Wednesday, July 7, 2010
I was skimming the abstracts of this week's PNAS early edition, when I came across this fun paper: Irrelevant Events Affect Voter's Evaluations of Government Performance.
I imagine most of my readers are sophisticated enough to know that people are not very rational, but still the outcome of this paper may be a little bit of a surprise. The study had two parts. In the first part, they did a regression data analysis on county level election results, 1964-2006, looking at the relationship of political elections and whether the local college football team had won or lost in the two weekends prior to the election. It turns out that if the local team wins, it boosts the incumbent party's share in the election, by, on average, 1.1%. That may not seem like that much, but obviously it could change the outcome of close elections (and the effects are statistically significant, though not by a mile).
Also, the effects seem larger when taking account of the fact that only about 10%-15% of voters make up their minds in the last two weeks. So amongst those voters, college football is apparently causing 7-10% swings in incumbent party vote share.
The general idea is that people are basically emotional association machines. Prior research has shown that when people are in an upbeat mood, they can generally more easily access positive memories of other things, and when they are in a bad mood, they can more easily access negative memories. So presumably, the excited supporters of the winning team, in a generally good mood, feel suffused with good-will which spills over into their feelings about their incumbent representative and makes them more likely to vote for him or her.
Given that college football is by no means the only thing that affects the mood of voters, but is utterly unrelated to the actual performance of politicians, these late deciding voters may not be a whole lot better than noise in terms of actually providing a useful informational feedback within the democratic system.
The second study was similar in theme but used survey data of approval of President Obama during the 2009 "March Madness" college basketball tournament. Again, the better your team was doing, the more likely you were to approve of President Obama. Also, intense sport fans showed a much stronger effect than fans who didn't have a strong emotional investment in their team.
The most interesting thing here was an experiment in which they explicitly told survey participants the score of the most recent game right before asking about presidential approval. When they did that, the effect of game outcome on presidential approval disappeared. The idea is that when an irrelevant fact is explicit in consciousness, people are able to correctly assess that it is irrelevant and ignore it in making other judgements. However, when it's just subliminally present in the background, it has an impact. This is consistent with a long history of research into subliminal effects of advertising.
Anyway, the political implications for November will be no surprise - voters are going to be pissed about the economy, and incumbents should look out.
I can't wait for the computers to take over, so we won't have to deal with this kind of thing...