Yes, You Can Be A Political Pundit

Here's a tip from the ivory tower: If you are looking to pick the winner in a political contest, go with your instincts.

As it turns out, just a glance at pictures of two rival candidates is enough for the brain to make a pretty reliable prediction of which one will win. On average, a snapshot judgment is right about 70 percent of the time, clearly better than a coin toss. Maybe not as good as the professional political pundits, but not too shabby.

Backing this up is an experiment organized by Princeton social psychologist Alex Todorov and recent graduate Charles Ballew, published in the October 22 issue of the Proceedings of the National Academy of Science.

It went like this: Each test subject was presented with a series of paired black-and-white mug shots, labeled A and B. First, the individual was asked to rate which face looked more "competent" based on gut reaction, and then he or she was told to rate how much more competent that face appeared on a scale of 1 to 9. This went on for a series of 64 pairs of faces presented in random order.

What participants were not told prior to the experiment was that the photos were from real matchups in political races for Senate or governor that had either occurred in the past decade or were ongoing at the time. The competency judgments predicted the outcome of about two thirds of the races correctly, even though participants did not know they were looking at politicians. (Students were dropped from the study if they recognized any of the faces, which excluded highly visible candidates like Hillary Clinton or Arnold Schwarzenegger.)

Experiments like Todorov's--conducted in October 2006-- are part of a growing body of research into the social science of political decision making, in which behavioral data supplement--and sometimes replace--more traditional means of polling people on their priorities.

Other researchers, for example, have shown short clips from gubernatorial debates with the sound off and asked participants to rate the stronger candidate, with similar correlation. And other permutations in Todorov's experiment found statistically significant correlations using photos of candidates in past races dating back to 1995, even when participants were exposed to the images for only a quarter of a second.

All of this may sound ugly for the future of political discourse, if two thirds of officials are getting elected based on split-second superficialities. But the correlation in this study raises more questions than it answers about the baffling process of candidate selection. As any self-respecting psychology major can testify, the correlation between two sets of results doesn't mean one is directly responsible for the other.

Still, Todorov says the results show that the process of selecting a candidate is less rational than one might imagine. "Most of the social psychology and decision making research in the last 20 or 30 years suggests that people don't have these stable, coherent preferences that we usually think about," he says. "In general, we think of people as rational agents. A lot of the work in decision making in psychology has been to show that these assumptions are really inconsistent."

A 70 percent prediction rate is well below the batting average for the best political pundits, some of whom predicted nearly all of the same 64 races correctly last year. But given what the participants had to work with, it amounts to a strong correlation between these kinds of snap judgments and the actual outcomes, particularly when one weighs these knee-jerk reactions against the many powerful forces that traditionally govern elections, like party identity and incumbency.

"This is pretty definitive evidence that people are making very rapid judgments based on even brief glimpses, from the way candidates hold their faces, as in this study, to the ways they present their emotions and other aspects of who they ae nonverbally," says Drew Westen, a psychology professor at Emory University and author of The Political Brain: The Role of Emotion in Deciding the Fate of the Nation.

Jon Krosnick, a political scientist at Stanford University who studies political behavior, offers one explanation. Like well-documented studies that show that attractive people are treated better, thus affecting the personality in a positive way, Krosnick suggests that a natural aura of competence can be a "self-fulfilling prophecy" for a future politician. Under this scenario, winning candidates would appear more competent even if this was not a basis for short-term political selections.

"The competent-looking candidates actually turn out more competent over the years, because people make them so," says Krosnick, who was not involved in Todorov's study. "But the reason that they win elections is because they are competent, not because they look competent." (Todorov responds that evidence also exists to the contrary: that people who look young and inexperienced can work harder to overcome this stereotype, eventually outpacing their peers.)

A list of which 2006 gubernatorial races were correctly predicted in the experiment, which Todorov provided to U.S. News & World Report, does not show any clear correlation between traits like gender or race of the candidates. In another of the three experiments discussed in the paper, the pairs of photos were limited to just cases where opponents were of the same gender and ethnicity, and the researchers found that the predictive capabilities actually improved.

By Chris Wilson