Justice Scalia argued in his dissent in Romer that:
This Court has no business imposing upon all Americans the resolution favored by the elite class from which the Members of this institution are selected
This fits a more broadly held perception that laws are more progressive on LGBT rights than voters would like them to be because elites are more liberal. A new working paper by Katherine L. Krimmel, Jeffrey R. Lax, and Justin H. Phillips (all Columbia) shows that this presumption does not hold for members of Congress. Indeed, the results imply the reverse. The authors use a method developed (in part) by our own Andy Gelman to estimate public support in Congressional districts for specific policies related to gay rights. They then compare public opinion to the actual roll-call votes of Congressmen. One of their key findings is that there is a conservative bias in Congress: on average public support in a district for a liberal policy towards gays needs to be well over 50% for a Congressman to vote in favor of that policy. Moreover, when opinion in a District changes, Democrats are much more likely to change their votes than Republican Congressmen.
The figure below illustrates their findings for the Don’t Ask Don’t Tell issue. There were only small pockets in the country where majorities opposed changing the policy. Votes against it in Congress were much more numerous than we would expect from public opinion. The paper is very rich in empirical detail, so go read the whole thing.