T-Paw and the Straw Poll
By I think is as good an account for Pawlenty’s failure as I’ve seen — I’m rarely persuaded by arguments that debates can be decisive turning points, but in this case I think it makes more sense.
Meanwhile, on the question whether the straw polls matter (see Tucker and Silver), what I’d add is that there’s no such thing as reliable data at this point, so more information is better than less. It’s obviously true that the straw poll produces a lot of false negatives and false positives, but it sometimes can provide useful information. This attack on the straw poll actually inadvertently makes this point:
The straw poll has no predictive (or even descriptive!) value. In 2007 Rudy Giuliani and John McCain together received a grand total of less than 2% of the vote. A visitor from Mars would have never guessed that one of these gentlemen was the indisputable front-runner at the time, topping 40% in nationwide polls of Republicans, and the other would eventually go on to win the nomination.
The poor performance of McCain shows the extreme limitations of the straw poll — obviously, some candidates take it more seriously than others, and you have to take that into account. But with respect to Guiliani, this is a lesson in the unreliability of public opinion data at this stage of the race, not the straw poll. Guiliani was not the “frontrunner,” or anything close to it — he was polling on name recognition. It should have been obvious that the pro-choice, radically pro-immigration candidate was drawing dead in the Republican primary, and in picking this up the straw poll was actually telling us something useful.
This year, again, Ames isn’t telling us much. But it did confirm that T-Paw — who spent a lot of time and money and couldn’t come close to Ron Paul — was a disaster. And I think it’s instructive that Perry got more support as a write-in than Romney got while being on the ballot — Perry’s not a lock by any means, but I think it should be clear that he has a lot more appeal to the Republican base.