An intrepid reader of Powerline does the math and finds that it indeed oversamples Democrats by a fairly significant margin.
I’m still looking over the methodology of the Gallup poll, but traditionally Gallup has been a more accurate poll with a better sampling methodology than others.
I’ve also started paying more attention to the Iowa Electronic Markets, which has a long tradition of being more accurate a predictor of Presidential elections than most polls. The current trading on Bush futures is at 52.4¢ while the Kerry future is at 47.5¢ – does this mean that Bush will get around 52% of the vote in November? Before Kerry’s speech I would have said it was unlikely, but now I’m not so sure. I’m still betting on a close race, but by this time next month the face of the game could be very different…
UPDATE: On the other hand, Jim Hood says Gallup has a history of being less accurate than other polls. After looking at the full results of the Gallup poll it does look as though the sample size is a bit on the low side. However, when one makes the distinction between registered and likely voters Bush’s numbers shoot up – which could very well be an indication that the GOP got a boost from the convention with likely voters. It will be interesting to see if these trends continue.
Unfortunately, the polls these days are can be as bad as the liberal media in general with their bias.
It’s also interesting how the polls tend to be reported on TV and radio mostly with just the percentage results, with no mention of sampling methodology. It’s also interesting how polls from different sources are reported as if each has the same credibility.
Bottom line: I don’t trust the liberal media, and so I don’t fully trust their polls.
The I.E.M. is in fact running _TWO_ futures markets. One is the “Winner-takes-all” market. That one indicates the probability of a particular candidate winning the Electoral College, and says nothing about the margin of victory. The other futures market is the “Vote-share” market. That one indicates the eventual share of voters each candidate will end up with.