Look and Act Like a Winner, and You Just Might Be One

By Shankar Vedantam
Monday, November 6, 2006

Quick, what three issues will decide your vote in the midterm elections?

Did you mention a sleek physique, a strong jaw and a penetrating gaze? Didn't think so.

Elections are supposed to be about issues. In tomorrow's election, those issues include the war in Iraq, terrorism and the spiraling cost of health care.

Hardly anyone mentions good looks and charisma, but a number of ingenious experiments show that how a politician looks and comes across to voters can make a huge difference in the outcome of an election.

In research that will be made public today, a pair of economists asked a large group to watch 10-second video clips of the Republican and Democratic gubernatorial candidates in 58 races from 1988 to 2002. Based on this limited information -- a quick, gut sense of charisma -- the volunteers were asked to guess which candidate won.

If people recognized either politician, their answer was dropped from the analysis, meaning that any knowledge about the outcome or expertise in politics was removed from the study. All the volunteers, in other words, were simply guessing and ought to have been no better at predicting the winner than someone tossing a coin.

Remarkably, the study found, the volunteers were quite good at guessing the winner.

If your jaw hasn't dropped already, this ought to do it: The 10-second video clips that the volunteers watched had the sound off, meaning that viewers saw only two white guys (all the elections featured two white guys) sawing the air and opening and closing their mouths. What the economists wanted to measure were those intangibles we pick up the moment we see someone, before we actually hear what they have to say.

The research did not show that any individual volunteers were exceptionally good at making predictions -- individuals regularly made predictions that were right and wrong. But when the answers were averaged over the whole group, the volunteers were able to spot winners more often than mere chance would dictate.

Curiously, when the sound was on and the volunteers could hear what each candidate said for 10 seconds, the viewers became much more confident in their guesses about who won, but their predictions became worse -- no better than chance.

The idea that better knowledge does not always improve judgment is sobering but aligns with previous research. In a book he wrote about expert political opinion, Philip Tetlock at the University of California at Berkeley found highly trained experts often do no better at predicting political events than nonexperts or even someone tossing a coin. Our faith in political pundits -- and the confidence these experts have in their own intuitions -- is often misplaced.

The new study is being released online today as a working paper from the National Bureau of Economic Research, a forum where economists share ideas. The study was conducted by Daniel Benjamin at the University of Michigan's Institute for Social Research and Jesse Shapiro at the University of Chicago.

CONTINUED     1        >

© 2006 The Washington Post Company