Self-Prophecy Effects and Voter Turnout in the United States

Anton Orlich
Jennifer K. Smith
Fieldwork by:
Northern United States
1,160 registered voters with a major-party affiliation
2000 - 2000
Target group:
  • Voters
Outcome of interest:
  • Electoral participation
  • Voter Behavior
Intervention type:
  • Nudges and reminders

A subset of psychological research suggests that predicting one's own future behavior may change actual future behavior. Voting presents a particularly useful context for examining this phenomenon because the prediction and the action occur at distinct times and the action of voting can be tracked. Researchers contacted registered voters in a New England town by telephone and asked a portion of these voters to predict whether or not they would vote. This allowed researchers to measure the impact of making a prediction about future voting behavior on actual voting behavior. The study found no significant effect of prediction-making on voter turnout.

Policy issue

Insights from psychology suggest that people may be more likely to follow through with a particular action if they are asked to predict their behavior in advance. For example, individuals with gym memberships who were asked to predict whether they would go to the gym were more likely to visit the gym than individuals who were not asked to predict their own behavior. The effects of this type of prediction-making or "self-prophecy" have also been studied in the contexts of volunteering and cheating on examinations. Another area where these insights may be relevant is voter turnout. If predicting future voting behavior increases voter turnout, this could provide an effective means of increasing electoral participation. Does asking voters to predict future voting behavior affect whether they actually vote?

Context of the evaluation

Researchers conducted this study in a New England town of about 50,000 residents. The investigation occurred in the days leading up to the March 7, 2000 presidential primary election, in which only registered Democrats and Republicans were eligible to vote. In addition, researchers limited the study sample to include only households with one or two registered voters.

Details of the intervention

Researchers examined the effect of predicting one' s future voting behavior on actual behavior. Voters were contacted by telephone on the Sunday and Monday prior to the election. Phone calls took place from 4 to 6 p.m. and 7 to 9 p.m. on Sunday and 6 to 9 p.m. on Monday. Interviews were conducted primarily by graduate students. Once an eligible voter was contacted, the interviewer requested that the voter participate in a brief survey and asked if the respondent knew the location and opening hours of the local polling center. Respondents who did not know were provided with the information.

Next, respondents were randomly asked to answer one of four possible questions. Interviewers determined which question to ask by drawing cards from a deck, creating three treatment groups and a control group:

  • Group 1: Voters were asked to predict their future voting behavior.

  • Group 2: Voters were asked to name the most important reason for voting.

  • Group 3: Voters were asked to both predict whether they would vote and name their reason for voting.

  • Group 4 served as the control group and was asked two demographic questions, which were also appended to all treatment interviews.

By randomly assigning questions, researchers were able to distinguish between the effects of self-prophecy and giving a reason for the importance of voting, which previous studies did not do and which may offer insight on tactics for future voter-mobilization efforts. Researchers collected data on the voting behavior of the survey respondents several weeks following the election.

Results and policy lessons

The study found no significant differences in voter turnout among the treatment and control groups. Neither predicting one’s future behavior nor giving an important reason for voting influenced voting behavior. Both together also had no impact on voter turnout. The researchers also examined whether the treatment had varying effects on individuals with different voting histories. Citing prior research in the field, the authors suggest that individuals with a very strong or very weak record of engaging in voting may be less swayed by self-prophecy, whereas individuals with more moderate records may be more susceptible to the behavioral influence of self-prophecy. However, using turnout records from five general elections prior to the presidential primary, voting history did not play a role in susceptibility to self-prophecy. In other words, the treatment did not change voter turnout for individuals with histories of low, moderate, or high voter turnout.

It is important to note that the authors only consider contacted voters in their analysis while voters who could not be contacted for whatever reason do not factor into the study. The researchers suggest their results may differ from those of previous studies because of differences in the study population. The results of this study, in combination with other studies of voter mobilization, suggest that telecommunications do not seem to influence voter turnout substantially.

Smith, Jennifer K., Alan S. Gerber, and Anton Orlich. 2003. "Self-Prophecy Effects and Voter Turnout: An Experimental Replication." Political Psychology 24 (3): 593-604.