but it's interesting what happens when you compare the differences between polls phrased as "should gay/same-sex couples <some wording about gaining the right to marry or legal recognition of their marriages>" versus polls phrased "should marriage be defined as <some comparison between man-and-woman or allowing same sex couples>".
Well, don't leave us hanging...
Er? Ok. In 2006, Rasmussen did a poll asking the question: "Should marriage be defined in terms of a union between a man and a woman? Or should marriage be defined as a union between any two people including same sex couples?" 68% said is should be defined as a union between a man and a woman and 29% said any two people including same sex couples. . In the same time period, other polls asking questions like "Should marriages between same sex couples be legally recognized?" or "do you oppose/support allowing gays and lesbians to marry legally?" tended to get numbers more like 40 something in favor 50 something opposed. It certainly appears as though directionality of the question itself affected the outcome of the poll by 10-15 points. All those questions are technically asking the same thing, but when it's presented as a change in legal definition of an existing thing, more people opposed it, while when presented as granting or allowing some group to do something (marry in this case) fewer people opposed and more people supported.
Which is kinda exactly what we should expect. It's human nature to lean in favor of helping people in some way, or not opposing people in some way. No one wants to be viewed as opposing the rights of a group or hurting them.
And yes, before you go off on some tangent, I did just grab a sample set of polls from a lolwiki
page. But the samples there were pretty indicative of polls I found scattered around the interwebs, so I'm not worried about it. Feel free to do your own examination of every poll done on the subject for the last 10-15 years if you think the results will be different, but every single poll I've run across which asked the question in terms of what the legal definition of marriage should be differs significantly in outcome to polls which ask whether the group in question should be allowed/denied marriage.
What's also interesting (and which supports some stuff I spoke about earlier) is the polls which do a breakdown of legal marriage versus civil union versus no recognition at all. The totals for marriage and civil union end out being very close (just slightly higher usually) than the "allowed to marry legally" results when that's the only option given. It suggests that when presented with that form of question, people don't want to be mean and deny gay folks something, so they'll tend to pick legal marriage if that's the only option other than denying them that. But when presented with a civil union option, they pick that. Um... Which supports the idea that many of those polls aren't really telling us how many people will vote to expand the definition of marriage to include gay couples. As I mentioned earlier, many who poll with that answer are using it as a substitute for civil unions. They are not given the option in the poll to say what they want, so they pick the closest answer. But when voting, they're not going to grant marriage status to gay couples.
Enough of the civics lesson yet?