Seamus Heaney on terrorism | Main | moral standing

July 15, 2005

experiments with get-out-the-vote

In 2004, my organization, CIRCLE, sponsored randomized field studies to test the effectiveness of various approaches to mobilizing voters. In these studies, some people are randomly selected to receive a "treatment" such as a phone call, a visit, or a mailing. Other people are deliberately left off the list. After the election, researchers consult voter rolls to see who actually voted. The difference in turnout rates between the treatment group and the control is the effect of the GOTV.

Professors Don Green and Alan Gerber (Yale) have revolutionized the study of voting by promoting the use of such experiments. Until the first Green/Gerber experiments of the 1990s, almost all research had counted the number of people reached by a mobilization effort or had polled citizens to ask them whether they had been contacted. These methods can be very misleading.

As an illustration, consider a study that CIRCLE funded in 2004, in which college students were sent emails, approved by their universities, that encouraged them to register and then gave them an opportunity to do so by clicking a link in the email. Tens of thousands of students clicked the link. By traditional standards of measurement, the project was a success. However, since this was a true experiment, students were randomly selected either to receive the email or not. Those who did not receive the email were just as likely to register as those who did. Thus the emails had no effect whatsoever and were not cost-effective, even at their low price. Every student who used the email link to register would have registered by another means without it. We believe that a considerable amount of offline voter mobilization has been similarly ineffective. Large numbers of people are contacted and then turn out, but they would have voted anyway.

On the other hand, the Green-Gerber studies proved, contrary to conventional wisdom, that young people can be mobilized cost-effectively with face-to-face visits and phone calls. Both major parties and many independent political groups have changed their campaign tactics as a result. The competition between the Democrats and Republicans to mobilize young voters in 2004 may partially explain why we saw the highest youth turnout in any election since at least 1992.

The dozen studies we sponsored in 2004 were selected to fill gaps in our accumulated knowledge. Some of the preliminary results involve messages. Before 2004, most studies had found no difference in the effectiveness of GOTV efforts depending on the message used. The medium, not the message, seemed to determine the results. This was perplexing, since we would assume that it would matter what a canvasser said to a potential voter. Presumably, telling people that Elmo loves them would not encourage them to vote, but all actual scripts and messages seemed to produce similar results.

In 2004, researchers pressed this question by deliberately comparing negative and positive messages, partisan and nonpartisan messages, information-rich messages versus simple exhortations to vote, and messages about potential barriers to voting, aimed at African Americans and Whites.

In general, it still appears that medium matters more than message. Results are available from an experiment that directly compared negative and positive messages; there was no difference in turnout. Various messages and combinations of messages did have different impacts in certain communities. For example, a self-interest appeal was more effective among Chinese Texans than an appeal to ethnic solidarity. In general, however, changing the message did not change the results dramatically. I continue to find this pattern puzzling.

Posted by peterlevine at July 15, 2005 07:47 AM


Hi peter,

Interesting post. I do however take issue with your conclusion. So far, you haven't taken into account the source of the message vs. the medium. For example, perhaps an email campaign from family, or friends or a trusted organization, would work but unsollicited email does not. Then again, I do occasionally ignore "helpful" emails from mom too...

- Mike

Posted by: Michael Weiksner at July 18, 2005 07:41 AM

Post a comment

This blog is under attack from comment spammers, who are causing a problem for the server. I believe I can block them by upgrading to a recent version of MoveableType. However, I do not have time to do that until late December. Therefore, I have temporarily disabled comments. Please feel free to email me feedback at

Site Meter