Nate Silver continues to urge caution:
[T]he margins matter as much as the victor: if Ms. Hochul wins by a single vote, that tells us almost exactly the same thing as if she loses by a single vote. Also, Ms. Hochul’s share of the vote matters: the lower Mr. Davis’s vote goes, the more we can read into the results. As I noted two weeks ago, if Ms. Hochul finishes with a vote share in the mid-to-high 40s, that would be consistent with how Democrats performed in the district in the strong Democratic years of 2006 and 2008 and is a result that Democrats could be pleased with.
Jonathan Bernstein seconds:
[W]hat everyone wants to know about NY-26 is how Medicare plays. And, if the polling is good enough, it should be possible to get at least a bit of information about that. But Silver is also right that it's really easy to read way too much into those data (for example, as he points out, those who would have supported the Democrat anyway may say that it was because of Medicare because that's what she's talking about). And what Silver doesn't mention is that knowing how Medicare plays in one place in spring 2011 doesn't necessarily predict how it will play elsewhere in November 2012. It's information, and you always want to add information, but you also have to be careful not to pay more attention to the data you happen to have than to the (potentially far more important) data you don't happen to have.
Screenshot from a vicious anti-Corwin parody website.