FACE IT: APPEARANCES SWAY VOTERS
By Gareth Cook, Globe Staff
The Boston Globe
June 14, 2005
SECTION: HEALTH SCIENCE; Pg. E1
In a vivid demonstration of the superficiality of politics, scientists have announced that they can predict the winners of American congressional elections using nothing more than black and white photographs of the candidates.
The scientists showed images of US Senate candidates to volunteers for one second and asked the college-age volunteers to pick the candidate who looked more "competent." On average, the one who appeared more competent won the election 68 percent of the time.
Political operatives know that unfolding campaigns are moved by many factors that cannot be predicted much less described in the precise language of science yet the results announced last week suggest voters are indeed judging their candidates by their covers.
"One can imagine cases where this is probably not the best way to select politicians," said Alexander Todorov, a Princeton University scientist who led the experiments.
These volunteers chose from candidates they didn't recognize, without knowing anything about the politician's party, character, or stands on the issues the supposed stuff of deliberative democracy. Their answers to questions like who looked more "intelligent" or who was more "attractive" didn't predict the election's outcome.
The new work, published in the June 10 issue of the journal Science, joins a large and growing body of research that shows people do not base their decisions on the careful deliberation that might be hoped for.
For example, an entire field called "behavioral economics" has shown that people often do not act in their objective best interests, undermining one of the bedrock assumptions of traditional economics.
And other research, popularized in the best-selling book, "Blink," has focused on the staying power of snap judgments, even when the judgments are incorrect. Author Malcolm Gladwell argues that Warren G. Harding, whom historians widely consider one of America's worst presidents, was elected in 1920 largely because he looked "presidential."
The findings are also likely related to a set of discoveries about how people respond to "baby-faced" features, according a companion paper in Science.
Conversely, people assume that people with more mature faces are more shrewd.
This attitude is probably part of our evolutionary heritage, encouraging parents to care for their young. "If people didn't see their babies this way, they probably wouldn't pass their genes along very well," Zebrowitz said.
This deeply-ingrained attitude associating babyish physical features with psychological features like helplessness shapes the quick judgments we make about all faces. On average, Zebrowitz said, women have more baby-faced features, and it is possible that this is a liability in the race for high office. The Todorov study didn't show that women are at a disadvantage.
The predictions, using only the photos, worked well wherever they tested them. They were correct in about 70 percent of the race outcomes for both the US House of Representatives and the Senate.
Altogether, the team looked at a total of 95 Senate races and 600 House races, from 2000 to 2004, according to the report.
The competence scores, gleaned from the photographs, were also correlated to the margins of victory with candidates who had more competent-seeming faces generally winning by a greater percentage of the vote, according to the paper.
To do the experiment, the volunteers were shown images of the candidates and then asked a variety of questions about the candidates, including whether they recognized the face. If the volunteer recognized one of the faces, their data was not used for that race. Much of the work was done after the elections they tested, but the predictions were just as accurate when the testing was done before the election, Todorov said.
The results are interesting because character has come to play such an important role in modern campaigns, and the research gives a very precise answer about what connects with people, said Chris Lehane, a California-based Democratic political consultant. The experiment, he noted, discredited "likability" and other factors that pollsters have long argued are important.
In the 2000 campaign, when Lehane worked for presidential candidate Al Gore, polls regularly showed that voters considered Gore more "competent" but found Bush more "likeble," and Bush won the election. Lehane added, however, that Gore did win more votes in that election.
"I guess the visage factor does not take into account the limits of the electoral college."
SIDEBAR 1:
TAKE THIS QUIZ
VOTE FOR WHO'S MORE COMPETENT
Look quickly at each pair of faces and decide which person you think is more competent. Then look on Page E4 to see who looked "competent" to study volunteers and how that compared to actual election results for these congressional candidates.
SIDEBAR 2:
Who looks more competent?
Researchers at Princeton University asked volunteers to look at pictures of rival Congressional candidates they did not recognize and say which one in each pair looked more "competent." Their first impressions correctly predicted the winner on average in nearly 70 percent of the races they studied. In this example, 55 percent of volunteers thought Senate incumbent Russ Feingold, who won the election with 56 percent of the vote, looked more competent than Tim Michels, who got 44 percent.
Here are the answers to the quiz on the cover, pairs from left, for the following 2000 Senate races:
In South Dakota, John Thune was seen as more competent by 63 percent of volunteers, and he won the race with 51 percent of the vote compared to Tom Daschle's 49 percent.
In Nevada, Richard Ziser earned 35 percent of the vote, while Harry Reid won the election with 61 percent, and was seen as more competent by 75 percent of the volunteers.
In California, Bill Jones (second row, far left) earned 35 percent of the vote, while Barbara Boxer won 58 percent of the vote, and was seen as more competent by 60 percent of volunteers.
In Illinois, volunteers threw researchers a curveball. Eighty-four percent thought Alan Keyes looked more competent than Barack Obama, but Obama won with 70 percent of the vote to Keyes' 27 percent.
|
