Clinton could use a 'Colbert bump'

Hillary Clinton is out of money. She recently loaned $5 million to her own campaign just as Barack Obama was setting records by raising $32 million in January. As the two of them race ahead, winning the money game now is more important than ever. But where can she turn? Two words: Stephen Colbert.

In character, Colbert claims that anyone who comes on his Comedy Central show, "The Colbert Report," receives a boost in popularity that immediately vaults the guest to stardom, fame and fortune.

Fans have pointed to several pieces of evidence that the "Colbert bump" is real. Colbert helped former Orleans singer John Hall win his seat in Congress. Ned Lamont came on the show and then promptly beat Sen. Joe Lieberman in the Connecticut primary. And in this year's presidential race, popular support for Ron Paul doubled and for Mike Huckabee tripled immediately after they appeared on the show (never mind that they both started at a paltry 1 percent). At that rate of increase, Colbert quipped, Huckabee could be the first candidate ever to win 88,128,000 percent of the vote.

But these efforts to measure Colbert's influence don't even qualify as "truthy." For example, one method involved averaging the percentage of the vote earned by each candidate who had appeared on the show. Because 89 percent of them were incumbents (several running unopposed), it is hardly surprising that they beat their opponents by 34 percentage points on average.

Such amateur statistics fall prey to the problem of self-selection — candidates with secure seats may be much more likely than others to risk televised humiliation. Just ask Georgia Republican Rep. Lynn A. Westmoreland, who advocates putting the Ten Commandments on the walls of Congress — he could only name three of them on the show — or Florida Democrat Rep. Robert Wexler, whom Colbert goaded into saying he liked cocaine and prostitutes "because it's a fun thing to do." Both were incumbents facing no serious competition.

But it is possible to control for the self-selection problem by also examining similar candidates to those who went on "The Colbert Report." So that's exactly what I did in my study. I matched Republicans to other Republicans, incumbents to other incumbents, and so on. (Colbert might call his guests the "Hall of Heroes" and the similar-looking "control" group that didn't go on the show the "Hall of Cowards.") I then used data from the Federal Election Commission to see whether being on the show had an effect on campaign contributions.

So what did I find?