I've had a curious thought tonight, concerning how to predict national elections.
This is the information age, and news services are giving us increasing down-to-the minute election coverage. If CNN is to give us 24-hour reports, they have to have something new to report all the time. Fueled by this demand, and backed by ever-beefier computers, predictions of the eventual winner are more finely calculated every year. We at last have the computing power to handle what statistics has always been theoretically capable of, and the connectivity to draw on the necessary balanced samples.
But the predictions themselves are getting no more accurate, even the exit polls which catch their samples minutes after the actual votes. (*cough* *cough* Florida *cough*) The true predictions, the ones coming weeks or even months in advance, aren't worth anything at all; the campaigners have the same statistical tools at hand, and adjust their platforms to accommodate, even as new results are calculated. Add the general ignorance of the average voter. (Oh, no, not me! Or you, of course. Just those other people over there.) Add the increasing centralization of the two parties to pinch the elusive swing vote. Add the escalating complexity of today's issues, or at least the seeming complexity as we gain more access to factoids. Small wonder so many voters shoot from the hip.
So if we're all ? or a heavy plurality of us ? shooting from the hip, and the proper issues are slipperier than ever, then the real predictor of the election should be the gut feelings of the voters towards the candidates themselves. Heck, that seems to have been the entire platform of Bush minoris. And if that's true, we should get pretty good results from asking questions concerning the ?jerk factor.? Such as:
- Which candidate can you most easily imagine double-parking on a busy city street?
- Which candidate do you think picks his nose more often in public?
- Which candidate can you picture being less popular as a child?
- Which candidate do you consider more likely to stiff a waitress her tip?
- Which candidate is less likely to wash his hands after going to the bathroom?
- Which candidate would more readily talk aloud to friends during a movie?
- Which candidate is less likely to clean up after his dog?
- Which candidate would more readily toss cigarette butts out his window?
- Which candidate would you more readily expect to have been teacher's pet?
- Which candidate can you more readily picture stiffing a waitress her tip?
Questions on the poll would need to be carefully designed not to tie readily to political issues, or even general party alignments. For example, ?Which candidate would sooner make fun of a single mother on welfare?? is heavily biased. ?Which candidate would more likely try to weasel out of military service?? might be fair in some elections, but wouldn't have been in '96.
I'd be willing to put a buck on this poll being as reliable as anything else out there, including exit polls. Given that the Economist outperformed ABC, CBS, and NBC by measuring hemlines and counting rabbit populations, it can't do much worse.