Predictions

"Prediction is hard, especially about the future." -- Author unknown

Quite a lot of people are nervous about the November elections. Some of
them aren't even office-holders. When people get nervous about a
possibility, they resort to trying to gauge its probability in advance
of the event itself.

That's when the fun really starts.

There have been so many wildly wrong predictions in American politics
that a sensible man should know better than to repose confidence in any
of them. Some have been based on bad statistical technique; others, on
correlations that proved not to be as firm as had been believed. (Anyone
else remember "As goes Maine, so goes the nation" -- ?) But all
predictions of any sort, whether or not they prove correct, share a
common, ineradicable failing: Those who want to believe them will invest
in them emotionally, usually to the exclusion of all objective evidence
and good sense.

Just now, several national polls have presumptive GOP nominee Mitt
Romney defeating Barack Hussein Obama on November 6. A few other polls
present the opposite result. The most interesting thing about this split
among the prognosticators is the cleavage between those who choose to
believe which poll. The pollsters themselves are dedicated to producing
a good, trustworthy prediction based on the soundest known techniques
and the most reliable evidence available at the moment. Their allegiants
mainly want a certain result, and will pledge their faith to whichever
poll best matches their preferences.

No doubt the folks whose opinions skewed the famous 1936 Literary Digest
telephone poll heavily in favor of Governor Alf Landon wanted to believe
that result, too. The election decreed otherwise, but the poll must have
been a comfort to Republican partisans and critics of the New Deal until
the balloting was held. A bare twelve years later, poll after poll had
Governor Thomas Dewey defeating Harry Truman, as well, similarly to the
chagrin of those who had put their faith in the polls beforehand.

The most important case of a recent prediction gone wildly wrong may be
the Election Night 2000 fiasco that had the major networks declaring Al
Gore the winner in Florida more than an hour before the polls actually
closed there. The result was massively embarrassing to many a talking
head: in part because no one wants to be wrong on camera, but in equal
part because they were almost undisguisedly Democrat partisans and Gore
backers.

I know, I know: this is all very interesting, a delightful study in the
presumptions of "specialists" and the arrogance of "experts," but so
what? This is what: How many of the Landon backers in 1936, the Dewey
backers in 1948, and Florida voters in 2000 refrained from voting
because they believed a prediction that turned out not to be accurate?

There's an exertion of effort involved in getting up off the couch and
going to the polls -- minor, to be sure, but it can be enough to keep a
"convinced" voter home. Why, after all, should you stir your stumps to
get to the polling place if you're dead certain that your side has it
"in the bag" -- or contrariwise?

That effect is a large part of the reason so many Americans disbelieve
in the objectivity of political polls. The use of a slanted poll to
depress one side's turnout seems just too tempting to resist, especially
since most polling experts are known to be partisans of one sort or
another.

Even if you're inclined to regard this as a "problem," there's no
"solution" applicable to large numbers of people. All you can do is
resolve to pay no attention to the polls...a rather difficult thing when
the latest poll results are displayed in 36-point type at the head of
every newspaper and the top of every Website, and trumpeted
authoritatively at the beginning of every broadcast news program.