Phil Klinker at the Monkey Cage jokes:
Tom Edsall has a good overview of the election predictions offered by various political scientists. The consensus? A big win for Obama, unless he loses.
In truth, though, what's striking about the roundup is how little real disagreement there is. First there's Alan Abramowitz, Tom Mann, and Larry Sabato and their essay "The Myth of a Tossup Election" arguing that Obama will win easily. James Campbell, on the other hand, thinks it'll be close. Then we learn that "Vanderbilt's John Geer, in turn, is by no means convinced that McCain will lose as badly as Adlai Stevenson in 1952." Robert Y. Shapiro says it'll be close, Michael S. Lewis-Beck and Charles Tien argue that Obama will win but only narrowly because his race will turn off a segment of the electorate, Helmut Norpoth has a model that predicts a narrow Obama win, and then Sandy Maisel agrees with the Abramowitz/Mann/Sabato analysis.
Basically, predictions range from Obama winning narrowly to Obama winning easily with one guy calling it a toss-up. In other words nobody thinks McCain is likely to win.
My take on this is that the election is more unpredictable than the "Obama in a landslide" crowd thinks primarily because the fundamentals themselves are unpredictable. I don't think it's likely that there'll be a marked turnaround in economic conditions over the next few months, but macroeconomic trends are famously hard to forecast. Similarly, none of us really know what's going to happen in Iraq over the next few months. Elections are primarily determined by the fundamentals, and thus are in that sense more predictable than journalists usually imply, but it's not as if the fundamentals are all that easy to predict.