[Ninong]

Are you as confused as I am by all the polls that are taken on exactly the same dates but seem to come up with very different numbers?

Ever notice that some of the polls state that they called "registered voters" but most of them say that they polled "likely voters?" OK, fine but how does that explain the huge difference in numbers between the various polls taken on the same dates that polled "likely voters?" Hmmm... Interesting question.

It turns out that each of the different polling organizations has their own method for determining what constitutes "likely voters." They then take the raw data and "adjust" it to account for what they believe will more closely reflect the actual voting patterns. There is only one little bitty problem there. Virtually all of them think that Republicans are "more likely" to vote than Democrats even though that was not true in the last two presidential elections. In fact, the Gallup Poll is still using an "adjustment" based on their outdated belief that Republicans are 6-8% more likely to vote in a national election than Democrats. Wow! No wonder we are getting all these different meaningless numbers. I decided to look into this when I noticed a couple of weeks back that Gallup was showing a 13% advantage for Bush at the same time that all of the other polls reflected a much closer spread.

I can tell you that if that's the way they are calculating their results, some people will be in for a huge surprise the night of Nov. 2nd. That's exactly what happened in Louisiana's recent gubernatorial election. The Republican candidate was favored by 3% the night before the election but the Democrat won the election. All of the polls were based on "likely voters" but someone forgot to tell that to the many "unlikely" voters who showed up to vote.

If I remember correctly, the polls all showed Bush with a lead of 2-3% over Gore the night before the 2000 presidential election, yet Gore won the popular vote by more than half a million. That doesn't tell you much about how the electoral college will break but I would think that if Bush is only 2% or less ahead of Kerry in the published polls of "likely voters" the day before the election, he's in trouble.

I don't know who will win but I do know that the polls are even less reliable than I once thought, even when all they are trying to predict is the total popular vote.

As Benjamin Disraeli once said, "There are lies, damned lies, and statistics."