Obama, flanked by Suzanne Stern, left, and Alexa Kissinger, visits the Organizing for America field office in Williamsburg, Va., where he made phone calls to thank volunteers. (Nikki Kahn / The Washington Post)
Obama, flanked by Suzanne Stern, left, and Alexa Kissinger, visits the Organizing for America field office in Williamsburg, Va., where he made phone calls to thank volunteers. (Nikki Kahn / The Washington Post)

Perhaps the most extraordinary feature of the now-retracted LaCour and Green study about persuasion was not that the supposed effects were large or that they were long-lasting. Rather, it was the substantive finding that gay canvassers convinced gay-marriage opponents to change their minds.

This was surprising because a large body of social science research suggests the opposite is more likely true: If you want to convince a white, suburban, straight, gay-marriage opponent to change their mind, recruit one of their white, suburban, straight neighbors to engage with them. Young, out-of-town, gay activists should actually be less effective.

In a recent article, Ryan Enos and I explore what we call the principal-agent problem of campaign mobilization. We argue that the kinds of people who typically volunteer for political campaigns are not well-suited to bringing swing voters to their side.

For Democrats, think of the liberal, racially diverse college students who came out in droves to volunteer for Barack Obama. Even if these activists are trained how to speak to swing voters, their demographics and behavior send subtle cues that could easily to rub many swing voters the wrong way.

Or consider Republican activists like those who canvassed for Ron Paul’s presidential bids. According to the New York Times, Paul’s volunteers needed to be told “to look, dress, shave, sound, and behave in a way that will not jeopardize Mr. Paul’s chances.” As the Paul campaign knew, in order to connect with voters, it’s better to accentuate your commonalities with them rather than your differences.

The literature on this is pretty clear. Want to mobilize blacks and Latinos? Send black and Latino canvassers to talk to them (see here, here, here, here, and here). Want to mobilize voters in a particular area? Recruit local canvassers, who are much more effective than out-of-area activists (here, here, here, and here). In other words, a strategic campaign ought to employ canvassers similar to the target population. For more, see research by Samuel Popkin and Kevin Arceneaux and Robin Kolodny.

But the LaCour and Green study reached a different conclusion, which is why it was a big deal. As Ira Glass concluded in his “This American Life” segment about the study, “When people most affected by an issue show up at your door and talk to you, that’s the thing that could change your mind.”

We now know that this conclusion has not been supported. The best evidence suggests that campaigns should really do the opposite. Do not send people most affected by an issue; send people who most resemble the people you are trying to convince. That’s the advice that political science literature can offer at this point.

Eitan Hersh is Assistant Professor of Political Science at Yale University.