Gerber, Green, Iyengar and Jackman 2005 - Interactive CDs and Young Voter Education
Findings from one of two experiments are equivocal about the effects of CDs containing voter education information targeting young voters due to low contact rates.
Gerber, Alan, Donald P. Green, Shanto Iyengar and Simon Jackman. 2005. "Using Information Technology to Mobilize Young Voters: A Field Experiment." Paper presented at the Annual Meeting of the American Political Science Association, Marriott Wardman Park, Omni Shoreham, Washington Hilton, Washington, DC, September 1, 2005.
Alan Gerber, Donald Green, Shanto Iyengar and Simon Jackman examine the effects of two different versions of CDs designed to educate young people about voting, mobilize voter turnout, and encourage political engagement. Gerber et al. argue that new media forms of canvassing could be effective in mobilizing young voters because the interactivity of CD/DVD or web content provides the opportunity to present information in a way that could intrinsically motivate voters, which is more likely to lead to repeating the action of voting than external or situational motivations such as classical GOTV efforts. Gerber et al. present the designs of two experiments of young voter mobilization by means of CD content across six states, the results of one of the studies. Due to a low contact rate, results that would indicate how effective CD campaigns may ultimately be in mobilizing young voters are inconclusive.
Electoral Context: The experiment was carried out in the context of the 2004 presidential election.
Study 1: The authors sent "Every Vote Counts" CDs to a representative sample of 15,000 18-24 year old registered voters in California, New York, North Carolina, Florida, Pennsylvania and Oregon, "six-state study."
Study 2: The authors drew a representative sample of 10,000 18-24 year olds with internet access. Polimetrix randomly assigned 10,000 young voters into a CD condition and 6,000 young voters to a control group. Voters in both groups filled out a survey after the election probing self reporting of presidential vote, sense of political efficacy and interest in the campaign. In this study, the authors have attitudinal and behavioral indicators of political participation.
Randomization Procedure: For study 1, in each of the six states, 2,500 subjects were randomly assigned to the treatment condition from a list of registered young voters for a pooled treatment group of 15,000 subjects. The remaining young registered voters were randomly assigned, therefore, to the control group (N=3,782,743 young voters). For study 2, 10,000 individual voters were randomly assigned to the treatment group and 6,000 voters were randomly assigned to the control group.
Treatment: The authors created two versions of a CD: one was interactive with games and puzzles and the other was informational. CDs were divided into six thematic segments and included information on youth voting, the candidates, issues that might be of interest to youth, candidate positions, ads and candidate speeches. In addition to the afore-mentioned content, the interactive version included games and puzzles.
Between October 12 and 17, the CDs were mailed to approximately 15,000 voters from the six-state list and 10,000 from the Polimetrix list. 1,500 CDs were returned with invalid addresses. When participants put the CDs in their computers they were directed to sign-in by entering a five digit number and their date of birth. If the computed had an internet connection, information was transmitted to the Political Communication Lab at Stanford and entered into a database. For participants who successfully completed the sign-in process, the CD randomly assigned them to the interactive or informational version of the CD. Of the recipients, 1,894 successfully carried out the sign-in providing an 8% contact rate. Respondents off the Polimetrix list (young voters with access to the internet and interest in politics) responded at a higher rate (13.4%) than the young voters in the six-state sample (3%).
Findings: Because data is not yet available for the study using the Polimetrix list, the Gerber, et al, only present results for the "six-state study." The authors' estimates imply that the mailings produced a 0.3 percentage point (standard error 0.4 percentage point) decline in voter turnout. Including covariates does not change the treatment coefficient or standard error. The cost per vote was $330; only 2.7 % of the CDs were utilized. It is impossible to calculate the ATT effect because of the low contact rate.