data for politics #27: Same Day Registration Can Increase Voter Turnout
By Charlotte Hill (@hill_charlotte) and Jake Grumbach (@JakeMGrumbach)
Voting reform is on the map in the 2018 elections. At least six states are considering updating how people register to vote, either through legislative action or ballot initiative. Some of these reforms, like Florida’s felon re-enfranchisement initiative, focus on boosting registration and turnout among specific populations; others, like a new Michigan ballot initiative, make it easier for people to register and cast a ballot on the same day.
Most voting reforms share a common goal: increasing overall voter turnout. (Even reforms explicitly aimed at increasing voter registration ultimately want higher turnout; after all, there’s little value in registering more voters if they fail to show up at the polls.) A critical question for policy researchers, then, is which reforms are most effective at getting people to vote.
Our analysis of census data is clear: relative to two other popular voting reforms—early voting and absentee voting—same-day registration (SDR) is great at boosting voter turnout. SDR has been adopted in 17 states, plus the District of Columbia.
SDR tackles one of the most annoying things about voting: registering to vote. When Election Day rolls around, many people are excited to get out and cast a ballot. But too often, people show up at the polling place only to find out they’re not registered. Maybe they recently moved. Maybe they’re living in a college dorm. Maybe their state election administrators mistakenly purged them from the voting rolls. Whatever the cause, registration problems are inconvenient and all too common.
It’s tough to estimate the effect of election reforms like SDR, since the states that implement these changes are likely to be different than those that don’t. We overcome this hurdle by using a “difference-in-differences” approach to analyze how voter turnout improved within states that adopted SDR. (Here’s a simple overview of how difference-in-differences works.) States that implemented SDR showed a 1.2 to 1.5 percentage-point increase in turnout. Early voting and absentee voting, by contrast, don’t appear to have much effect on turnout.
We also re-ran our analysis using a newer form of difference-in-differences (described here), which weights the estimates based on how likely states are to implement SDR. Using this new method, we find that SDR boosts turnout by more than 5 percentage points. Again, there is not much action from early voting or absentee voting reforms.
Why is SDR so effective at boosting turnout? For starters, it makes a complicated process much simpler. For many Americans, the seemingly straightforward act of registering to vote is a big pain. In fact, more than a third of unregistered voters say they want to register but simply “haven’t gotten around to it” or find registering “not convenient.” SDR reduces the amount of total effort it takes to register and vote. You no longer need to go somewhere to register, wait for confirmation that you’re in the voting system, and then turn out to cast a ballot; with SDR, you can register and vote at the same time.
SDR also has the nice side-benefit of improving the effect of early voting laws, which, on their own, can actually decrease turnout. To be sure, early voting is great at giving existing voters more flexibility. But it can also make Election Day seem less significant, and it reduces incentives for political campaigns to get out the vote in the final days of election season (because, presumably, many people have already turned in their ballots).
As a 2014 study on early voting put it, “local news coverage, discussions with peers, and Election Day activities all help spur turnout by providing information about candidates and the process of voting, introducing some normative pressure to vote, and enhancing the social benefits of taking part in a collective enterprise.” Early voting laws dilute these activities by spreading them out over time, and this leads to lower Election Day turnout.
Supplementing early-voting laws with SDR gives us the best of both worlds. Existing voters get more flexibility, which makes them more likely to keep voting each election. New voters benefit from being able to register and vote at the same time—either before Election Day, or on Election Day itself. Yet of the 37 states with early voting laws on the books, only 15 states also have same-day registration. The takeaway is clear: voting reform advocates should make implementing SDR in the remaining 22 early-voting states a top priority.
Matching Design
Our diff-in-diff design exploited variation within states across time to estimate the effect of SDR. Matching, on the other hand, exploits variation across states by comparing similar individuals in states that do or don’t have SDR. In particular, we match individuals in the “control” group (no SDR) to those in “treated” group (with SDR) based on their race, income, age, gender, education, and election year. (We use exact matching on race, gender, and year.) In total, we match 55,685 treated individuals with 55,685 control individuals.
We use genetic matching, which searches the data to find the best matches between similar individuals. The demographics of the “treated” and “control” units before and after the matching procedure are shown above. As you can see, after matching, treated and control individuals have very similar demographics on average. We weren’t able to match especially well on education, but we matched individuals perfectly on race, gender, and election year, and nearly perfectly on income and age.
A paired t-test gives us an average treatment effect on the treated (ATT) that is very similar to our earlier estimates: SDR increases an individual’s likelihood of turnout by about 1.72 percentage points (95% CI: 1.189, 2.245; p-value < 0.001).
Jake Grumbach (@JakeMGrumbach) is a Postdoctoral Fellow at the Center for the Study of Democratic Politics at Princeton University.
Charlotte Hill (@hill_charlotte) is a Ph.D. student of public policy at UC Berkeley.