THE EXPERTS PREDICT A BUSH WIN

According to this article from the National Review, a group of political scientists have published their predictions for the upcoming election. They predict a Bush victory, at least through an analysis of the total popular vote.

The authors and co-authors of the seven models include: Alan I. Abramowitz of Emory University, James E. Campbell of SUNY-Buffalo, Robert S. Erikson of Columbia University, Thomas M. Holbrook of the University of Wisconsin-Milwaukee, Michael S. Lewis-Beck of the University of Iowa, Brad Lockerbie of the University of Georgia, Helmut Norpoth of Stony Brook University, Charles Tien of Hunter College, and Christopher Wlezien of Nuffield College, Oxford University.

As Professor Campbell notes in an introductory essay, most of the models share some common features. First, they generally incorporate some measure of public opinion toward the president, such as job approval, into the model. Second, most use measures of the pre-election economy to help generate predictions. Finally, many of the models incorporate some measure of "incumbency advantage" as a part of their prediction.

While all the models diverge after this point in terms of the variables they include to generate predictions, each spits out a percentage of the two-party vote for President Bush. Here are the results, with each scholar listed next to the percentage of the vote he predicts President Bush will win:

Abramowitz: 53.7 percent
Campbell: 53.8 percent
Wlezien and Erikson: 51.7-52.9 percent
Holbrook: 54.5 percent
Lewis-Beck and Tien: 49.9 percent
Lockerbie: 57.6 percent
Norpoth: 54.7 percent
Median Forecast: 53.8 percent

You can read the whole thing by following the link. One of the problems with these models, like any model of human behavior, is that it relies on making the correct assumptions about who turns out and whether or not those that do turn out will repeat past behavior patterns. The author points out that these same guys predicted a substantial Gore win in 2000.