Effect of Vote-by-Mail Postcards With and Without Return Postage on Behavior of Registered Democrats in Sarasota County

February-March 2020

Investigators

Peter Grey, Richard Segan, Julia Rabkin, JoAnne DeVries, John Loewenstein, Susan Labandibar

Abstract

A previous study demonstrated that a two part postcard was effective in increasing enrollment of registered Democrats in Vote-by-Mail (VBM) in Sarasota County Florida in the fall of 2020.  The current study sought to determine the effect of putting a stamp on the reply postcard to the local Supervisor of Elections (SOE).  Reply postcards without return postage were more effective than no intervention over the period from February 1, 2020 to March 31, 2020 (7.36% enrollment versus 6.61%, P<0.00578).  In the same period reply post cards with return postage were more effective than those without return postage (12.2% versus 7.36%, P<0.0001).  Enrollment overall increased during this period, perhaps because a primary election was scheduled for March.  The COVID-19 epidemic may have also encouraged people to enroll in VBM.

Objective: Determine whether pre-paid return postcards increase enrollment in Vote by Mail (VBM).

Study population: 42,682 Registered Democrats in Sarasota County Florida who were not enrolled in VBM.

Design: Prospective cohort study. 

Three groups were part of the study.  Different voter precincts were chosen for inclusion in each group.
Group 1) Voters who were mailed a two-part postcard encouraging them to enroll in VBM.  The return tear off card to be returned to the local SOE did not include postage.
Group 2) Voters who were mailed the same two-part postcard, with postage on the return tear off reply card to be returned to the local SOE.
Group 3) Voters who did not have any mailing from Swing Left Greater Boston.

Study period: Data was collected on the number of registered Democrats in each precinct, and the number of registered Democrats enrolled in VBM in those precincts, for the period February 1, 2020 to March 31, 2020.

Introduction

Studies have shown that Vote-by-Mail (VBM) programs increase voter turnout in Florida. Enrolling registered Democrats in Florida’s VBM is one of the top priorities of the Florida Democratic Party.

In 2018, overall Florida Democratic voter turnout was 64.2%. Democrats who were registered to VBM turned out at 87.6% whereas those who were not registered to VBM turned out at 55%. In 2020, the Florida Democratic party anticipates that over 90% of Democrats enrolled in the VBM program will participate in the general election.

In Florida, registered voters can request an absentee ballot for all elections through the end of calendar-year 2022.  A VBM request form may be made by the voter, the voter’s legal guardian, or an immediate family member of the voter or the voter’s spouse. Voters who request to VBM will receive an absentee ballot 35 days before every election at the mailing address they specify on the request form.

In a previous study, conducted in the fall of 2019 in a small number of precincts in Sarasota County, two-part postcards were effective in increasing Democratic voter enrollment in VBM.  Part of the two-part card could be torn off and sent to the Supervisor of Elections to enroll in VBM.  The individual seeking to enroll had to add postage to the return card.  In this study we sought to evaluate whether identical postcards with return postage already applied would yield a higher rate of enrollment.

Methods

The study was conducted in Sarasota county with the study period February 1, 2020 through March 31, 2020.  Postcards were sent to registered Democrats not yet enrolled in VBM, over a period from about mid-February to early March.  Precincts were selected for voters to receive either no intervention, a two-part postcard with no stamp on the reply card to the Supervisor of Elections (SOE) or a two-part postcard with a stamp on the reply card to the SOE.  The selection was at the precinct level as this is how the SOE provided the data.  The SOE provided the total number of registered Democrats and the number of registered Democrats who enrolled in VBM in each precinct.  They sent the data monthly.  For example, end of March totals were sent in early April. The number of new VBM enrollments was divided by the number of registered Democrats who were not signed up for VBM (for each category (control, no stamp on reply card,  stamp on reply card), the calculation was (number registered for VBM as of March 31 – number registered for VBM as of Jan 31) / (total registered Democrats as of Jan 31 – Democrats registered for VBM as of Jan 31).  This yielded the percent of newly targeted registered Democrats who responded positively to the intervention.  The number of registered Democrats not signed up for VBM are referred to later in this paper as the targeted voters.

For testing significance we used a z score calculator for 2 population proportions. (socscistatistics.com).

Results

The study population consisted of 42,682 registered Democrats who were not enrolled in VBM at the beginning of February.  Of these, 15,161 were in Group 1 (no postage on return card), 6002 in Group 2 (postage on return card), and 21,519 in Group 3 (to whom we did not send postcards). There were 1,116 new enrollments in VBM in Group 1, 732 in Group 2, and 1424 in Group 3.  (See table 1.) 

As cards were mailed from about mid-February to mid-March, to evaluate the total effect of the postcards, we used the 2 month totals to calculate the number of new enrollments in VBM and the number of targeted voters for this period.  Group 1 voters enrolled at a rate of 7.32%, Group 2 voters at 12.2%, and Group 3 voters at 6.6%.  (see Figure 1.)

The differences in enrollment rates were statistically significant (Group 2 vs. Group 1 (P<0.0001), Group 2 vs. Group 3 (P<0.00001), and Group 1 versus Group 3 (P<0.00578).

Cost Effectiveness

Sending postcards with no stamp on the reply card to the SOE yielded an additional 0.76% enrollment in VBM over the control situation.  It cost $0.103 to print one postcard, $0.35 for one stamp, and $0.02 for 2 sealing labels to secure the foldover of the 2-part card, for a total cost of $0.473. Sending 100 cards would cost $47.30 and yield 0.76 additional enrollments; therefore each additional enrollment cost $62.24. Sending 100 cards with an $0.35 postage stamps on each reply card to the SOE cost ($0.473 + $0.35)X(100), or $82.30. This would yield an additional 5.6 voter enrollments over the control situation, or $14.70 per enrollment. Postcards with stamps on the reply card were therefore 4.23 times as cost effective as one stamp cards.

Discussion

In this prospective study with a large sample size, we demonstrated that two-part postcards with return postage were significantly more effective than two-part postcards without return postage.  The selection of precincts for the control group, no stamp on the reply card, and stamp on the reply card group was done at the precinct level.  Precincts were selected for inclusion in the study arms on the basis of whether their size fit the needs of the postcard writer groups. The precincts not chosen became the control arm. Nothing was known about the precincts other than their size.

A primary election was scheduled in Florida for March 2020.  Voter registration typically increases in the months just prior to election.  Various groups, both partisan and non-partisan contact potential voters during this period to encourage registration and voting.  It is therefore not surprising to see increases in both voter registration and VBM enrollment in all groups during the entire study period.  The COVID-19 epidemic began growing significantly in the United States in February and March.  This may also have motivated voters to enroll in VBM.

VBM conversions were higher in both intervention groups as measured at the end of both February and March. Our hypothesis that cards with a pre-paid return to the SOE would result in a higher rate of enrollment in VBM was validated.  We also found that cards with return postage are about four times more cost effective than cards without return postage.

There are many variables that influence voter decisions to enroll in VBM.  These change over time and can be specific to local factors.  We therefore cannot be sure that our results will apply to other districts, states, or regions.

Table 1. Targeted Voters and New VBM Enrollments

Figure 1.  Percent of registered Democrats newly enrolled in Vote-By-Mail February-March 2020