Gaby Goldstein, SDAN Director of Research;
Mallory Roman, SDAN Associate Director of Research
We ran a randomized controlled trial to determine: 1) if people who received a handwritten postcard encouraging voter registration, along with an official voter registration form from Voter Participation Center (VPC), in our March and Sept 2018 studies had higher odds of turning out to vote in the November 2018 general election; and 2) if the effect of those postcards differed based on whether or not those targets had returned the registration form sent to them by VPC earlier in the year.
The point of voter registration is to expand the pool of voters. But registered voters don’t always show up to vote, come election day. While little is known about the effect of postcarding on voter registration, even less is known about its downstream consequences for voting behavior.
Sister District Action Network (SDAN) and VPC partnered on two voter registration experiments in March and September 2018. In each, SDAN volunteers sent handwritten postcards to eligible, unregistered individuals living in Michigan, Arizona, and Pennsylvania. These postcards were sent in conjunction with 3 of VPC’s voter registration programs, which sent pre-filled official voter registration forms to all of the people included in this experiment. You can read about the March study here and the September study here. The current analysis went a step further and looked at whether or not the targets of the March and September studies were more likely to vote in the general election in November 2018, and further, if the effect of the handwritten postcard on voter turnout was higher among people who had returned the pre-filled VPC registration form.
1. Voter Registration Postcard Recipients Did Not Vote More Than Non-Recipients.
- People who received a handwritten postcard along with a pre-filled voter registration form did not vote at higher rates than people who did not receive the postcard.
- People who received a postcard before (p = 0.841) or after (p = 0.177 in March; p = 0.276 in September) receiving a pre-filled voter registration form did not have significantly higher (or lower) odds of voting in the November 2018 general election than people who did not receive postcards.
2. Chaser Recipient Returners Had a Slight, But Not Statistically Significant Boost for Turnout
- Among people who received handwritten postcards encouraging them to register to vote after receiving VPC’s pre-filled registration form (Chaser Recipients), who then returned the completed registration form (Responders), having received the handwritten registration postcard had a suggestive effect on turnout. However, the effect did not rise to statistical significance.
- March Chaser Recipient Responders voted at a 2.6% higher rate than March Responders who were not Chaser Recipients (they didn’t get a postcard). Similarly, September Chaser Recipient Responders voted at a 2% higher rate that September Responders who were not Chaser Recipients (they didn’t get a postcard).
- For the March sample, this interaction is not statistically significant for people who received a postcard before the VPC registration form (Primers, p = 0.349)or Chaser Recipients (p = 0.146). For the September sample, the interaction is also not significant for Chaser Recipients (0.782).
- Note, that we see an effect but it is not statistically significant may be due to the study being underpowered to detect the effect (in other words, we may not have enough people in the experiment for the results to rise to statistical significance).
3. Basic Takeaway.
Overall, sending eligible, unregistered individuals a handwritten postcard encouraging them to register to vote had no effect on later voter turnout. However, there is some suggestive evidence that receiving a handwritten postcard encouraging the target to register to vote did provide an extra boost in voter turnout among people who returned the VPC registration form in order to register to vote. Both studies are underpowered, so it is possible that these effects would have risen to statistical significance in a larger sample (specifically a larger sample of people who registered to vote through the VPC+SDAN registration efforts).
- Studies Underpowered. This Voting Pipeline Study, and the underlying March and September studies, were each statistically underpowered, which means we didn’t have enough people in the experiment to be able to rely on the findings. This study would need to be replicated with a larger sample of people to get a better picture of whether encouraging people to register to vote using handwritten postcards has a downstream effect on turning those people out to vote later on.
- Non-Representative Sample. This study focuses on VPC’s voter registration targets, who are considered part of the rising American electorate (young people, people of color, and unmarried women) and are not generalizable to the entire population of voting-eligible Americans.
Differences Between Samples. Since the voting pipeline analyses were conducted on existing data from two different samples (our March and September voter registration studies), those differences may affect the outcome of the voting pipeline analyses as well. There are two main reasons that we might expect differences between the results of the March and September Studies to affect the voting pipeline analyses:
— First, there are far fewer treatment participants (people who got a postcard) in the September sample than in the March sample, and as noted above, the March Study is already statistically underpowered.
— Second, there were more voters in the September Study sample than in the March Study sample. This indicates that people who registered closer to the election (unregistered people we targeted in September 2018) voted more than people who registered further out from the election (unregistered people we targeted in March 2018).
Contributions and Future Directions:
Beyond the initial effects on voter registration, this analysis helps us understand how postcarding alongside a voter registration effort may (or may not) affect the later choice to vote. Unsurprisingly, people who registered to vote during the VPC+SDAN registration efforts voted at a much higher rate than people who did not. This suggests that encouraging newly registered voters to vote may be a good area for future research. We will continue to look at the downstream voter turnout effects of other voter registration studies.
A more detailed report of these findings 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 standardized/contextualized effect sizes for all models tested. 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. These findings were peer reviewed by an independent statistician.