The impact of universal vote by mail and permanent absentee voting on turnout in 2020 and 2022

Last updated June 4, 2024

Authors and Analysts
Debra Cleaver (CEO, VoteAmerica)
Jennifer Lauv (Data Engineer, VoteAmerica)
Scott Minkoff, PhD (Associate Professor, SUNY New Paltz and Senior Research Fellow, VoteAmerica)

Map 1. 2024 vote by mail rules by state*

*10 states have semi-permanent absentee voting and will accept one request for all elections in the calendar year. One state, Florida, has semi-permanent absentee voting where voters can submit one request for all elections during a general election cycle (example: 2023 and 2024, or 2025 and 2026).

Key Takeaways

• In universal vote by mail (UVBM) states, all registered voters are sent a ballot in the mail without needing to proactively request one.  In permanent absentee voting (PAV) states, voters can sign up once to receive ballots in the mail for all subsequent elections. Both methods significantly reduce the obstacles in the way of voting by automatically bringing the polling place to the voter rather than making the voter go to the polling place. 
• In 2020, turnout in UVBM states was 0.6 percentage points higher than turnout in battleground states, where battleground states are defined as those where partisan groups spent at least $90 million on broadcast TV. 
• Turnout in UVBM states was higher across all age cohorts. Notably, turnout among younger voters (18 to 24) in UVBM states was about 3.7 percentage points higher than in battleground states.
• PAV adoption is high in both Montana and Arizona: in 2022, it was 74.1 percent of registered voters in Montana and 75.5 percent of registered voters in Arizona. The differences in turnout between those with and without PAV status were substantial.
• Overall, PAV turnout was about 30 percentage points higher than non-PAV turnout in Arizona’s 2020 and 2022 elections. The difference was about 20 percentage points for Montana. Differences persist consistently across age, gender, race/ethnic, and voter propensity groups.
• Our multivariate regression analysis that controlled for age, gender, marital status, race, voter history, and county turnout found an average turnout lift from PAV of 13.53 percentage points for Montana 2022 and 12.89 percentage points for Arizona 2022. 
• While states like Arizona and Montana have been using PAV for some time, it is newer in several states including DC, Maryland, Michigan, Minnesota, New Mexico, New Jersey, and Virginia. Work will need to be done to build and ease adoption among existing and new voters.    

Section 1. Overview: Examining turnout effects for PAV and UVBM

This memo presents evidence on the efficacy of universal vote by mail (UVBM) and permanent absentee voting (PAV). With UVBM, all registered voters are sent a ballot in the mail without needing to proactively request one. With PAV (also known as “single signup”), voters sign up once to receive ballots in the mail for all subsequent elections.

Like other state voting laws (ID, early voting, etc.), UVBM and PAV may have a significant impact on whether people actually turnout to vote. Both methods significantly reduce the obstacles in the way of voting by bringing the polling place to the voter rather than making the voter go to the polling place. VoteAmerica has conducted a number of analyses aimed at assessing this impact and find that both methods substantially boost voter turnout.

Map 2: 2020 Universal vote by mail and battleground states

*In 2020, Montana counties had the option of using UVBM elections. 46 of 56 counties opted for UVBM. 

Section 2. The impact of Universal Vote By Mail (UVBM) on turnout

To assess the impact of UVBM, we compared 2020 aggregate turnout in battleground states with aggregate turnout in universal vote by mail states. Battleground states were defined as those where partisan groups spent at least $90 million on broadcast television. These states were presumed to be the most competitive ahead of Election Day, they were the places where the most partisan and nonpartisan voter mobilization spending occurred, and they are the states where turnout is typically expected to be the highest. It also happens to be the case that none of them had UVBM in 2020 (see map 2 above for UVBM and battleground states). As such, battleground states provide a good, conservative comparison with UVBM: How did turnout in UVBM states compare with the states where there was the most external pressure to vote? [1]

The results of this analysis are presented in Graph 1 by age cohort [2]. Overall, we find that UVBM state turnout was slightly higher than battleground turnout in all age categories. Notably, and somewhat remarkably, turnout in UVBM states was 3.7 percentage points higher among those 18 to 24 and 2.4 percentage points higher among those 25 to 34. This suggests that UVBM can be particularly helpful at boosting turnout among younger voters. 

Graph 1. Universal Vote by Mail v. Battleground States (2020): Aggregated general election turnout by age cohort

Consider Utah. Utah was not a competitive state in 2020: President Trump won by just over 20 percentage points, there was no Senate race that year, only one of its 4 House districts was remotely close, and Spencer Cox won the governor’s race by more than 30 points. Overall turnout among active registered voters was 89.33 percent, which is higher than any of the battleground states except Wisconsin (which had a turnout rate that was only 0.21 points higher). 

