Monday, September 15, 2008

Latest Polls, Questionable Accuracy

Two state polling results were announced this morning, one in New York and one in Virginia. The people commenting over at Hedgehog are having a field day with them. The accuracy of both polls is extremely questionable.

Virginia first. The poll is from Survey USA and has Obama over McCain, 50% to 46%. But the internals of this poll are almost laughably bad. They suggest that Democrats will outnumber Republicans by about 4% on election day, but historically the opposite of that has been true. They show Republicans in Virginia with relatively weak support of McCain which is also suspect.

On to New York. This poll is from Siena College and has Obama only leading by McCain by five points, 46% to 41%. I'm having trouble locating good internals for this poll, but I would doubt it on the following reason alone. If McCain is only down five in New York he would winning in a landslide across the electoral map in other polls, which we do not see.

That argument also holds for the Virginia poll, too. If Virginia is blue, then Ohio and Florida would be too. While the relative leanings of the states can change based on who the candidates are, where the VP hails from, and specific issues the fact remains that some things remain constant from one election to another. If Texas goes for a Democrat that candidate would have to be so popular that it would be a landslide across all the other states. Similarly if Massachusetts or California favor a Republican, all the traditional battleground states would also be supporting that candidate.

Rasmussen has few state polls due out at 6pm EDT, including one for Virginia. Hopefully his Virginia internals are more sane and we can get a better handle on what is actually happening in that state.

No comments: