Political science isn't.

Folks who claim that elections can be confidently predicted using complicated formulas and past voting patterns remind me of an ex-girlfriend who kept trying to convince me that psychology is a hard science, because it relies on statistics.

Look. Statistics might be difficult, but they don't make psychology a hard science. In the end, it all boils down to the DSM-IV, which is based on opinions about observed human behavior – and human behavior is anything but scientific, as the great case study of Life Here On Earth easily proves.

And when the subject is politics, then humans become about as predictable as an ADD kid off his Ritalin who just ate an entire bag of M&M's, washed down with shots of double espresso.

All of which leads us to today's David Broder column in The Washington Post. Yes, you guessed it – the subject is the upcoming election, and the political scientists think they know what's what:

In the run-up to the 2000 election, a number of prominent political scientists embarrassed themselves and their profession by predicting -- on the basis of formulas they claimed to have tested repeatedly with historical data -- that Al Gore would handily defeat George Bush. Undeterred by that experience, some of their colleagues who met here over the Labor Day weekend for the annual convention of the American Political Science Association were once again trying to see into the future -- or in some cases debunking pundits and politicians who offer their own prognostications about the direction of U.S. politics.

But wait, there's less:

It remained for David Mayhew, a Yale political scientist and author, to throw cold water on the whole debate. "There is no emerging majority," he said, at least when it comes to electing the president. Despite what happened to Jimmy Carter and the first George Bush, an incumbent president seeking reelection probably starts with a six-point advantage over his challenger. But when no repeat candidate is on the ballot, "the results are essentially like flipping a coin. The result of the previous election gives no clue."

So what's the point of Broder's column?

Mayer said that Bush may have a slight advantage going into 2004 (a point Teixeira readily conceded), but basically, Mayer said, "another 2000 is what you should expect -- a random result."

That is not a very satisfying forecast, but it is probably more realistic than predictions based on some rigid, pseudo-scientific formula.

Amen.