Wired to mobilize: The effect of social networking messages on voter turnout

An earlier version of this paper was presented at the 2012 annual meeting of the Midwest Political Science Association, Chicago, IL. Our thanks to Peter Aronow for his statistical advice and to Nicole Ellison, James Fowler, Donald P. Green, Shanto Iyengar and the anonymous reviewers of Social Science Journal for their feedback on earlier drafts. All errors, of course, remain our own.

Abstract

Recent scholarship has documented the effect of online social networking on political participation, a relationship hypothesized to be due to the generation of social capital. This paper tests the hypothesis that impersonal get-out-the-vote messages delivered via an online social network can increase voter turnout. Specifically, this study uses a field experiment of randomly assigned students from a large southern public university to test the effect of exposure to political messages via Facebook on the likelihood of them voting in the November 2010 election. The results indicate that encouragements to vote delivered through a social networking site can have substantively large effects on political behavior.

Keywords:

• Civic participation
• Social media
• Social networks
• Facebook
• Experiment

Notes

An earlier version of this paper was presented at the 2012 annual meeting of the Midwest Political Science Association, Chicago, IL. Our thanks to Peter Aronow for his statistical advice and to Nicole Ellison, James Fowler, Donald P. Green, Shanto Iyengar and the anonymous reviewers of Social Science Journal for their feedback on earlier drafts. All errors, of course, remain our own.

1 Some readers may have concerns about the external validity of the findings. As Druckman and Kam note: (2011: 70): “student subjects do not intrinsically pose a problem for a study's external validity.”

2 Participants who may have been registered in another state are coded as not registered due to resource constraints. When matching participants to the voter file, the birth date field for people born after 1989 had not yet been updated. Instead of listing the voter's full birth date the voter file showed 01/01/1990, 01/01/1991, or 01/01/1992. Therefore, for anyone born in 1989 or earlier, a match was considered positive if a voter file search resulted in a single record in which the name available on Facebook (or a name for which the name on Facebook is an obvious derivative) and the entire birth date were identical. For anyone born in 1990 or later, a match was considered positive if a voter file search resulted in a single record in which the name available on Facebook (or a name for which the name on Facebook is an obvious derivative) and the birth year were identical. Regardless of birth year, if multiple records were returned, data available on the participant's Facebook profile was used to match them to the proper record if it existed. This included any email addresses they listed, their current city, the address of people they identified as family members, or the town where their high school is located. See Appendix B for examples of matched and unmatched records.

3 Four individuals had more than 1675 friends (they had 1917, 1982, 2064 and 2305 friends); six individuals had fewer than 50 friends (1, 16, 27, 29, 34 and 39).