Andrew Gelman is stunned that support for gay marriage has increased more in states with already liberal attitudes:
In the past fifteen years, gay marriage has increased in popularity in all fifty states. No news there, but what was a surprise to me is where the largest changes have occurred. The popularity of gay marriage has increased fastest in the states where gay rights were already relatively popular in the 1990s.
In 1995, support for gay marriage exceeded 30% in only six states: New York, Rhode Island, Connecticut, Massachusetts, California, and Vermont. In these states, support for gay marriage has increased by an average of almost 20 percentage points. In contrast, support has increased by less than 10 percentage points in the six states that in 1995 were most anti-gay-marriage–Utah, Oklahoma, Alabama, Mississippi, Arkansas, and Idaho.
Here’s the picture showing all 50 states:
I was stunned when I saw this picture. I generally expect to see uniform swing, or maybe even some “regression to the mean,” with the lowest values increasing the most and the highest values declining, relative to the average. But that’s not what’s happening at all. What’s going on?
Gelman offers two possible explanations: First, gays are more likely to come out of the closet in liberal states, and so the average person in these states knows more openly gay people. Second, politicians in tolerant states have electoral incentives to argue for the liberal position, thus pushing public discourse in that direction.
I suspect both of these factors have some influence, but I think another may be more important. I suspect that public opinion on highly salient issues with strong signalling value tends to be self-reinforcing. A useful framework for thinking about this is Timur Kuran’s model of preference falsification.
Kuran distinguishes between an individual’s public opinion – the views he openly expresses on a particular issue – and private opinion – the way he privately feels about this issue. Since there are social costs to expressing unpopular views, public opinion will be systematically biased towards the social consensus compared to the underlying distribution of private opinions. There is social pressure for those in Utah to express opposition to gay marriage, while there could well be social pressure to express support in New York. If few people share your opinion, you’re more likely to keep quiet or actively falsify your view. Via this mechanism, social pressure leads to homogeneity in public opinion, but leaves private opinion unchanged.
If we think socialization matters in creating private opinion, though, preference falsification will also affect private opinions. When there seems to be a strong consensus on some issue, social learning will bias the underlying distribution of private opinions towards the consensus view relative to the situation without preference falsification. This is obviously pretty closely related to the psychological phenomenon of group polarization.
The preference falsification view is pretty close to Gelman’s hypothesis that liberal states encourage gays to come out of the closet. Social environments which reduce the costs of coming out as gay also reduce the costs of coming out as gay-tolerant. Beyond some threshold, the social payoff from expressing gay-tolerant attitudes becomes positive, which means we’re likely to have closeted bigots rather than closeted liberals. This pro-gay preference falsification will then reinforce pro-gay private opinion, accelerating the liberalization of attitudes in already liberal states.
I heartily endorse Gelman’s call for further study:
We can look at other issues, not just on gay rights, to see where this sort of divergence occurs, and where we see the more expected uniform swing or regression-to-the-mean patterns.
My guess is that issue salience would be a good predictor of divergence. It would also be informative to try and break things down to areas smaller than states. If the preference falsification explanation works, opinion converges within social networks. County or town data should show a stronger effect, as people are more likely to interact with those geographically close to them. Social networking sites could provide some pretty awesome data if you could get entire tightly interconnected networks to share their opinions.