A Few Extra Thoughts on 2004 Elections

I choose to wait a few days before commenting on the 2004 elections. After this, then it is on with the rest of my life.

Polling

I am not going to dwell on the exit polling fiasco but instead review the last polls by both Gallup and Zogby. The goal is to examine some of the miscalculation of these pollsters and the consequences of those mistakes.

Many conservatives are quite giddy at Zogby being proven wrong but this is one conservative who will cut Zogby a break. John Zogby made two serious miscalculations. First, he underestimated Bush’s strength, in particular dealing with social conservatives. When asked about Karl Rove’s missing 4 million evangelical Christian voters in 2000, Zogby dismissed this as nothing more than fantasy. This proved not the case. Most pollsters work from particular models and in many cases, uses past elections as their guides. By dismissing the possibility of Rove being right, Zogby missed an important source of new voters that went undetected in his poll.

The second mistake that Zogby made was the same that Gallup made, assuming that many of the undecided or new voters would overwhelming go for Kerry. Most polling models on a Presidential levels assume that the challenger may get a small additional boost from undecided voters but not an overwhelming boost. More often than not, both challenger and incumbents will nearly split this vote. Other polls like Rasmussen, Harris and the Battleground polls did allocate the undecided based on past historical models and proved more correct in their analysis.

What would Zogby and Gallup poll looked liked if they followed the other polls lead? They would have been more accurate in their final poll. Gallup, for example, had Bush up by 2 before the pollsters allocated their final numbers. If they had based allocation on past voting history, Gallup would had Bush winning by a 50-48 margin. While this would still have shorted Bush, it least would have been within the margin of error and Gallup would have nailed Kerry final vote perfectly. (I will add that Gallup allocated far more to Nader than what Nader actually received, thus overestimating Nader strength.)

I suspect that Zogby own Democratic background and being on the limb after predicating a Kerry victory in the spring skewed his interpretation of the data.

With media obsession over the Democratic registration machine, they overlooked the registration drive put on by the GOP and many churches. In Ohio, churches registered voters and many of these voters voted for Bush. The conventional wisdom that new and undecided voters would break Kerry’s way was wrong. And this affected many pollsters’ final polls or thinking.

Article Author: Tom Donelson

Visit Tom Donelson's author page — Tom Donelson's Blog