Thursday, August 26, 2004

More on Polls

The problem with polls is that you don't know whose to believe. As reported below, the Economist/You Gov poll has Kerry with a comfortable lead. On the other hand, the very recent LA Times poll has Bush ahead by 3% (49 to 46 with 5% undecided). There appears to be no question that the Swift Boat Liars For Butthead has had an impact on the electorate. That of course is predictable but also lamentable. It is a subject, however, that I will refrain from addressing further, at least at this juncture.

Back on topic, what can we glean from the polls. Most polls, along with averages of polls still have Kerry in the lead, both in the popular vote and in electoral votes. (see, and see Electoral, Tis' Poll of Polls, Polly's Poll).

Other things on the plus side include Bush's decreasing approval rating (see Professor Polkatz Pool of Polls ).

On the minus side, most quantitative analytical models (see down the page) show Bush winning the election (though most of these are months old). Moreover, the Iowa Electronic Markets, a unique market where you "bet" real money on who will be the eventual winner, still has Bush with a small lead.

Lastly, if you are interested in bias in the polls, Polkatz has a good analysis on this. As you could guess, Fox News is clearly biased toward Bush. Zogby, on the other hand appears to be biased toward Kerry. His conclusion is that the rest are pretty much trustworthy. He had a great graph on potenial bias, but for some reason I can't find it now.

If you are really interested it is best to look at the polls done state-by-state and view how they have changed over the past 6 months. Real Clear Politics is a place to start for this type of analysis.

Good luck and good hunting.