Mann 2006 - Absentee Ballot Recruitment
Receiving an absentee ballot in the mail mobilizes absentee voters, but is not found to increase overall voter turnout.
Mann, Christopher. 2006. "Cost Effective Voter Mobilization: An Experiment on Recruiting Absentee Voters." Unpublished Manuscript. Institution for Social and Policy Studies, Yale University.
Christopher Mann reports the findings of a randomized field experiment examining the effects of a single absentee ballot mailer on voter turnout in a Colorado statewide election in partnership with the Colorado Conservation Voters Education Fund (CCVEF). Mann argues that absentee ballot recruitment is logistically easier to execute and can reach more geographically dispersed voters than canvassing. Results from the experiment suggest that receiving an absentee ballot in the mail increases the likelihood that an individual will vote by absentee ballot. Absentee ballot voter recruitment does not however increase overall turnout with the exception of some segments of voters when analyzed by patterns of voter history.
Electoral Context: This experiment was conducted during the 2005 Colorado statewide election in partnership with the Colorado Conservation Voters Education Fund (CCVEF). Colorado allows "no excuse" absentee voting, but voters must sign up to vote absentee each year. The only statewide items on the ballot in November 2005 were Referenda C & D, known collectively as the Economic Recovery Act. Referendum C lifted the revenue limitations imposed on Colorado's state budget the Taxpayer Bill of Rights initiative in 1992 and Referendum D allowed bonding against some of the Referendum C revenues for transportation and other capital projects. The campaign for and against these ballot measures set Colorado records for the amount of money spent on ballot measures and odd year election turnout. As a result of extensive TV advertising and other paid communications, public awareness of this election was quite high for an odd year election. Referendum C passed 52%-48%. Referendum D failed 49%-51%. There were also numerous local ballot measures and some local elections for city government and school board.
Subject Population: CCVEF is a 501(c)3 public charity dedicated to informing and mobilizing pro-conservation voters. CCVEF's voter turnout program targets registered voters in Colorado who have been identified as concerned about the environment through membership in environmental groups or past identification programs by environmental groups to recruit voters to take part in civic activities. In 2005, 36 of Colorado's 65 counties choose to conduct the November vote on Referenda C & D entirely by mail. This experiment was conducted in the 6 largest counties which did not conduct the election by mail. These counties include: the City and County of Denver, the Denver suburban counties of Adams and Broomfield, El Paso County (including the City of Colorado Springs), Pueblo County in southern Colorado, and Weld County in northern Colorado.
Randomization Procedure: There were 9,219 low-propensity voters propensity pro-conservation voters (voted in two or fewer of the past four elections) on CCVEF's pro-conservation voter list. A control group of 1,816 (20% of the universe) was randomly selected from CCVEF's pro-conservation list and assigned to be in the treatment group. The remainder of the list was in the control group.
Treatment: The 2005 absentee ballot recruitment mailer was addressed to each individual in the treatment group of pro-conservation voters (see Appendix B for a copy of the mailer). It is possible that a household received multiple mailers if multiple individuals in a household were on CCVEF's list of pro-conservation voters. The mailer included two tear-off absentee ballot request postcards which were pre-addressed for return to the appropriate County Clerk's office. The voter was required to fill in her or his information, sign the application, place a stamp on it, and mail it. The mailer was sent on October 14, 2005 and the deadline for returning the postcard applications to the appropriate County Clerk was October 21, 2005. Ballots had to be received by the county clerks no later than the close of polls on November 1st. The total average cost of each mailer for design, printing, sorting, and postage was 49.7 cents. CCVEF's internal costs for producing the mailer were $1,869 (salary and overhead for 40 hours of staff time) or 25.3 cents per mailer. The average total unit cost of each mailer was 65 cents per mailer.
Findings: This experiment is only able to report the effect of being assigned to the treatment group (being sent the mailer). Mann finds that being sent an absentee ballot recruitment mailer (Treatment) caused a statistically significant increase 2.8 percentage points (p-value: .004) in the probability of casting a vote in 2005 by absentee ballot. Increases in the number of absentee ballots cast does not necessarily translate into increased overall voter turnout. Absentee ballot recruitment may simply cause voters to substitute voting by mail for voting in person on Election Day. However, Mann finds that absentee ballot recruitment mailer (Treatment) caused a 1.1 percentage point increase in overall voter turnout (p-value=0.146, one-tailed test). Controlling for past voting, Mann finds that that there is no statistically significant effect on turnout among registered voters who have voted rarely in the past four elections (0 or 1 out of 4 elections) or among frequent voters (voted in 3 or 4 of the past 4 elections). Receiving an absentee ballot increased the likelihood that voters who voted in two of the last four elections voted by 5 percentage points (p-value: .004, 1-tailed test).