Last month, the MIT Election Lab released the 2016 Election Performance Index (EPI), which sought to provide a measure of election performance in all 50 states during the 2016 general election. To do so, the index uses several key variables.
When the EPI was first developed, the inclusion (or omission) of the individual indicators it uses was not taken lightly. Each one was carefully considered and debated. One of the attributes that the EPI measures as part of its evaluation is the voter registration process, or the way by which eligible voters are placed on the voting rolls before casting their ballots.
How much control do states have over voter registration?
Since the National Voter Registration Act (NVRA) of 1993, states have been required to allow voters to register through a wide variety of means, including state agencies, schools, and libraries, unless they were exempted because they had Election Day registration. In that light, it might seem that state election administrators are fairly constrained in how they handle the voter registration process. And if that’s the case, why should voter registration be part of an index like the EPI at all—if states are limited in how they manage a particular piece of the election process by national policy or law, why should that factor into an evaluation of their election management?
However, while each state adheres to national requirements, voter registration processes still vary considerably from state to state. Many states now allow voters to register online. A few states (such as Maine) still limit remote registration to old-fashioned post cards. There has been a renewed interest in some states in beginning Election Day registration, and 13 states have now passed laws to implement automatic registration for eligible voters. Clearly, within the overarching national framework, election administrators and policymakers in each state remain reasonably in control of the process.
Why does the voter registration process matter?
There is a strong consensus that registration processes matter for voters on several fronts. First, voter registration requirements have been shown to be linked to turnout. Using data from Ohio and New York, Ansolebehere and Konisky demonstrated in 2006 that implementing voter registration requirements reduced turnout by 3–5 percent. This result was later corroborated by Burden and Neiheisel in their analysis of the implementation of voter registration in several Wisconsin counties, where they found that new barriers to voting led to a 2 percent decrease in voter turnout.
These studies support the idea that the various “costs” of registering to vote — monetary costs, the cost of taking time off at work to register in person, or simply the mental cost of confusing processes — depress voter turnout because the perceived costs of voting outweigh the perceived benefits. However, that impact has not been as definitive in research from the past few years. In a recent blog post, University of Wisconsin professor Barry Burden notes explicitly that the relationship between registration and turnout is not exact. Instead, there is a tendency for turnout in a given state to lag behind registration rates, seemingly indicating to scholars that there is little election officials can do to enhance turnout, at least through changes to voter registration processes.
While this may lead to questions about the value of including voter registration processes in the EPI, I argue that this is instead another example of the “Roseanne Roseannadanna” principle that MIT Election Lab founder Charles Stewart referred to in a recent Monkey Cage post: it’s always something. Often, seemingly random issues arise to challenge election administrators: hanging chad in 2000; long lines in 2008. With voter registration processes, their history of impacting turnout, paired with the variations in policy and process between states, means that it remains important to include them in the EPI.
How does the EPI measure the voter registration process?
The EPI looks at voter registration processes through two variables: registration or absentee ballot problems, and rejected voter registrations. Why these two? There are a couple of reasons why they can serve as a reasonable proxy for voter registration processes.
First, previous research (see Rosenstone and Wolfinger and Powell, for example,) has shown that problems with the registration process are one of the most frustrating barriers to voters wishing to cast a ballot. As we have seen above, registration processes have been shown to affect turnout; measuring these variables allows the EPI to look more closely at whether registration processes are impacting voters in each state.
Second, problems with registration and rejected registrations may be seen as an indicator of a confusing or opaque process, either to the voter or to officials who process registrations. This is similar to evaluating surveys for confusing questions, which have been shown to lead to mistaken responses. A more appropriate analogy might be to confusing mail ballots, which have been shown to lead to increased residual vote rates (see Alvarez, Becket, and Stewart).
Finally, these two variables allow us to look at voter registration processes in a way that is agnostic to the individual policy positions taken by politicians and election officials. The EPI largely doesn’t care how a state improves the quality of its registration process; it just cares that it does. (The exception to this statement is that the EPI does care whether a state offers online voter registration, and whether a voter can verify his or her registration status online.)
Voter registration and the EPI: adapting to new challenges
In this post, I have sought to further clarify the reasons behind including the voter registration process in the EPI, and the specific variables used as proxies for the registration process. Even while following national requirements laid out in laws such as the NVRA or the Help America Vote Act, states have extensive autonomy in how they implement their voter registration processes. This fact, compounded with the consensus that the registration process can serve as a barrier to voter turnout, means it remains an important indicator in any metric for election administration. Finally, relying on rejected registration data and self-reported registration problems allows the EPI to analyze the process and its impacts in a way that is strongly grounded in theory and methodology, while keeping it agnostic in regards to the political landscape and partisan election reform.
While the EPI may represent the voter registration process well, advances in the field of election management mean we should also be on the lookout for potential improvements in the 2018 and 2020 indices. As the election landscape changes, the EPI will also adapt, providing a powerful tool for understanding election administration for the foreseeable future.
