Can Pundits Actually Prognosticate? Answer: Mostly, No

Predictions

from the nice-to-see-some-lookback dept

I’ve discussed in the past how loath the tech industry often is to ever look back at analyst “predictions.” The big research firms come out with all sorts of ridiculous predictions, and no one ever goes back and figures out how accurate they were. It seems that the same thing is frequently true with political prognosticators, so it’s interesting to see a Hamilton College public policy class analyze the predictions of 26 political pundits over the period of 15 months (September 2007 to December 2008) and measure how good the pundits were. It turns out most were “no better than a coin toss.” I’m not entirely convinced of the methodology, since it seems to make use of some subjective analysis from the description, but for the politically minded, it’s at least interesting to note that the most “accurate” pundits all fall on the left of the traditional political spectrum, while the least successful tended to fall on the right. I do wonder how much of that has to do with the timing (the period covered the financial decline and the Presidential election). It would be interesting to see a similar test run during a different period of time as well. I wonder if a similar analysis, say, prior to the election of a Republican president, would have turned up the opposite results. In other words: was there a fundamental quality in the predictions, or was it just that “the winning team” looks smarter in retrospect? Either way, it’s still great that people are going back and looking at how well some of these prognosticators did.

Filed Under: politics, predictions