Evil demon

By M.S.

A NEW poll by Public Policy Polling (PPP) asks respondents the charmingly reflexive question of whether they think pollsters are "deliberately skewing their polls this year to help Barack Obama, or not". One might wonder: but will respondents who see the results of this poll believe the pollster has faithfully reported the data, or will they assume they're just skewing the poll on whether or not people think they're skewing the polls? You could imagine this sort of thing launching a thousand logic-problem-type polling questions and spicing up the campaign season considerably. ("Pollsters always lie. I am a pollster. Who will I report you are going to vote for?" And so forth.) Predictably, the PPP poll results break down on party lines. By a 71-13 margin, Republicans think pollsters are deliberately biasing the polls in favour of Mr Obama. By a 65-14 margin, Democrats think they aren't. The overall result is a near-even split, 42-40, in favour of the bias thesis, which isn't too far off the near-even split along party lines in the overall electorate.

I don't want to spend any time at all explaining why the idea that pollsters overall are deliberately tilting their results in Mr Obama's favour is ridiculous. It's ridiculous, and the fact that a plurality of respondents in a large survey say they believe it to be true is just the latest addition to the colossal pile of evidence showing that we human beings spend our lives inhabiting an ornate palace of self-flattering delusions. That said, three points leap to mind.

Point one: It could be a moderately healthy development for people to lose faith in the accuracy of election polls, if that distrust leads polls to have less influence on their voting behaviour. Because voters tend to vote for candidates who seem like winners, and the media treats candidates who are ahead in the polls as winners, candidates who do well in polls tend to benefit from something of a positive feedback loop. Diana Mutz, a political scientist at the University of Pennsylvania whose work on the so-called "bandwagon effect" is widely cited, thinks the mechanism here isn't so much tribal group identification as a cognitive shift: the information that a candidate is popular prompts people to consider arguments for that candidate which they wouldn't otherwise have considered. Now, I don't have a complete normative theory of what factors people should base their voting decisions on, but I'm pretty sure there's little rational benefit to being influenced by this kind of self-reinforcing crowd dynamic. In a primary or multiparty election, sure, jumping on the bandwagon rather than wasting your vote on a marginal candidate may be a rational tactical-voting decision; but not in a two-candidate race. If people start disregarding polling data entirely and voting for candidates based on their programmes, their party affiliations, or even something as nebulous and potentially misleading as their perceived character, that's probably all to the good.

Point two: Nonetheless, the gain here would come at a price that's far too high. Generalised popular mistrust of the reliability of independent institutions and information gatherers makes our society stupid. Rational thinking demands that people accept the available evidence, and not speculate that they may be brains in vats whose perceptual worlds are actually artificial productions of an all-powerful Cartesian evil demon. "A wise man proportions his belief to the evidence," as David Hume put it. It is, of course, possible that pollsters are all collaborating in an effort to re-elect Barack Obama, just as it is possible that pollsters are all members of the Order of the Knights Templar who are conspiring to effect the return of the Emperor Barbarossa and reinstate the Holy Roman Empire. These, like 50 quadrillion other possibilities, cannot be entirely ruled out; but my lifetime contains only about 2.5 billion seconds, which means that out of all possible cases of the universe, I have already spent too much time thinking about these ones.

Point three: Getting back to point one, that point about the two-candidate race is important. If there are more than two candidates in the race, then signals between voters about which candidates they intend to vote for become an absolutely crucial element of group decision-making. If distrust of polls among Republicans were to become so pervasive that, for example, right-wing libertarians came to believe that Gary Johnson was a serious candidate in the presidential race, it might lead to large-scale defections to Mr Johnson that handed the election to Mr Obama, which would be a sub-optimal result for those right-wing libertarian voters. For this reason, polls in multi-candidate elections serve as an important channel of information, and voters who mistrust polls may end up doing themselves no favours.