If you thought the health care polling was all over the place I imagine the numbers on repeal will be even more so as they start to come out in the next few weeks.
Because most voters didn't really have a handle on the details of the health care bill the poll numbers, especially in the early stages, were highly influenced by question wording and in some cases ordering. Voters may understand repeal even less than they did the bill itself and because of that I think the poll numbers that come out on it will be highly variable.
Want a poll that shows support for repeal? Ask folks if they want the health care bill passed by the Democratic Congress overturned. The Democratic Congress is very unpopular. If you ask folks if they want the health care bill Barack Obama signed into law repealed you're going to get less support for repeal because compared to his Congressional colleagues he's pretty popular.
Want a poll you can say shows opposition to repeal? Ask folks if they think Congressional Republicans should focus on repeal or move on to other issues. Hard core conservatives might want the party to fight this fight, but I don't think independent voters who are mainly unhappy with Obama because of the state of the economy do.
Question order's a big thing here too. If you ask folks if they support the health care bill first, and then whether they think it should be repealed, you're going to get a higher level of support because people who are opposed to the bill will be likely to say it should be repealed so their answers are consistent. But if you ask folks if the bill should be repealed first, and then whether they support it you may get less support for repeal because a good number of folks opposed to the bill might be ready for Congress to move on from this issue.
I know what's going to happen- both sides are going to have polls whose results they like and they're going to get cherry picked big time. On an issue like this I think it's going to be hard to get a handle on what the public really thinks but the best way to do that, as is usually the case, is going to be to look at all the polls as a whole rather than fixating on individual ones.