolina last night, are particularly low turnout affairs— giving opponents to gay marriage the edge.

Yeah, I think demographics are the key. According to the Pew poll I linked up top, fully 56 percent of seniors still oppose gay marriage. Among voters 18 to 29, it’s just 30 percent. Grandma and grandpa can be guaranteed to turn out while junior really can’t, so it’s grandma and grandpa who ultimately make the laws. (See also: Entitlements.) Beyond that, the national polls are typically of adults, not actual voters. It may well be that the average American adult shrugs at gay marriage, but shruggers tend not to make it to the polling place. In all likelihood opponents of gay marriage are more motivated, which means they’ll be overrepresented in the voting booth. And finally, it could be that there’s a slight NIMBY problem at work in state votes as opposed to national polls. Some people, when asked whether they support gay marriage in the abstract, might say “sure” because they’re dealing with a hypothetical. When suddenly they’re not dealing with a hypothetical but rather the prospect of lots of gay couples moving to their state to marry if no ban is enacted, the calculations for some fraction of those voters might change.

Any other theories? Given the trendlines on this issue across demographics and the prospect of more young voters coming onto the rolls, I think the gap between the national polls and state votes will disappear in time. How much time, though, I don’t know.