____ 3/27 3/12 2/26Siena is the only non-partisan pollster to have surveyed this race. (The Tedisco camp disputes the result, citing internal polling that still shows them with a lead). Siena's record as a pollster is, to be frank, fairly average, although this particular race is almost literally in their back yard, as NY-20 wraps around the town of Loudonville, where Siena College is located. The usual caveats about polling special elections, which feature very low turnout and are notoriously hard to predict, also apply.
Tedisco (R) 43% 45% 46%
Murphy (D) 47% 41% 34%
Nevertheless, we shouldn't be horribly surprised if the Democrat has turned this race into a toss-up. In the past three Presidential elections, NY-20 been slightly more Republican than the country as a whole -- but only slightly so, and it is arguably becoming less so. And of course, the country as a whole has been Democratic. Note that Barack Obama received the majority of NY-20's vote in November:
VOTE SHARE RECEIVED BY DEMOCRATIC CANDIDATEDemocrats have generally been winning about 60% of the elections for the Congress in districts with PVI's of +2 or +3 over the past couple of electoral cycles; this is how they have earned their majority. So while, on the one hand, the blue team may sneak away with a victory here, they are probably also somewhat guilty of having downplayed expectations.
Year NY-20 USA PVI
2008 (Obama) 51% 53% R+2
2004 (Kerry) 46% 48% R+2
2000 (Gore) 44% 48% R+4
UPDATE: Jeff, in the comments section, asks:
Are we really supposed to believe that only 64% of republicans are Tedisco voters, as this poll claims? Come on.Well, I don't know. But note that the poll also implies that Republicans have a fairly substantial registration advantage in the district. According to the poll, Murphy has a much larger lead among Democratic voters (84-11) than Tedisco has among Republicans (64-27), while the two candidates are essentially tied among independents (Tedisco leads 45-44). If you assumed that the district was composed of, say, 40% Republicans, 40% Democrats and 20% independents, then Murphy would have a 53-39 lead according to the poll's internals. But in fact, the poll only shows Murphy with a 4-point lead. To arrive at those numbers, there have to be quite a bit more registered Republicans than Democrats in Siena's sample.
So suppose that Siena is right -- and I have no reason to doubt them -- that Republicans have a substantial registration advantage in NY-20? What do we know about those Republicans? Well, we know that a fair number of them probably voted for Barack Obama, because Barack Obama won this district, and we know that a fair number of them probably voted for Kirsten Gillibrand, because Kirsten Gillibrand won this district. So crossing over to vote for a Democratic candidate in elections for national office is nothing new for these Republicans; some of them have been doing it routinely.
Personally, I suspect that the relatively high incidence of Republican registration in NY-20 has more to do with politics at the state level. If you're a voter in NY-20, your vote frankly isn't going to matter much for the Presidency and it isn't going to matter much for the U.S. Senate -- the Democrat is going to win those elections. It will matter some of course, for the U.S. Congress. But the most important vote you cast may be for the New York State Senate, which has been almost evenly divided between the two parties, largely along upstate versus NYC lines. That is, the Republicanness of the district may partly reflect its conservatism relative to New York City and will not necessarily translate into elections for national office.