The Oprah Effect

By Shankar Vedantam
Monday, September 1, 2008

Political conventions, like all forms of public relations, carry the risk of whiplash. Last week in Denver, for example, Bill Clinton said Barack Obama "has a remarkable ability to inspire people, to raise our hopes and rally us to high purpose" -- all of which sounded a little different from the former president's earlier take on Obama.

"This whole thing is the biggest fairy tale I have ever seen," Clinton said of Obama's stance on the Iraq war, at a Hillary Rodham Clinton campaign event during the Democratic primary. Referring to Obama's supposedly inspirational campaign, Clinton added: "The idea that one of these campaigns is positive and the other is negative when I know the reverse is true, and I have seen it and I have been blistered by it for months, is a little hard to take."

John McCain's campaign manager, Rick Davis, publicly accused George W. Bush's operatives of spreading rumors in 2000 that McCain had fathered an illegitimate black child -- a smear campaign that many McCain loyalists think cost him the Republican nomination that year. But at the Republican convention this week, expect Bush to warmly endorse McCain, and for the Republican nominee to delightedly accept the endorsement.

Given the choice between believing Clinton when he slammed Obama and when he praised Obama, most people are likely to believe that Clinton was being sincere when he was critical. A variety of experiments have shown that people pay much closer attention to criticism than to praise, when it comes both to themselves and to others. When we hear equally passionate positive and negative evaluations of a political candidate, the negative evaluation carries greater weight.

The research suggests that our bias toward negative information comes about for two reasons. Negative information has a greater propensity to harm than positive information has to help -- the boss sending you an unflattering note indicates you might be on your way to getting fired, but a flattering note hardly suggests a raise is imminent. Also, we have learned that people are more likely to be speaking their minds when they are critical. In terms of using information in the most efficient manner, it makes sense to discount praise and inflate criticism.

Given their naked insincerity, do political endorsements carry any weight? There are lots of hints they don't -- after Sens. Edward M. Kennedy and John F. Kerry and Gov. Deval Patrick all endorsed Obama in the Massachusetts Democratic primary, for example, Hillary Clinton whipped Obama in the state by 15 percentage points.

A new study, however, suggests that -- at least in some cases -- political endorsements can have resoundingly powerful effects. Employing an unusual methodology, Craig Garthwaite and Tim Moore at the University of Maryland have shown that Oprah Winfrey's endorsement of Obama not only netted him upward of a million votes in the Democratic primary season but also probably played a central role in his winning the nomination.

"It does appear to have been a decisive, if not a deciding, factor," said Garthwaite, who has not yet published his analysis.

Garthwaite and Moore, who are PhD students in economics, matched sales of Winfrey's hugely popular "O" magazine and the spike in sales that usually accompanies a book getting nominated to Oprah's Book Club to Obama's performance in the Democratic primary.

After controlling for a large range of confounding factors -- including the obvious fact that Obama and Winfrey draw substantial support from African Americans -- Garthwaite and Moore showed that votes for Obama spiked in precisely the same geographic areas where Winfrey is the most popular. The researchers applied the same methodology to Obama's 2004 Senate race -- when he did not have Winfrey's endorsement -- to show there was no relationship between Obama votes and her popularity in Illinois, which suggests the two do not merely have crossover appeal.

In "The Role of Celebrity Endorsements in Politics: Oprah, Obama, and the 2008 Democratic Primary," Garthwaite and Moore also showed that the connection is not because people who read women's magazines tended to like Obama: The politician got less support where women's magazines such as Self and People are popular. After controlling for racial factors, the researchers also found no relationship between the popularity of Ebony magazine, whose readership is strongly African American, and support for Obama. Winfrey's support, the researchers also found, boosted campaign contributions to him in those counties where she is most popular.

Garthwaite and Moore argue that the endorsement may have been effective because it was unique: For one thing, Winfrey may be the biggest celebrity in the country. For another, she endorsed Obama early, well before he assumed front-runner status, which suggests her decision was not part of the usual bandwagon effect. And, finally, this is the first time Winfrey has thrown her weight behind a candidate in a political race.

© 2008 The Washington Post Company