Voter Purges

Purges:

A Growing Threat to

the Right to Vote

by Jonathan Brater, Kevin Morris,

Myrna Pérez, and Christopher Deluzio

Brennan Center for Justice at New York University School of Law

About the Brennan Center for Justice

The Brennan Center for Justice at NYU School of Law is a nonpartisan law and policy institute that works to reform, revitalize — and when necessary defend — our country’s systems of democracy and justice. At this critical moment, the Brennan Center is dedicated to protecting the rule of law and the values of constitutional democracy. We focus on voting rights, campaign finance reform, ending mass incarceration, and preserving our liberties while also maintaining our national security. Part think tank, part advocacy group, part cutting-edge communications hub, we start with rigorous research. We craft innovative policies. And we fight for them — in Congress and the states, in the courts, and in the court of public opinion.

About the Brennan Center’s Democracy Program

The Brennan Center’s Democracy Program works to repair the broken systems of American democracy. We encourage broad citizen participation by promoting voting and campaign finance reform. We work to secure fair courts and to advance a First Amendment jurisprudence that puts the rights of citizens — not special interests — at the center of our democracy. We collaborate with grassroots groups, advocacy organizations, and government officials to eliminate the obstacles to an effective democracy.

About the Brennan Center’s Publications

Red cover | Research reports offer in-depth empirical findings.

Blue cover | Policy proposals offer innovative, concrete reform solutions.

White cover | White papers offer a compelling analysis of a pressing legal or policy issue.

its entirety as long as the Brennan Center for Justice at NYU School of Law is credited, a link to the Center’s web pages is provided, and no charge is imposed. The paper may not be reproduced in part or in altered form, or if a fee is charged, without the Center’s permission. Please let the Center know if you reprint.

Acknowledgments

The Brennan Center gratefully acknowledges the Carnegie Corporation of New York, Change Happens Foundation, craignewmarkphilanthropies, craigslist Charitable Fund, Democracy Alliance Partners, The Ralph and Fanny Ellison Charitable Trust, FJC – A Foundation of Philanthropic Funds, Ford Foundation, The JPB Foundation, Susheel Kirpalani, The Kohlberg Foundation, Betsy Krieger, Open Society Foundations, Barbra Streisand, and Vital Projects Fund for their generous support of our voting work.

The authors would like to thank the numerous Brennan Center colleagues without whose help the publication of this report would not have been possible. Brennan Center President Michael Waldman and Democracy Program Director Wendy Weiser provided valuable strategic and drafting guidance for this report. Research and Program Associate Phoenix Rice-Johnson made substantial research contributions throughout the project. Isabella Aguilar, Jaya Aiyer, Faith Barksdale, Max Feldman, Sean Morales-Doyle, Michael Pelle, Wendy Serra, Ani Torossian, and Makeda Yohannes also provided valuable research and editing assistance. The editorial and design assistance of Yuliya Bas, Lisa Benenson, Stephen Fee, Theresa Raffaele Jefferson, Jim Lyons, Zachary Roth, and Jennifer Woodhouse allowed this report to reach publication.

The authors are grateful to Dr. Terry-Ann Craigie for her peer-review of the econometrics presented in this report.

The authors sincerely appreciate the assistance and cooperation of the election officials and advocates who took time out of their busy schedules and allowed themselves to be interviewed for this report, and the Brennan Center gratefully extends our thanks to them for doing so.

About the Authors

Jonathan Brater is counsel for the Democracy Program at the Brennan Center, focusing on voting rights and elections. In this capacity, he has worked on litigation to block enforcement of restrictive voting laws and policies in state and federal court. An expert on voter registration law and policy, he has drafted legislation and published analysis on automatic voter registration and has testified before state and national bodies including the Presidential Commission on Election Administration. His work focuses on registration list maintenance and the prevention of harmful voter purges. His work also includes extensive analysis of state legislation affecting voting access. He graduated cum laude from Michigan Law School, where he served as Executive Editor of the Michigan Law Review He received a bachelor’s degree from Columbia. Prior to law school, he was a legislative analyst for the House Committee on Energy and Commerce.

Kevin Morris is a quantitative researcher with the Democracy Program, focusing on voting rights and elections. His research focuses on the impact of laws and policies on access to the polls, with a particular focus on rights restoration and voter list maintenance. Prior to joining the Brennan Center, he worked as an economic researcher focusing on housing at the Federal Reserve Bank of New York and an economist at the Port Authority of New York and New Jersey. He received a bachelor’s degree from Boston College and is currently pursuing a master’s degree in Urban Planning at NYU’s Wagner School, with an emphasis in Quantitative Methods and Evaluation.

Myrna Pérez is deputy director of the Democracy Program, where she leads the Voting Rights and Elections project. She has authored several nationally recognized reports and articles related to voting rights. Prior to joining the Brennan Center, she was the civil rights fellow at Relman & Dane. She graduated from Columbia Law School, where she was a Lowenstein public interest fellow. Following law school, she clerked for Judge Anita B. Brody of the U.S. District Court for the Eastern District of Pennsylvania and for Judge Julio M. Fuentes of the U.S. Court of Appeals for the Third Circuit. Her undergraduate degree is from Yale University, and she received a master’s degree in Public Policy from Harvard University’s Kennedy School of Government, where she was the recipient of the Robert F. Kennedy Award for Excellence in Public Service. She is currently an adjunct professor at Columbia Law School.

Christopher Deluzio is counsel for the Democracy Program, where he focuses on voting rights and elections. Prior to joining the Brennan Center, he was a litigation associate in private practice with Wachtell, Lipton, Rosen & Katz. He served as a law clerk to Judge Richard J. Sullivan of the U.S. District Court for the Southern District of New York. He graduated magna cum laude from Georgetown Law, where he was elected to the Order of the Coif, served as an executive articles editor of the Georgetown Law Journal, and was selected as the top oralist in the Robert J. Beaudry Moot Court Competition and the Thurgood A. Marshall Memorial Moot Court Competition. He received a bachelor's degree from the U.S. Naval Academy and, following graduation, served as an active-duty naval officer.

Table of Contents

Introduction ....................................................................... 1

Methodology ......................................................................2

The 2018 Purge Landscape ..............................................2

CURRENT FINDINGS

Purge Rates Are Higher Than a Decade Ago

Purge Rates Increased More in Jurisdictions
Previously Subject to Federal Oversight

States Continue to Conduct Flawed Purges

Federal Role in Voter Protection Diminished

New Flaws in Voter Purges ................................................ 6

Interstate Voter Registration Crosscheck Program (Crosscheck)

Electronic Registration Information Center (ERIC)

Efforts to Purge Noncitizens Are More Frequent and Often Rely on Flawed Data

Activist Groups Pressing for More Aggressive Purges

Solutions .............................................................................. 11

Endnotes .............................................................................. 12

Appendices ......................................................................... 24

Appendix A: Federal Statutory Regulation of Voter Purge Practices
Appendix B: What Explains a Jurisdiction’s Purge Rate?
Appendix C: Relationship Between Purge Rates and Provisional Ballot Rates

VOTER PURGES: A GROWING THREAT TO THE RIGHT TO VOTE

Introduction

On April 19, 2016, thousands of eligible Brooklyn voters dutifully showed up to cast their ballots in the presidential primary, only to find their names missing from the voter lists. An investigation by the New York state attorney general found that New York City’s Board of Elections had improperly deleted more than 200,000 names from the voter rolls.

In June 2016, the Arkansas secretary of state provided a list to the state’s 75 county clerks suggesting that more than 7,700 names be removed from the rolls because of supposed felony convictions. That roster was highly inaccurate; it included people who had never been convicted of a felony, as well as persons with past convictions whose voting rights had been restored.

And in Virginia in 2013, nearly 39,000 voters were removed from the rolls when the state relied on a faulty database to delete voters who allegedly had moved out of the commonwealth. Error rates in some counties ran as high as 17 percent.

These voters were victims of purges — the sometimes-flawed process by which election officials attempt to remove ineligible names from voter registration lists. When done correctly, purges ensure the voter rolls are accurate and up-to-date. When done incorrectly, purges disenfranchise legitimate voters (often when it is too close to an election to rectify the mistake), causing confusion and delay at the polls.

Ahead of upcoming midterm elections, a new Brennan Center investigation has examined data for more than 6,600 jurisdictions that report purge rates to the Election Assistance Commission and calculated purge rates for 49 states. ¹

We found that between 2014 and 2016, states removed almost 16 million voters from the rolls, and every state in the country can and should do more to protect voters from improper purges.²

Almost 4 million more names were purged from the rolls between 2014 and 2016 than between 2006 and 2008.³ This growth in the number of removed voters represented an increase of 33 percent — far outstripping growth in both total registered voters (18 percent) and total population (6 percent).

Most disturbingly, our research suggests great cause for concern that the Supreme Court’s 2013 decision in Shelby County v. Holder (which ended federal “preclearance,” a Voting Rights Act provision that was enacted to apply extra scrutiny to jurisdictions with a history of racial discrimination) has had a profound and negative impact:

For the two election cycles between 2012 and 2016, jurisdictions no longer subject to federal preclearance had purge rates significantly higher than jurisdictions that did not have it in 2013. The Brennan Center calculates that 2 million fewer voters would have been purged over those four years if jurisdictions previously subject to federal preclearance had purged at the same rate as those jurisdictions not subject to that provision in 2013.⁴

In Texas, for example, one of the states previously subject to federal preclearance, approximately 363,000 more voters were erased from the rolls in the first election cycle after Shelby County than in the comparable midterm election cycle immediately preceding it.⁵ And Georgia purged twice as many voters — 1.5 million — between the 2012 and 2016 elections as it did between 2008 and 2012.

Meanwhile, the Justice Department has abdicated its assigned role in preventing overly aggressive purges. In fact, the Justice Department has sent letters to election officials inquiring about their purging practices — a move seen by many as laying the groundwork for claims that some jurisdictions are not sufficiently aggressive in clearing names off the rolls.

This new report follows an extensive analysis of this issue in a 2008 Brennan Center report entitled Voter Purges.⁶ In that report, we uncovered evidence that election administrators were purging people based on error-ridden practices, that voters were purged secretly and without notice, and that there were limited protections against purges. In this year’s report, we discovered that little about purge practices has improved and that a number of things have, in fact, gotten worse.

This study also found:

●

In the past five years, four states have engaged in illegal purges, and another four states have implemented unlawful purge rules. Federal standards for purges were set in the 1993 National Voter Registration Act (NVRA). Since 2013, Florida, New York, North Carolina, and Virginia have conducted illegal purges. Moreover, Brennan Center research has uncovered that four states (Alabama, Arizona, Indiana, and Maine) have written policies that by their terms violate the NVRA and provide for illegal purges. Alabama, Indiana, and Maine have policies for using data from a database called the Interstate Voter Registration Crosscheck Program (Crosscheck) to immediately purge voters without providing the notice and waiting period required by federal law (Indiana’s practice has been put on hold by a federal court).

Arizona regulations permit Crosscheck purges during the 90 days prior to an election, a period during which federal law prohibits large-scale purges. These eight states are home to more than a quarter of registered voters across the nation.

●	States use inaccurate information. Although states have improved the way in which they use data to purge the voter rolls in some respects, several jurisdictions rely on faulty data to flag potentially ineligible voters. And some of the new sources of information that have come into widespread use since our 2008 report, such as Crosscheck, are especially problematic.
●	A new coterie of activist groups is pressing for aggressive purges. Most purging litigation brought by private litigants before 2008 contended that voter removal efforts were overly aggressive. Today, a different group of plaintiffs is hauling election officials into court, claiming that purging practices in their jurisdictions are not sufficiently zealous.
This report makes the following recommendations:
●	Enforce the NVRA’s protections. The NVRA, one of the major federal laws governing how states and localities can conduct purges, permits voters and civic groups to sue election officials if they violate the law’s provisions. Monitoring jurisdictions to ensure they are complying with the NVRA — and bringing litigation when necessary — is especially important in an era when election officials are under pressure to mount aggressive purges.
●	States should set purging standards that provide even more protections than the NVRA. The NVRA sets out federal standards for purges and requires that voters removed from the rolls for certain reasons be given notification. But these are minimum guidelines. States can and should do more to protect against disenfranchisement caused by improper purges — for example, providing public and individual notice before purging names from the rolls.
●	Pass automatic voter registration. Automatic voter registration is a popular reform that minimizes registration errors and allows for easy updates, making rolls more accurate and current.

Methodology

We analyzed purge statutes, regulations, and other guidance in 49 states.⁷ We interviewed 21 state or local election administrators in 18 states and reviewed documents from 20 states in response to public records requests.⁸ .

We also calculated state and county purge rates using voter registration data from the Election Administration and Voting Survey (EAVS), which is administered biennially by the U.S. Election Assistance Commission.⁹

Our analysis used EAVS data from the 2008, 2010, 2012, 2014, and 2016 reports. In each two-year period, we calculated a jurisdiction’s voter removal rate by dividing the number of removed voters by the sum of

registered voters (i.e., both active and inactive registered voters) and removed voters.¹⁰

The 2018 Purge Landscape

Between the 2014 and 2016 elections, roughly 16 million names nationwide were removed from voter rolls.¹¹ The federal law governing purges¹² allows a voter’s name to be purged from the voter rolls on the following grounds: (1) disenfranchising criminal conviction; (2) mental incapacity; (3) death; and (4) change in residence. In addition to these criteria, individuals who were never eligible in the first place, such as someone under 18 or a noncitizen, may be removed. Voters may be removed at their own request (even if they remain eligible). While all 49 states with voter registration lists have affirmative policies to remove names from the rolls (typically for several or all of the four delineated categories), states vary in the manner in and frequency with which they conduct voter purges.¹³

●

Disenfranchising Conviction

Except in Maine and Vermont, states disenfranchise at least some voters convicted of a crime for some period of time, which means that there are states that purge voters because of a criminal conviction. States have different policies about what causes a voter to become ineligible and different procedures for removing those who have been disenfranchised.¹⁴ They also draw upon different lists to identify individuals with felony convictions, which may in turn be maintained with different levels of regularity and precision by courts or law-enforcement officials at the state or federal levels.

●

Mental Incapacity

Though less ubiquitous than some other bases of removal, 28 states have specific rules requiring removal from the rolls of a person determined not to have mental capacity to vote.¹⁵ Definitions vary, and reform attempts have had some success limiting the instances in which those with alleged mental incapacity lose their right to vote.¹⁶

●

Death

Federal law mandates that states take steps to remove the deceased from the rolls. Yet there is no uniform standard among the various state laws detailing the sources of information to be consulted to determine which voters are deceased. Some jurisdictions use information from state agencies, some review obituaries, and some rely on the Social Security Administration’s Death Master File.¹⁷

●

Residency Changes

States vary in how they perform list maintenance for changes of address. Some of that variation is in timing. Montana, for example, conducts address removals every odd-numbered year,¹⁸ and Connecticut conducts address removals annually.¹⁹ There is also variation in which source of information is used. Two common sources are drivers’ license updates and the postal service’s National Change of Address (NCOA) database, but states also utilize other sources, such as interstate databases, returned mailings, or voter inactivity.

●

Noncitizenship

While election officials generally remove names of persons when it is made known to them that a noncitizen has gotten on the rolls, at least six states also have laws that require state officials to use jury declinations, drivers’ license information, and/or federal databases to actively identify noncitizens on the voter rolls, to remove names of noncitizens so identified, or both.²⁰

CURRENT FINDINGS

Purge Rates Are Higher Than a Decade Ago

In the two-year period ending in 2008, the median jurisdiction purged 6.2 percent of its voters.²¹ At one end of the spectrum in 2008, Salt Lake County, Utah, purged less than 0.1 percent of its voters, and at the other end of the spectrum, Milwaukee County, Wisconsin, purged more than 34 percent of its voters. Of the 2,534 counties that reported purge rates to the Election Assistance Commission in 2008, only 97 had purged more than 15 percent of its registered voters in a two-year period.

Between the federal elections of 2014 and 2016, almost 4 million more names were purged from the rolls than in 2006-08. In this same period, more than twice the number of counties — 205 — had purged more than 15 percent of their voters than between 2006 and 2008.

Although a higher removal rate is not inherently bad, more purging means increased potential for eligible voters to be removed, especially given that we identified no state with the desired level of voter protections against purges.

Purge Rates Increased More in Jurisdictions Previously Subject to Federal Preclearance

Prior to 2013, the Voting Rights Act required certain jurisdictions with a history of discriminatory election practices to obtain federal certification that any intended election change, including voter purge practices, would not harm minority voters and was not enacted with discriminatory intent. This monitoring process was known as “preclearance.”²² In 2013, however, the Supreme Court concluded in Shelby County v. Holder²³ that Congress had inappropriately determined which jurisdictions should be subject to preclearance. As a result, jurisdictions subject to (or “covered” by) preclearance requirements were freed from making the case that minority voters would not be harmed by a proposed election change.

Across the board, formerly covered jurisdictions increased their purge rates after 2012 more than noncovered jurisdictions. Before Shelby County, jurisdictions that were subject to preclearance requirements (“covered jurisdictions”) had removal rates equal to other jurisdictions (“noncovered jurisdictions”).²⁴ After 2013, the two groups sharply diverged. For the 2012-14 and 2014-16 two-year

FALLOUT FROM SHELBY COUNTY

Increases in purge rates in previously covered jurisdictions weren’t the only changes after Shelby County¹ Following the decision, many states and jurisdictions proceeded to enact or implement laws that would have been subject to preclearance. In fact, states formerly under preclearance requirements were more likely to pass legislation restricting their voting and election practices than the nation as a whole. Of the nine states once fully covered by the Voting Rights Act, seven have passed restrictive legislation since 2010. Of the 41 states not fully covered, only 18 passed restrictive laws over the same period. Two of these states (Florida and North Carolina) each had several counties subject to the Voting Rights Act.²

1 Shelby County v. Holder, 570 U.S. 2 (2013).

2 See Brennan Center for Justice, New Voting Restrictions in America, May 2017,
We include in this count legislation that was enacted and subsequently struck down by courts. See, e.g., Applewhite v.Pennsylvania, No. 330 M.D. 2012, 2014 WL 184988 (Pa. Commw. Ct. Jan. 17, 2014) (striking down Pennsylvania voter ID law).

election cycles, the removal rate for noncovered jurisdictions did not budge. The story was entirely different for covered jurisdictions, whose median removal rate was 2 percentage points higher after the Shelby County decision than the noncovered jurisdictions. Though 2 percentage points may seem like a small number, more than 2 million fewer voters would have been removed if these counties had removal rates comparable to the rest of the country. Previously covered jurisdictions ended up removing more than 9 million voters between the presidential elections of 2012 and 2016. These increases were not concentrated in just a few small counties: 67 percent of residents in previously covered jurisdictions lived in areas where the removal rate increased, compared to just 46 percent of residents in non-covered jurisdictions. These calculations are restricted to jurisdictions that reported their data each year, but there is evidence that the same trend happened in counties that did not report each year, as our Texas analysis below shows.

The increase in removal rates in counties previously covered by the preclearance provision is not attributable to geographical or partisan factors (see footnote 25 for more information). We also conducted a difference-in-differences regression analysis²⁶ to see if population, minority presence, income, or other factors could explain the increase in removal rates in these counties. Even after controlling for these factors, a jurisdiction’s former status under the Voting Rights Act was strongly associated with higher voter removal rates. Although this effect was larger in the two-year period coinciding with the lifting of the preclearance requirement, it continued even into the two-year period ending with the presidential election of 2016.

To be absolutely clear, our analysis cannot establish what percentage, if any, of these post-Shelby County purges were done erroneously. What we do know is that provisional ballots, which are given to voters who are missing from the voter rolls, had a statistically significant relationship to purge rates in previously covered jurisdictions.²⁷ This means that as the purge rates increased, so did the number of people who showed up to vote but were unable to do so, either because their names were not on the rolls or for some other reason.

Another factor is that between the presidential elections of 2012 and 2016, a handful of states implemented strict voter ID laws that required voters to cast provisional ballots if they did not have one of the limited number of accepted identifications. The implementation of these laws could, of course, have led to an increase in provisional ballot rates. (To isolate the impact of increased purge rates on provisional ballot rates, we performed a regression analysis in which we controlled for the implementation of strict voter ID laws and other sociodemographic factors. The regression specification and a closer look at a few counties with big increases in purge rates and provisional ballots can be found in Appendix C.)

