Nickerson 2006d - Thirteen Email Experiments Across Eight States
Mass email GOTV messages in thirteen experiments exhibit negative effects on registration and turnout that are both substantively and statistically insignificant.
Nickerson, David W. 2006. "Demobilized by e-Mobilization: Evidence from Thirteen Field Experiments." Unpublished Manuscript. Department of Political Science, University of Notre Dame.
David Nickerson presents the results of thirteen experiments testing the extent to which e-mail messages can be effective in increasing voter registration or mobilizing turnout. Three non-partisan organizations randomly divided their target audience into a treatment group, which received a series of emails encouraging registration and turnout, and a control group that received no attention from the organization. Given that the thirteen combined experiments involved 232,716 subjects, any differences in registration and turnout between the treatment and control groups would have been directly attributable to receipt of the emails. However, no mobilization from email was detected with regards to either registration or turnout. Therefore, the conclusion is drawn that e-mail is not currently a const-effective means of boosting rates of registration or increasing voter turnout.
Votes for Students 2002, Five Universities:
Electoral Context: The five Votes for Students experiments were carried out in anticipation of the 2002 Congressional elections
Subject Population: Subjects were drawn from lists of student names and e-mail addresses that were purchased from or donated by five universities (California Polytechnic Institute, Eastern Michigan University, University of Minnesota Twin Cities, University of California at Irvine and University of Georgia Atlanta) during the 2002 Congressional election. While all students on the list were assigned to either treatment or control groups, only students living in specified regions were included in the analysis.
Randomization Procedure: Students with valid e-mail addresses were randomly assigned to control and treatment groups.
Treatment: The treatment groups received a battery of email messages encouraging voter registration and voter turnout, while students in the control group received no communication from Votes for Students prior to Election Day. Because the email addresses were provided by the school administration, only a handful of addresses were found to be undeliverable. Similarly, very few students made use of the "opt out" link in the Votes for Students email messages. The average open rate across the five schools in the experiment was 20% with a high of 26% at the California Polytechnic Institute and a low of 11% at the University of Georgia in Atlanta. Subjects deleting emails prior to reading the messages may still have been "treated" to some extent by virtue of the blandishment to vote included in the subject line.
Youth Vote 2003, Houston
Electoral Context: . The Youth Vote Houston experiment was conducted during the November 2003 election in Texas.
Subject Population: The target sample consisted of roughly 145,000 registered voters under the age of 26 in Houston. The list of registered voters was given to a political consulting/lobbying firm specializing in Internet outreach, who matched registered voters against the subscriber list of large Internet Service Providers in the area. The email addresses were only for individuals who at some point indicated that they were willing to receive messages from third parties. The 13,185 (a yield of roughly 9%) subjects with a matching email addresses were then emailed an invitation to participate. The 12,772 individuals who did not opt out over the next few days (a yield of 98%) were then randomly placed into treatment and control groups.
Randomization Procedure: Individuals from the target sample were randomly assigned to treatment and control groups.
Treatment: The treatment group in each city was sent a series of 3 emails leading up the election. Emails typically began with a short quiz and an invitation to explore the Youth Vote website. The conclusion of the email included a brief blandishment to vote. Because the Youth Vote Coalition is a non-partisan organization that seeks to engage young people in politics, the content of the email did not endorse a candidate, party, or side in a referendum. Open rates were not available for the campaign although click-through rates to the Youth Vote website were measured between 5-8% for each email.
Working Assets 2004, Seven States
Electoral Context: The seven Working Assets experiments were conducted prior to the 2004 national election.
Subject Population: Subjects were drawn from Working Assets lists of voters living in California, Colorado, Michigan, Minnesota, Missouri, North Carolina and Clark County, Nevada.
Randomization Procedure: Individuals were randomly assigned to treatment and control groups.
Treatment: Treatment group subjects were sent a series of email blandishments to vote prior to the election. Control group members were sent no emails.
Findings: Pooled results from the twelve Votes for Students and Working Assets suggest a slightly negative and statistically insignificant effect of email on voter registration: - 0.3 percentage points with a standard error of 0.2 percentage points. Pooled results from all thirteen experiments (including Youth Vote Houston) suggest a slightly negative and statically insignificant effect of email on voter turnout among registered voters: -0.2 percentage points with a standard error of 0.3 percentage points.