Election officials are constantly verifying the eligibility of voters.

Elections professionals across the country are constantly working to make sure the voter rolls are current and accurate. Many check with other government agencies to make sure that they are allowing people to vote when they have already proven voting eligibility through another government interaction. They verify drivers’ license and social security information provided by voters with federal and state government databases to ensure registrations are valid. And they train poll workers to address any questions that emerge around a person’s identity at the polls.

Policymakers are constantly evaluating proposals to change how we verify voter eligibility.

Policymakers in Congress, state legislatures, and election agencies regularly consider changes to how election officials verify the residency, identity, and citizenship of voters. Some of these proposals would actually change the eligibility requirements themselves. These are proposals like making people with felony convictions eligible to vote or changing the voting age to 16 or 21.

Other proposals leave the current eligibility requirements in place but change the processes to verify that voters meet these requirements. For example, currently voters can prove eligibility by providing the last four digits of their Social Security Number on the national voter registration form. But new proposals would require voters to bring a physical copy of a passport, birth certificate, naturalization certificate or other citizenship document to their local election office in order to register to vote. In another example, some states with laws requiring voters to show photo ID at the polls accept a government issued ID from another state while other states require the ID to be issued by the state where the person is voting. These nuances in how voter eligibility is determined can have big consequences for how many eligible voters are disenfranchised because they are not able to comply.

3 ways to measure who lacks the proof of residency, identity, and citizenship they need to vote

In order to consider proposals for how we verify voter eligibility, policy makers and the public need a clear understanding of how many eligible voters would not be able to vote or would struggle to vote if the proposal was implemented. The population is in constant flux. People turn 18 and become eligible to vote every day. People die. People go to the DMV. People get married. Every day people access new forms of identification or change their names. People become US citizens and get naturalization papers. Others misplace or lose access to their own vital records through weather related disasters, divorce, or disorganization. Gaining an accurate and timely estimate of who lacks easy access to the particular vital records they might need to verify their eligibility to vote in a wide range of policy scenarios is an ongoing challenge.

Here are some questions policymakers (and you!) might have when considering a policy proposal to change how voter eligibility is verified:

• How many eligible voters don’t have easy access to the documents they need to verify they are eligible to vote under the proposed policy?

• Are some groups of eligible voters more likely to lack easy access to the documents they need to verify they are eligible to vote under the proposed policy?

• Could state and local election officials implement the proposed policy in a way that doesn’t exclude eligible voters or include people who ineligible to vote?

There are three main complementary strategies we use to answer these questions.

Strategy 1: Questions on existing large surveys!

Surveys work because when a small group of people is selected at random from the general population, their responses can accurately predict what is happening to everyone. Statistics are amazing!

And surveys are able to more accurately predict what is happening in the general population and among key subgroups when they include more responses (assuming those people are sampled in a representative way from the general population). One really good strategy for researchers is to pool their resources, get a really big representative sample, and ask a lot of questions! Some examples of studies like this include the American National Election Studies (ANES), the Cooperative Election Study (CES), and the Survey of the Performance of American Elections (SPAE). These studies often include some questions about whether respondents have and can access a driver’s license, birth certificate, passport, and other types of documents.

In 2023, our team at CDCE began our partnership with VoteRiders to study who lacks the proof of residency, identity, and citizenship they need to vote with an exploratory analysis of responses to questions about ID possession in the ANES.

Our colleague Charles Stewart at the MIT Election Data and Science Lab has done similar analysis using the SPAE.

• Strengths of using existing questions on large surveys: This strategy is relatively fast, inexpensive, and precise! The data are already available, the analysis can be done by any research assistant with basic survey analysis skills, and subgroup analysis is often possible (i.e. how many men between 50 and 65 have easy access to their birth certificate?) because the sample size is so large that subgroup results (for many groups) are valid.

• Weakness of using existing questions on large surveys: The questions are almost never precise enough estimate how many voters would actually be disenfranchised by any particular proposal. So we can have a sense of how many people don’t have a passport or drivers’ license but we aren’t sure if those particular documents could be used for voting purposes. For example, we often don’t know if the documents have their current name and/or address or are unexpired, which is critical to knowing if they can be used for voting purposes in many states. We also don’t know if they have another type of document that they could use for voting purposes.

Strategy 2: Field new original surveys with questions tailored to specific policy proposals!

Fielding an original survey allows researchers to ask more precise questions that are tailored to the details of policies and policy proposals in particular places. For example, the national survey we fielded with partners in 2024 asked questions about a long list of potential forms of ID so that we could more accurately estimate how many eligible voters could not comply with policies requiring voters to show ID to vote.