An even starker contrast appears when you compare UVBM states with the states that were the least supportive of vote by mail in 2020. These are states that require an excuse to receive an absentee ballot and some also require a witness signature on the ballot. The states that required an excuse in 2020 were Indiana, Louisiana, Mississippi, Tennessee, and Texas (note that many states waived their excuse requirements during the pandemic). Not surprisingly, turnout among voters in UVBM states was 6.8 percentage points higher than in the states that made it difficult to vote by mail. The difference was biggest among the youngest voters with UVBM turnout higher by 12.5 percentage points among those 18 to 24 and 8.6 percentage points higher among those 25 to 34. 

Graph 2. Universal VBM states v. excuse required states (2020): Aggregated general election turnout by age cohort

Section 3. PAV adoption rates across age, gender, race, and propensity score

Before examining the impact of PAV on turnout, we look at overall PAV adoption rates and how they vary across groups in these two states. PAV adoption is high in both states: In 2022, it was 74.1 percent of registered voters in Montana and 75.5 percent of registered voters in Arizona [3]. Graph 3 shows PAV adoption across levels of voter propensity and demographic groups. PAV adoption mostly increases with voter propensity. PAV status among low-propensity voters in Arizona is notably low (below 50 percent) relative to the other propensity groups. We also observe that PAV adoption rates increase with age and are at least slightly higher among women (relative to men), those who are married (relative to those who are single), and those who are White or Asian (relative to those who are Black, Hispanic, or Native American). In particular, PAV status among Native Americans is below 60 percent in both states. [4]    

Graph 3. Montana and Arizona (2022): Permanent absentee voter (PAV) status by group

Section 4. The effect of PAV status on turnout in Arizona and Montana

We conducted both overall turnout rate comparisons and more sophisticated regression analyses to identify the turnout effects of PAV. The turnout comparisons compare turnout among those with and without PAV status, broken down by age and voter propensity. The regression analysis uses a statistical model to estimate the average change in the likelihood of voting associated with PAV status, controlling for other factors typically associated with turnout. Both analyses yield the same result: PAV status significantly boosts voter turnout.

Section 4.1. Turnout comparisons

Here we look at three elections across two states where PAV was an option for voters: Arizona in 2020 and 2022 and Montana in 2022 [5]. PAV status varies among voters within each state that offers it (some people have opted in, some have not). This means we can hold state election-effects constant and examine how PAV status influences turnout in the state, no need to compare across state lines.

When we compare turnout between registered voters who have and have not adopted PAV, we find that turnout is substantially higher among those who have opted-in to PAV. This is true in both states and across age cohorts (graphs 4 and 5), gender groups (graph 6), race/ethnicity groups (graphs 7 and 8), and voter propensity groups (graphs 9 and 10). 

In Arizona, there is a consistent difference of just less than 30 percentage points overall and across age cohorts. In Montana, the difference is about 20 percentage points overall and between 15 and 20 percentage points across age cohorts, with higher differences among younger voters. With respect to gender, we observe turnout differences between 18 and 30 percentage points. Notably, the turnout gap is greater among men such that PAV sign-up boosts turnout among men a bit more than it increases turnout among women. The differences across race and ethnic groups look similar; interestingly, the turnout difference is consistently greater among the non-white groups in all three elections. For example, PAV turnout was 24.7 percentage points higher among white people in Arizona’s 2020 election, but 33.6 percentage points higher among Black voters, 33.7 percentage points higher among Asian voters, 32.6 percentage points higher among Hispanic voters, and 31.7 percentage points higher among Native American voters. 

When comparing across voter propensity groups, we find that the turnout difference is largest among lower-propensity voters. By way of example, in Arizona 2020, PAV turnout was higher by 33.3 percentage points among low-propensity voters, 23.9 percentage points among mid-propensity voters, and 12.5 percentage points among low-propensity voters. In Arizona 2022 and Montana 2022, the differences were a bit less but PAV voters still turned out at a much higher clip than non-PAV voters. 

