http://en.wikipedia.org/wiki/Interracial_marriage_in_the_Uni...
Note that the "Approve" line doesn't cross the "Disapprove" line until the mid-1980s, long after the USSC had spoken. With same-sex marriage on the other hand, the "Approve"/"Disapprove" cross-over happened a bit over three years ago and the USSC has yet to speak as definitively as it did in Loving v. Virginia.
(One could argue that in invalidating most of DOMA, the USSC has already spoken strongly - if not definitively - in favor of same-sex marriage, and therefore that we should use that decision when making analogies to Loving. I'm not sure about that.)
Not very convincingly, because it did not hold that States cannot deny equal marriage, which would be the equivalent of the holding in Loving.
The "only" here is pure speculation. The fact is that the US Supreme Court has not issued a Loving like decision on same-sex marriage, and that same-sex marriage remains illegal in many states because of that. Therefore, it does not make sense to treat United States v. Windsor (which struck down Section 3 of DOMA) as an equivalent, in the domain of same-sex marriage, to Loving v. Virginia, in the domain of interracial marriage.
The federal government never did something like DOMA on interracial marriage (there was, relevant to Section 3, no federal policy against recognizing state-sanctioned interracial marriages), so Windsor naturally has no parallel among the interracial marriage cases.
I have a strong suspicion that the results would indicate that America itself doesn't really change its mind all that quickly even after the legal changes take place.
Not quite, but close enough: http://xkcd.com/1431/
For an extreme case, consider some issue where opinions are evenly spread out among all the states, and it very slowly changes, at a rate of, say, 0.5% per year. It would take two centuries to go from all-against to all-for at that rate, yet once you pass 50% you'd expect all the states to switch over more or less simultaneously.
The real world isn't so clean, but the same idea applies. Looking at the data, the recent examples of marijuana and same-sex marriage are changing faster, at more like 1-2%/year, but you can still see that the rate of change in state legalization is way faster than the rate of change of public opinion, as one would generally expect.
A couple of random pages with relevant poll data:
http://www.gallup.com/poll/117328/marriage.aspx
http://www.gallup.com/poll/1657/illegal-drugs.aspx