*Swing Blue Alliance (SBA) volunteers sent postcards to a targeted group of WI voters with the goal of increasing election turnout for the more liberal nominee running for an open seat on the WI State Supreme Court. The postcards yielded less than a half percentage point increase in the overall turnout, which did not reach statistical significance. However, further in-depth analysis revealed that among targeted voters aged 45 and under turnout was boosted by 1.3%- a significant effect. These results suggest that postcards can have a deciding impact on voter turnout when voters are carefully targeted, and elections expected to be close.
On April 4, 2023, WI held a statewide election to replace a justice who retired after 20 years on the WI Supreme Court. The retirement set off a highly charged battle for ideological control of the Court. The election offered voters an opportunity to change the complexion of the court whose conservative majority typically voted together. With continuing battles over state policies governing abortion rights, voting rights, and redistricting, liberals sought to replace the conservative judge with a liberal leaning judge thereby flipping the court to a more liberal leaning majority.
In collaboration with Indivisible, Swing Blue Alliance conducted an experiment to evaluate the impact of a GOTV (get-out-the-vote) postcard campaign to elect the more liberal candidate in the WI Supreme Court election. Designed as a randomized controlled trial comparing the April 4 voting records of persons who were sent a postcard prior to the election (the Treatment group) with the records of persons who were not sent a postcard (the control group) offered an opportunity to determine the impact of the postcards on turnout. Assuming general equality between the two groups, the degree of improvement in turnout rates of the Treatment group over the Control group would give a measure of the extent of the impact and determine its significance.
Voters in the treatment group were mailed postcards prior to the election. The postcards included a QR code that opened a webpage containing voter information. SBA volunteers hand wrote on the postcard (1) the voters first name, (2) a short question “What’s your plan to vote?”, and (3) the sender’s first name. Additional data collected during the study found that only a handful of voters (115) opened the webpage by scanning the QR code. (pictures of a sample of the front & back of the cards)
The sample frame for the study included 80,000 registered WI voters with high prochoice scores (greater than 70) and a promising off-year election voting score (between 10 and 44 inclusive). The prochoice score was suggestive of a voter who was more likely to vote for a candidate who supports prochoice initiatives. The off-year election score was suggestive of a voter who would more likely vote in an off-cycle election. The criteria were developed by Indivisible and applied to Indivisible’s voter files, from which we drew two groups of 40,000 names that matched the criteria.
Prior to analyzing the results, we studied the success of the randomization procedure by comparing the treatment and control group on critical variables, including age, race, gender, partisanship score, index of support for Biden, and voting records from 2018, 2020, and 2022. The randomization was highly successful. There was not one significant difference between the two groups concerning any variable of interest. Most of comparisons were nearly identical and within a percentage point of the other. This result provided confidence that the two groups were identical in the most important ways.
The study results revealed an overall increase of .45 percent in voter turnout for the postcard group. While nearly half of a percentage point improvement, the increase was not significantly different from zero (0). Thus, we cannot be confident that the increase is due to the postcard alone.
Fortunately, further analysis of the data improved our understanding of using a GOTV postcard in this study. Employing various forms of regression analyses we found that the small overall effect of the postcard masked a much greater effect among some younger voters. In general, the postcard increased voter turnout among voters 45 years old and under by 1.3 percentage points; a statistically significant finding. Another finding further suggested that voters 25 years old and under actually had a decrease in voter turnout, though that was not significant.
This study confirms what other similar research has found: that carefully targeted GOTV campaigns with well-designed postcards have small effects on voter turnout, and that these effects can impact results in a tight election where voter turnout is small. While the QR code went largely unused, the postcards held a significant and positive effect on voter turnout among the 26 to 45 year old age groups. (Swing Blue Alliance Research Team) (Loewenstein, 2024)