It is not an uncommon sight: a political candidate, standing on a stage with some rock star. They smile, wave, and clap each other on the back. Many people running for office go out of their way to secure these celebrity endorsements. In the 2012 Republican primary race, for example, Mitt Romney has racked up the support of Donald Trump, Kid Rock, and Cindy Crawford. But whether or not these endorsements have an effect has always been an open question. So Craig Garthwaite, an assistant professor of management and strategy at the Kellogg School of Management, and Timothy Moore, a doctoral student at the University of Maryland, decided to take a look.
The effect of celebrity endorsements has in the past been a controversial issue. On one side are pundits and academics who say that celebrity endorsements have no effect whatsoever. On the other side are people who run campaigns. For these people, time is a very valuable resource and they often expend a lot of effort to line up endorsements with celebrities.
“We have this sort of contrasting set of opinions that we haven’t been able to reconcile,” Garthwaite says. But, “it seems odd that these campaigns could be somehow systemically irrational and attempt to get support from people that have absolutely no effect.”
Someone obviously had to be wrong here, Garthwaite thought. And the researchers’ intuition pointed to the academics, that celebrity endorsements do have an effect on voters. “If endorsements affect all kinds of behavior, why would we think that they wouldn’t affect voting behavior?” asks Garthwaite. However, he explains, it is generally difficult to measure the impact of an endorsement.
“Elections are a one-shot game,” he says. The situation is usually like this: there are a few candidates, and one or more may have a celebrity endorsement before the election. Then the election happens and somebody wins. But less certain is whether the same person would have won without the endorsement and how many votes the endorsement was worth.
Garthwaite and Moore found a unique situation in the 2008 Democratic primary, however. In this election, TV host Oprah Winfrey endorsed then-Senator Obama as a presidential candidate. The mega-celebrity had never publicly backed a candidate before. She is also extremely popular, with a high amount of influence.
To gauge Oprah’s popularity and how it varied geographically, Garthwaite and Moore started by finding the number of subscribers to O, the Oprah Magazine who lived in each county in the United States. They then compared the number of subscribers in a given area to Obama’s electoral success in that same region. After controlling for a wide range of socioeconomic and demographic factors—sex, education, race, and income, to name a few—the authors found that Obama did better in areas with more O subscribers.
In the study, the researchers wanted to make sure that they were picking up an effect of the endorsement, and not just an underlying preference for Obama that was correlated with preferences for Oprah. To check they were only detecting Oprah’s influence on voters, Garthwaite and Moore looked for geographic variation in O subscribers and in President Obama’s past electoral experiences. In particular, they studied his 2004 Democratic primary race for the U.S. Senate, which is the only multicounty race he was involved in prior to running for president.
“We find no relation between the popularity of Oprah and his county-level outcomes in that election. That suggests that there’s really not some underlying preference that’s driving this, but it’s the endorsement’s effect,” Garthwaite says.
“We find that if anything, the types of people that read these magazines would be expected in the absence of the endorsement to support then-Senator Clinton in this election, and not then-Senator Obama. So we don’t think that there’s some sort of inherent preference for then-Senator Obama among the people that read these types of magazines.”
Overall, they calculated that Oprah’s endorsement was worth about 1 million votes for Obama, which is a significant number, Garthwaite says. “In politics, if you could guarantee that you could turn a million more people towards your candidate you’d be happy,” he adds. “The other thing you have to remember is that this was a really close election. So in that sense I think that a million votes is a pretty important number.”
Obama beat out Hillary Clinton by only about 278,966 votes in the states they used in their sample, Garthwaite points out in the paper. Oprah’s endorsement had an effect on the 2008 presidential primary, but whether other celebrities would also have had an effect is less clear. “We can’t obviously test that in the paper,” Garthwaite says. “The question we really have to ask is whether the fact that Oprah is such a powerful celebrity means that the magnitude of her endorsement is greater than others’, or if it means that only her endorsement, and no one else’s, would have an effect.”
Gauging the Celebrity Effect
If this were simply a question of magnitude, that would mean Oprah’s endorsement would be worth a million votes and someone else’s, depending on their popularity, would be worth maybe 100,000, for example. “I think that the effect is increasing in the popularity of the celebrity, which means lesser celebrities would still have an effect. On the other hand, there could be a nonlinear relationship, but that would be a more difficult story to tell,” Garthwaite says.
But their finding, that Oprah did really have an effect on how many votes Obama received in the 2008 Democratic primary, may or may not convince everyone about the value of celebrity endorsements.
“I think it’s changed some people’s minds in that we definitely have an example here where there was a strong endorsement,” Garthwaite says. But whether or not this will convince pundits is anyone’s guess, he adds. “I think one unique feature of the punditry in the United States that we’ve seen over the past three years is that facts don’t necessarily seem to change their minds,” he adds.
Related reading on Kellogg Insight
The Trouble with VPIN
New economic indicator fails to live up to its promises