Abstract: In 2020, SDAN ran a Two Primaries Postcard Study, which was a randomized controlled trial that found that primary voter turnout among voters with low-mid turnout propensity for presidential primaries (turnout scores of 25-60) and high Democratic support scores (partisanship scores of 80+) could be improved by sending such voters a handwritten GOTV/voter education postcard ahead of both the 2020 state and Presidential primaries in Florida, Georgia, and Minnesota. After the general election, we then took a look at whether voters who received primary GOTV postcards in that study might also have had higher odds to vote in the general election, and found that while this was not the case overall, the primary postcards did boost turnout among very high general election propensity voters.

In addition, we also conducted a substudy during the two primaries study: all of the participants from Minnesota and Florida in the original study were included, and half of the people who received the primary postcards received an additional GOTV postcard before the November general election. This substudy sought to determine if a third handwritten postcard ahead of the general election would further boost general election turnout in 2020.

Objective: This analysis explores the efficacy of sending an additional GOTV postcard ahead of the general election in reactivating the increased voter turnout effects seen from receiving primary election postcards in the original study.

Background: SDAN’s Two Primaries Postcard Study increased turnout rates in both 2020 primary elections in Florida and Minnesota. In that study, participants were sent one postcard ahead of each of the two primary elections in Florida and Minnesota (voters in Georgia also received primary postcards but these data were removed from the analysis due to pandemic election issues in Georgia). In this substudy, we aimed to determine if voters who received a third postcard right before the general election, after having received postcards ahead of the primaries, might have higher odds of voting in the general election, compared to folks who didn’t receive any postcards, as well as folks who only received primary postcards.

Specifics

This substudy enrolled all of the participants in the original two primaries RCT (see the original study blog for more information). Only treatment subjects from the original study were eligible to be randomized into the substudy treatment condition (primary postcards + GOTV postcard in the general) or control condition (only postcards during the primaries). This yielded 3 groups: controls – participants from the original study that were always assigned to the control condition and have received no postcards from SDAN as at any point in 2020; primary postcards only – participants who received primary postcards in the original study but were assigned to receive no additional postcard in the substudy; and primary + general postcard – participants who received primary postcards in the original study and were also assigned to receive an additional postcard in the substudy ahead of the general election. This resulted in a sample size of 19,999. Among this sample, 5,000 received a postcard during the substudy. Sister District volunteers wrote the GOTV postcards and sent them to in-state partners on October 20, 2020 for local mailing. It is estimated they were delivered to homes between October 27-30, 2020.

Basic Takeaways

  • The substudy appears to have had no effect. A third postcard did not boost turnout.
    • The conditions in the substudy where people received postcards (people who received only primary postcards and people who received primary postcards as well as a GOTV postcard) had almost identical turnout in the general election, and were only 0.06% different.
    • Voters who received a third postcard just before the general election did not have higher odds of voting than people who had only received the postcards during the primaries (who had the highest turnout of all 3 substudy conditions), or people who had received no cards during the study.
  • The inclusion criteria for the original study and the fact that this study was a substudy within a larger study indicate that these findings should be seen as an educational starting point as opposed to the last word on determining the efficacy of additional GOTV postcards that chase an earlier primary postcard effort.
  • Since the postcarding conditions had higher raw turnout numbers than the control group, even though not statistically significant, these results suggest that the effects of this tactic might rise to significance in a larger sample. Also, this study targeted voters based on primary turnout scores, so future studies may consider running a similar study with a group of voters with lower likelihood to turnout to vote in general elections.

Key findings

  • People who received a general election GOTV postcard in addition to the primary postcards from the original study voted at a slightly higher rate than control participants, but this difference was not statistically significant.
    • Voter turnout among voters who received the additional general election GOTV postcard was 89.82%, while voters in the control condition (who received no postcards in either the study or substudy) turned out to vote at a rate of 89.29%. The difference between these rates (0.53%) was not statistically significant (p = 0.385). This indicates that receiving an additional GOTV postcard in the mail was associated with voting in the general election, but it was not statistically meaningful in this sample.
  • Effects did not vary for people who received primary postcards, regardless of whether they received general election postcards as well.
    • Neither postcard condition was statistically significant compared to the control condition, but both had larger effects, as demonstrated by their higher turnout rates. However, voters who received primary postcards only, with no additional general election GOTV postcard actually had the highest voter turnout in the general election. This indicates that the additional general election GOTV postcard did not improve voter turnout. The primary + general postcards condition was actually associated with slightly lower turnout than primary postcard recipients who did not receive a general election GOTV postcard.

Caveats and considerations

  • The substudy could only enroll original study participants. The original study targeted specific districts in a narrow range of states, and only voters with primary voter propensity scores of 25-60 and partisanship scores of 80+. That criteria remains for the substudy, and thus likely makes this data ungeneralizable.
  • This sample was targeted based on primary turnout scores. The targets were originally chosen for their low-mid propensity turnout scores for primary elections. However, voters were considered much more likely to vote in the general election than in the primary, changing this sample into a base voter sample by the general election. ,
  • This study was couched in another study. This substudy operated within the confines of another study, meaning that both the original study and this substudy are influenced by one another. That makes the effects of substudy impossible to divorce from the effects of the original study. This tactic should be replicated on its own to determine its true efficacy.

Contributions and Future Directions:

This substudy joins the other research that SDAN has done on the efficacy of postcards. In the case of the substudy, there was no discernible effect of sending another postcard during general election GOTV, but there was some suggestive evidence that the primary postcard receivers did vote at a higher rate (though this was not statistically significant). This is an intriguing suggestive finding, especially as this study is statistically underpowered, and it should be explored further. If primary tactics can influence voters into the general election, and an additional general election nudge was unhelpful, this could be reason to contact voters more during the primary season, especially in states with two primaries (because this allows for ongoing contact).

Further, this substudy was the first study SDAN has done to look at the effect of layering multiple waves of mail, with some treatment voters getting 2 waves and some voters getting 3 waves. This allowed us to see the added benefit of sending an additional postcard, though in this case, we did not find such a benefit. More research is needed to determine if and when layering voter contact is an effective strategy to boost turnout.

If you’re interested in reading more about this study, a longer report can be found here.

SDAN’s commitment: It is SDAN’s intention to provide as much context as possible to allow for the nuanced interpretation of our data. SDAN’s convention is to contextualize effects by reporting p values, confidence intervals, and effect sizes for all models tested (these items may be in the longer report linked in the blog). Additionally, SDAN always differentiates between planned and exploratory analyses and a priori and post hoc tests, and reports the results of all planned analyses regardless of statistical significance. If you have questions about these findings please email Mallory.