Reproducibility of findings has been a hot-button issue in social science over the last year, and as the election approaches, the reproducibility of findings related to voting is especially relevant. A recent study published in PNAS reassesses previously published data about ways to increase voter turnout and finds that language cues actually don’t have a significant effect on voter behavior.

The study revisits some field experiments conducted by Bryan et. al., which were also published in PNAS. These earlier experiments looked at voter behavior during the 2008 presidential election in California and the 2009 gubernatorial election in New Jersey. Using online surveys to prime voters, the previous study had shown that use of nouns or verbs affected whether voters went to the polls—verbs did better at getting out the vote.

This new study follows up on that work, looking at how subtle language manipulations can prime the behavior of voters.