Much has been made of the use of social media this campaign cycle, from the Democratic National Committee's Web site mocking Mitt Romney's tax plan to President Obama's citation of "Mean Girls" on his own site's Tumblr page to the Romney campaign's tweets showing the nominee's family playing Jenga before the first presidential debate. But does this really have an impact on potential voters?
A new study suggests not. Columbia's Donald Green, a leading scholar on campaign effects, and Berkeley's David Broockman (last seen proving that state legislators show racial bias in handling constituent requests) paired up with a state legislature candidate and bought Facebook ads targeting a random sample of his potential voters.
They bid an extremely high amount for the ads, so that "a voter in the treatment group was exposed to an ad every single time they loaded any Facebook page during the entire week." They tested three types of ads: one aimed at boosting name recognition, one that touted the candidate's experience and one that focused on his policies.
So did the ads work? Nope. Broockman and Green concluded that "the ads had no politically consequential effect on knowledge of the candidate, his favorability or support for his election among voters." So, there's no evidence that the ads even worked to get the candidate's name out there, much less increase support for his candidacy.
The experiment lasted only a week, and the effects might have been greater for a congressional or gubernatorial race. But Broockman and Green really carpet-bombed those voters, and even those who said they used Facebook during the ad week weren't any more likely to have heard of the candidate than those who did not use the social network. Given that Green and his co-authors have in the past found positive (though short-lived) effects on potential voters from TV campaign ads, this study suggests that campaigns may want to keep to tradition when it comes to ad strategy.