Pundits trying to predict election outcomes usually rely on polls, but a new study suggests a different method could prove useful: tallying the cheers and laughs during candidates’ speeches. In a new paper in the Journal of Language and Social Psychology, Peter Bull and Karolis Miskinis, psychologists at the University of York, analyzed the content of eleven speeches delivered by Obama and Romney on the campaign trail in 2012—and discovered a strong relationship between electoral outcome and the frequency of audience responses like laughter, cheering, or applause.

Bull and Misnikis selected speeches delivered to bipartisan audiences at informal public gatherings in five swing states: Wisconsin, Florida, North Carolina, Ohio, and Iowa. They watched videos of the speeches alongside transcripts, marking each instance of audience response and sorting them into categories like: applause, laughter, cheering, collective chanting, and booing. In all five swing states, the candidate who elicited a higher number of positive responses per minute was the one who carried the state. And overall, Obama drew an average of 2.57 positive responses per minute, compared to Romney’s 2.23.

Journal of Language and Social Psychology

That response rate also highlights just how interactive political speeches can be. In the 1980s, political speechwriter Max Atkinson argued that the use of a rhetorical device—like naming or thanking another politician—serves as a signal to the audience that they can react. According to Bull and Miskinis, Atkinson’s “key insight was to compare political speech-making with the way in which people take turns in conversation. Thus, just as conversationalists take it in turn to speak, so speaker and audience may also take turns, although audience ‘turns’ are essentially limited to gross displays of approval or disapproval (such as applause, cheering, or heckling).”

Bull and Miskinis also incorporated cross-cultural data into their study, comparing what they saw in American audiences with what other researchers had observed in Japanese and British crowds. As they expected, in the more individualistic U.S. society, Americans sometimes piped up on their own—providing “a continuous flurry of asynchronous and uninvited individual remarks.” In a sample of 74 speeches delivered during the 2005 and 2009 elections in Japan, on the other hand, every audience response was in unison.

Another interesting cross-cultural difference was in the proportion of laughter to cheering: Japanese audiences appear to be more easily amused by their politicians, with laughter accounting for 25 percent of the positive audience reactions in the 36 speeches of the 2005 elections, compared to just 9 percent in this U.S. sample. Cheering was by far the most common American positive response; in Japan, it was applause.

Booing also appears to be a predominantly American behavior. Most often, though—in 45 of the 48 instances of booing in the U.S. sample—it was actually an expression of support for the speaker, directed against a rival who’d been mentioned (a rhetorical device the authors call “negative naming”). When British politicians employed the same trick—bringing up a political opponent for the purpose of derision—audiences would applaud rather than boo. And in the Japanese sample of 74 speeches, not a single “boo” was heard.