Friday, October 17, 2008

The Bride of the Gallup Anomaly

OK now I have no explanation. The Gallup likely voter filter gives results completely unlike those of any other poll. Currently the Gallup likely voter poll has Obama up by 2% against an average of around 7% !

Matt Drudge has noticed. This has made news around the world. Well at least in Italy where La Repubblica reports the result (and the Zogby result). Drudge's aims are clear. I think that the un-named La Repubblica reporter is just web illiterate. He or she notes that the Gallup likely voter filter excludes the young and stresses that Obama is still ahead. I don't see why someone is paid to generate an article which contains much less information than is easily available on the web.

I have no idea why the Gallup likely voter filter is giving results so different from other pollster's filters. For months Gallup and I have been arguing that it is only useful close to the date of the election, since it is based on, among other things, the question "do you know where your polling place is." Not knowing in August has little to do with voting intentions, in October is looks like carelessness.

The anomaly is sooo strong that Gallup won't stand by their traditional likely voter filter to which they had been loyal for decades. Instead they have chickened out and introduced the Gallup likely voter filter II (or Bride of the Gallup anomaly) which is clearly designed to be less anomalous. Now the justification for not going with the traditional filter is no longer that it is too early for it to work, but that Obama is special. I quote "Gallup's "expanded" likely voter model determines likely voters based only on current voting intentions. This estimate would take into account higher turnout among groups of voters traditionally less likely to vote, such as young adults and minorities."

Very plausible, but 4 years ago Democrats predicted that young voters were going to turn out that time (fear of a possible draft remember ?). Every election is unique although Obama is clearly the uniquest candidate for whom I have ever had the chance to vote.

As an Obama supporter, I am a bit concerned. The traditional Gallup filter has worked well soon before elections. The many other pollsters with other filters don't have all that much of a track record. Gallup may just be panicking because they are all alone out there. That is the race might really be closer than the average poll suggests. Still ahead is ahead and one poll is one poll.

But I just can't wait for "Final Battle for the Gallup Anomaly" on your monitor late night November 4.

No comments: