Wednesday, October 1, 2008

Can We Believe the Polls?

Over the last month I've heard more than a couple of Political Scientists, as well as pundits and other political commentators mention that they don't trust this years presidential polls.

Some blame the increasing use of cell phones, others simply can't believe that almost half of the electorate are in support of John McCain.

A professor in my department brought up an interesting point last week. To date he has not heard a single American Politics specialist who believes McCain is going to win.

So the question has to be asked. Does the academy and the media have such a strong bias towards Obama that it is incapable of accurately assessing whether or not he will win the election?

Can we believe the polls? I'd love to know what you all think... What reasons do we have to believe that they aren't accurate?

(and of course we're talking about reputable, time-proven polls, i.e. from survey centers like Gallup... not surveys with heavy respondent bias like those CNN text messaging polls).

1 comment:

tom said...

hi alex

I've really enjoyed reading your blog so far this election.

One big reason why I think most people interested in American politics are predicting an Obama victory is the Senator's performance, relative to polling, in the primaries and caucuses. Belying the erstwhile "Bradley effect", Obama's vote share exceeded, on average, the predictions of pre-election polling.

The relative extent of this effect seems to be related to the racial composition of different states, and the challenges pollsters had in predicting who would comprise the eventual turnout. A clear explanation of this claim is here

So, when we conflate polls' underestimation of Obama's true support with the myriad structural advantages he possesses, it suggests it's a very speculative year to back the gop.