Graph 4. Arizona Permanent Absentee v. All Other Registered Voters (2020 & 2022): General election turnout by age cohort

Graph 5. Montana Permanent Absentee v. All Other Registered Voters (2022): General election turnout by age cohort

Graph 6. Arizona and Montana Permanent Absentee v. All Other Registered Voters (2020, 2022): General election turnout by gender

Graph 7. Arizona Permanent Absentee v. All Other Registered Voters (2020, 2022): General election turnout by race and ethnicity

Graph 8. Montana Permanent Absentee v. All Other Registered Voters (2022): General election turnout by race and ethnicity

Graph 9. Arizona Permanent Absentee v. All Other Registered Voters (2020 & 2022): General election turnout by voter propensity

Graph 10. Montana Permanent Absentee v. All Other Registered Voters (2022): General election turnout by voter propensity

Section 4.2. Regression analysis 

While the turnout comparisons above have the luxury of being very straightforward, they lack the ability to control for multiple variables simultaneously so as to ensure that the observed effect of PAV is not being impacted by some other variable – this is particularly helpful for understanding the impact of voter propensity because it is a function of other variables. To accomplish this goal of better isolating the effect of PAV, voter-level 2022 turnout regression models were estimated for Montana and Arizona that included a variable indicating whether the person had PAV status or not, while controlling for county turnout, voter propensity, age, marital status, gender, and race/ethnicity. Moreover, one of the challenges of estimating PAV effects is that there is a likely selection effect such that those who opt-in to PAV are probably more likely to vote than those who do not. The regression approach is helpful with this because we can estimate the turnout boost at different levels of voter propensity while holding the other variables constant.

The regression analyses indicate that the average turnout lift from PAV status was 13.53 percentage points for Montana and 12.89 percentage points for Arizona (Graph 11). Like most voter mobilization tactics, we also find that this effect was largest among mid-propensity voters (Graph 12). For mid-propensity voters we observe a turnout boost between 15 and 20 percentage points, compared to a boost between 5 and 10 percentage points for low- and high-propensity voters. 

The turnout comparisons above found that the biggest PAV-related turnout gap was among low-propensity voters. This individual analysis indicates that once other variables are controlled for simultaneously, it is the mid-propensity voters that get the biggest bump from automatically receiving a ballot in the mail–this is in keeping with most voter mobilization interventions. Even among high propensity voters, who are very likely to vote regardless of whether they automatically receive a ballot in the mail or not, we observe a 5 to 10 percentage pointer turnout boost from PAV status. 

We also find that the effect of PAV reduces slightly with age. For example, PAV lift for an 18 year old Montanan was about 13.71 percentage points compared to 11.24 percentage points for an 83 year old Montanan. The difference was slightly less in Arizona. 

Graph 11. Montana and Arizona (2022): Effect of PAV status on turnout (with 95% confidence interval)

Graph 12. Montana and Arizona (2022): Effect of PAV status on turnout at different levels of voter propensity (with 95% confidence interval)

Section 5. Conclusion

When we talk about voter mobilization tactics we generally talk about external interventions that increase the likelihood that a person will vote in an election. Organizations are consistently seeking out new and different tactics that will get the attention of voters and bring them to the polls. What this analysis reminds us is that some of the best tactics for boosting turnout come from the government itself. Automatic mail voting (either by way of UVBM or PAV) brings the polls to the people and produces turnout lift that is beyond what can reasonably be expected from standard external mobilization interventions. 

The results presented in this brief study are not perfect. UVBM cannot be directly tested because it is constant within state (there is no one in a UVBM state that does not vote by mail to compare to). The effect of PAV adoption on turnout is mitigated by the expectation that those who opt in to PAV are probably more likely to vote in the first place. Nevertheless, what we observe in our analyses strongly supports significant turnout effects from receiving a ballot in the mail without needing to consistently request it. The ballot itself becomes the mobilization tactic and it is an effective one.  

Reaping the turnout benefits of UVBM and PAV will require systematic change in states that do not currently have those systems in place. However, there are a number of states that have more recently implemented PAV systems and voters need to be made aware that it is now an option. Campaigns to increase PAV adoption in these states will require interventions that are subject to all the challenges – and limited effect sizes – that interventions to increase turnout face. However, the effects of PAV adoption treatments will have the benefit of persisting into the future as ballots will continue to come in the mail even if organizations stop reminding people to vote. 

Footnotes

[1] We treat Arizona as a battleground and not a UVBM state. However, it is worth noting that about 75 percent of its registered voters are permanent absentee voters (PAV). 

[2]  This is aggregated turnout which means that the total number of active registered voters in all states in each group (UVBM and battleground) is added together and divided by the total number of people who voted. It effectively treats each group as its own combined election. 

[3]  Vote propensity is defined based on TargetSmart’s 2022 voter turnout model scores, where low-propensity is 0-33, mid-propensity is 34-66, and high-propensity is 67-100.

[4]  This is not surprising given the challenges that Native Americans face with getting their mail (https://abcnews.go.com/Politics/experts-worry-push-2020-mail-voting-leave-native/story?id=70411683). 

[5]  Amidst the COVID-19 pandemic, most Montana counties in 2020 utilized UVBM.