We also asked about “easy access” to citizenship documents which became the basis for the stat that 21.3 million U.S. citizens lack easy access to documentary proof of citizenship which has been cited widely in debates about the SAVE Act.

At the state level, fielding original surveys allows us to tailor questions to the idiosyncrasies of particular state laws and get a sample that is representative of that state. For example, our Texas survey asks questions about ID expiration dates because Texas allows people to use expired IDs only if they have not been expired for more than 4 years (or if the voter is over 70). This crucial detail allows us to estimate not just how many people have certain documents but whether those documents are sufficient for voting purposes when asked to a set of respondents that is representative of that particular state. The significant variation in the proposals that end up being considered and adopted in different states requires us to do this type of state specific work to get clear answers to the top questions policy makers have in those places.

• Strengths of fielding original surveys: Original surveys give researchers control over both the questions and the sample. This allows us to tailor questions and sampling to particular policy proposals and geographies and respond directly to the most pressing questions policymakers and the public have about these laws.

• Weaknesses of fielding original surveys: It’s really expensive, time consuming, and requires a lot more expertise! Even with significant resources, the sample size of a new survey fielded by an individual researcher will never approach the scale of the ANES or CES samples. So the precision of the questions is higher but the extent to which answers are predictive of the general population and subgroups can be lower than with large surveys.

Strategy 3: Use administrative data about how many people obtain certain documents and whether they use them to vote!

The last way that scholars figure out who lacks the proof of residency, identity, and citizenship they need to vote is by looking at “administrative data” that is publicly accessible from federal, state, and local governments. This type of data could be data from the DMV about who is getting IDs or licenses, or data from state or local election agencies about which documents people are using to vote.

Here at CDCE, we have done some work in this area by matching data from the Maryland Motor Vehicle Administration about the age of people getting driver’s licenses in Maryland by county with data from the Maryland State Board of Elections about the age of Maryland voters. This has helped us understand why some Maryland counties are struggling to serve all their eligible 18-year-old voters.

Other scholars and journalists have done great work using administrative data to understand who lacks the proof of residency, identity, and citizenship they need to vote.

To explore which eligible voters lack the proof of citizenship they need to vote, Kae Petrin and Jen Fifield of VoteBeat did an analysis of Arizona’s voter rolls in 2024, which showed voters without proof of citizenship on file were less likely to vote overall and more likely to live on tribal land, college campuses, or homeless encampments. Michael McDonald did an extensive analysis in 2016 studying how Kansas implemented a law requiring voters to present documentary proof of citizenship to vote in state elections. He found that young voters and politically unaffiliated voters were more likely to be purged from rolls because they did not yet have proof of citizenship on file with the election agency.

Numerous studies have also used administrative data to study who lacks the proof of identity they need to vote. Phoebe Henninger, Marc Meredith, and Michael Morse used affidavits signed at the polls by people without ID in Michigan and Bernard Fraga and Michael Miller have done similar analysis with “reasonable impediment declarations” in Texas. John Johnson has used DMV data in Wisconsin to estimate who lacks photo ID there. These are just a few examples of this kind of work.

• Strengths of using administrative data: Administrative data is precise. It’s an observation of something that occurred in the world. It helps us capture how things are actually implemented and not just how we predict people would be affected based on responses to a survey.

• Weaknesses of using administrative data: It’s not always available! We can only analyze the data we can get from the government. And it’s hard to use administrative data to study people who are not accessing vital records and not voting because they don’t show up in the data. This is the most important population to understand when considering these types of policies.

The upshot? When the challenge is that the laws and population are constantly changing…continuous engagement with different forms of expertise is the answer!

There is no single research project that can fully answer the question about who lacks the proof of residency, identity, and citizenship they need to vote. Ongoing collaboration between scholars from diverse disciplines, local election officials, advocacy groups, and other stakeholders is necessary to answer this question. This type of collaboration allow us to deploy the three strategies described above in a complementary and additive way to build a coherent base of knowledge. That’s why at CDCE we have engaged in an ongoing way with a research team that includes our faculty fellows, colleagues from other institutions, local election officials, advocacy and direct service organizations like VoteRiders and Foundry UMC ID Ministry, and others. This type of engagement - part of our broader commitment to “inclusive consensus” across our work - deepens our understanding of this issue in powerful ways.

We hope to continue providing clear and actionable research to shape and inform debates about how voters verify their residency, identity, and citizenship! Please let us know what questions you have about this topic or if you have ideas or suggestions for where to take this line of work.

Read on Substack