Andrew Sullivan has been blogging up a storm in the wake of the Iowa Supreme Court's decision to legalize gay marriage, as well as the Vermont Legislature's decision to legalize gay marriage. Today Sullivan, citing Nate Silver, chides National Review for misusing polling data (NR claims that Gallup shows little movement in the 2000s, while Silver's more fulsome data show fairly consistent improvement over time), and claims that "[i]n California, everyone concedes that support keeps going up - as the 2008 vote was far closer than the 2000 one":

Setting aside completely the merits of the gay marriage debate, I'm not sure Sullivan is correct on these two points. First, while Silver's LOESS regression shows a pretty smooth trendline, LOESS regression tends to show smoother trendlines than, say, rolling averages, especially if you use a large smoothing parameter. Even then, Silver's data show a flattening of support in 2008-2009.

For a different perspective, you can view Pollster.com's estimate, which shows a less-dramatic rise in support. This estimate only goes through 2008 (when Silver begins to show a flattening), but it shows support today at 40%, and opposition about where it was before the Supreme Court of Massachusetts weighed in. This is consistent with NRO's argument, though NRO also seems to ignore that a large reason for the "flatness" is the step back that occurred in the wake of the Massachusetts decision). To be sure, same-sex marriage seems likely (based on the information we have about people's attitudes today) to win out in the end – if for no other reason than attrition. But its arrival may be farther off than Sullivan anticipates, and support has been growing rather slowly, and in fits and starts, across the last decade.

More importantly, while I would think that support for gay marriage has grown in California, I would be wary of reading too much into the fact that the 2008 vote was far closer than the 2000 vote. With Prop 22, gay marriage proponents were seeking to preserve the immediate status quo and were not actually faced with the prospect of stripping anyone of a right that was more than theoretical. Neither was true for Prop 8, meaning that at least some of the tightening was likely due to the differing factual contexts for Prop 8 and Prop 22.

It is also probably worth noting that the next time gay marriage comes up for a vote in California, these contexts will be more akin to those present for Prop 22 than for Prop 8, and that in the future, proponents of gay marriage will have the added disadvantage of being on the “yes” side of a ballot initiative, which is generally thought to be the more difficult side to be on.