Who Will Win An Election? Snap Judgments Of Face To Gauge Competence Usually Enough

Princeton psychologist Alexander Todorov has demonstrated that quick facial judgments can accurately predict real-world election returns. Todorov has taken some of his previous research that showed that people unconsciously judge the competence of an unfamiliar face within a tenth of a second, and he has moved it to the political arena.

His lab tests show that a rapid appraisal of the relative competence of two candidates' faces was sufficient to predict the winner in about 70 percent of the races for U.S. senator and state governor in the 2006 elections.

"We never told our test subjects they were looking at candidates for political office -- we only asked them to make a gut reaction response as to which unfamiliar face appeared more competent," said Todorov, an assistant professor of psychology and public affairs. "The findings suggest that fast, unreflective judgments based on a candidate's face can affect voting decisions."

Todorov and Charles Ballew, an undergraduate psychology major who graduated from Princeton in 2006, conducted three experiments in which several dozen participants had to make snap judgments about faces. Participants were shown a series of photos, each containing a pair of faces, and asked to choose, based purely on gut feeling, which face they felt displayed more competence. The differences among the experiments largely concerned the amounts of time an observer was allowed to view the faces -- as brief as a tenth of a second or longer -- and to pass judgment afterward.

What was unknown to the participants in the third experiment was that the image pairs were actually the photographs of the two frontrunner candidates for a major election being held somewhere in the United States during the time of the experiment in late 2006. The races were either for state governor or for a seat in the U.S. Senate. In cases where an observer recognized either of the two faces, the researchers removed the selection from the data.

Two weeks later elections were held, and the researchers compared the competency judgments with the election results. They found that the judgments predicted the winners in 72.4 percent of the senatorial races and 68.6 percent of the gubernatorial races.

"This means that with a quick look at two photos, you have a great chance of predicting who will win," Todorov said. "Voters are not that rational, after all. So maybe we have to consider that when we elect our politicians."

Todorov's paper on the findings, written with Ballew, appears in the Oct. 22 issue of the journal Proceedings of the National Academy of Sciences. The paper has inspired researchers elsewhere to re-examine their assumptions about visual images and their effect on decision-making among the public.

"Political scientists have spent 50 years documenting only modest effects of the media on voting behavior, but Todorov's research suggests we may have been looking in the wrong place," said Chappell Lawson, an associate professor of political science at the Massachusetts Institute of Technology. "Most of these previous studies have relied on transcripts or printed records of what the media say, with much less attention to visual images."

Lawson, who called Todorov's work "pioneering," added that some of his own work corroborates the new findings, indicating that competence appears to be a universal quality, recognizable across cultures. His research shows that American observers could predict the outcome of elections in Mexico based on the same gut reactions.

"Both of these papers speak to the seminal quality of appearance in candidate success," Lawson said. "Our findings surprised us, because Mexican politicians often emphasize very different aspects of their appearance, such as facial hair, which American political figures avoid. But Americans could still pick out the Mexican winners. Our data show effects at least as strong as those Todorov found."

Political scientists, Todorov said, are likely to be most interested in his findings, primarily because they will want to identify which voters are most strongly influenced.

"It's still unclear how these effects operate in the real world," he said. "Not every voter is going to be affected. Obviously, some people vote according to their values, but many others are uninformed about candidates' policy decisions. So we need to do the hard work to find out."