On the day before the election, Republicans are largely reduced to hoping the polls are wrong–not a good position to be in. But Zombie, a very smart resident of the Berkeley area best known for her photojournalism, delivers the most sophisticated argument yet that the polls could, indeed, be skewed against John McCain.
Zombie relies on a couple of famous psychological experiments, one involving the German horse Clever Hans, who seemed to be able to do mathematics. Her theme is that the Left has blundered by inflating the extent of Obama’s support; that most people who make calls for pollsters are Obama supporters; that a certain number of respondents deliberately conform to social norms and the caller’s desires by saying they support Obama when they don’t; and that therefore the polls aren’t a reliable indicator of the true level of the candidates’ support.
Zombie’s argument is more sophisticated than conventional references to the Bradley Effect, but I don’t buy it. It’s true that social pressure to say that one supports Obama is high, with implicit threats of “racism” leveled at those who buck the Obama wave. But that’s a matter of degree: it’s always true that poll callers are mostly Democrats (like exit pollers) and that the socially “right” answer to poll questions is that the respondent intends to vote for the Democrat. If Zombie’s theory were right, we should have a track record of Republicans pretty consistently outperforming polls. To my knowledge, no such record exists.
So while it’s possible that late-deciding voters may swing toward McCain, and that for other reasons, too, McCain might outperform the polls, I don’t see any strong reason to expect that to happen.
SCOTT adds: Ann Coulter takes a gimlet-eyed look at past presidential polls in “Eighty-four percent say they’d never lie to a pollster.”
To comment on this post, go here.