The changes were particularly notable in three states: Georgia, Texas, and Virginia.

In Georgia, 750,000 more names were purged between 2012 and 2016 than between 2008 and 2012. Although Georgia did not report provisional ballot rates in 2012, their provisional ballot rates in the federal elections of 2010 and 2014 correspondingly increased as the removal rates increased. Of the state’s 159 counties, 156 reported increases in removal rates post-Shelby County. This included the state’s 86 most populous counties. The increased purge rate occurred during a period when Georgia was criticized for several controversial voter registration practices. For example, Georgia was sued for blocking registration applications between 2013 and 2016 because information (including hyphens in names) did not match state databases precisely. Georgia agreed to cease the matching rule as a result of the lawsuit but then enacted legislation reinstating a very similar practice the next year.²⁸

Texas did not report removal rates for the two years ending in 2012 and is thus excluded from our high-level analysis of the previously covered jurisdictions. Nonetheless, the state exhibited a substantial increase in removal rates when we compare the two-year periods ending with the federal elections of 2010 and 2014. Between 2012 and 2014, approximately 363,000 more voters were removed than in 2008-10.²⁹ Unsurprisingly, the provisional ballot rate also increased between the midterm elections of 2010 and 2014. Consistent with the broader trend, these increases were not driven only by small counties: Fourteen of the 20 most populous counties increased their removal rates. Of the 183 Texas counties that reported their removal rates in both periods, 121 saw an increase after the Shelby County decision. Among the Texas counties that consistently reported their data and increased their removal rate after the Shelby County decision, the median increase was 3.5 percent. This increased purge rate did not occur in isolation but was joined by restrictive voting legislation. In 2014, a federal district court ruled that the strict photo ID law that Texas passed in 2011 was motivated in part by a discriminatory purpose of reducing minority political participation.³⁰ The Court of Appeals of the 5th Circuit did not decide whether the law was motivated by discriminatory animus but did conclude it had a discriminatory effect.³¹ In 2017, Texas passed a new voter ID law. Litigation regarding the new law is ongoing.

In Virginia, previously covered counties removed 379,019 more voters between 2012 and 2016 than between 2008 and 2012. Once again, the increase in purge rates in these counties was not driven by small counties purging more voters. All the previously covered counties except one increased removal rates after Shelby County. The one previously covered county that showed a decrease — Highland County — is the least populous county in the state, home to just 2,230 people. More than 99 percent of Virginia’s voters live in counties that increased their removal rates after Shelby County. As later discussed in more detail, a contributing factor may have been a highly problematic purge process that Virginia mounted in 2013.

States Continue to Conduct Flawed Purges

Broadly speaking, purges go wrong for one of two basic reasons: bad information about who should be removed from the rolls or a bad method for removing them. There are tools to catch and correct these mistakes, some of which are legally mandated. For example, federal law sets forth some important and relevant safeguards, such as requiring that systematic purges — those in which voter rolls are compared with lists of potentially ineligible individuals to remove groups of voters at the same time — occur well in advance of an election. Another is making sure certain categories of voters get a notice and waiting period before removal.³² Yet as both a legal and practical matter, many states lack sufficient safeguards to detect and correct problems so that any harm can be repaired in advance of an election.

Two states’ recent experiences illustrate the basic reasons purges go wrong — Arkansas used bad information, while Texas used a bad method.

In June 2016, the Arkansas secretary of state sent county officials a list of more than 7,700 records from the Arkansas Crime Information Center (ACIC) of persons who were supposedly ineligible to vote and should be removed from the rolls.³³ (Those convicted of felonies in Arkansas lose their right to vote until their sentence is complete or they are pardoned.³⁴) But the list included a high percentage of voters who were indeed eligible,³⁵ yet appeared on the list because they had had some involvement with the court system, such as a misdemeanor conviction or a divorce.³⁶ Also included were names of those whose voting rights had been restored.³⁷ The error became public in July 2016, and despite the public outcry, the records of fewer than 5,000 of the more than 7,700 erroneously listed voters had been corrected by September 2016.³⁸ Pulaski County, the largest county in the state, explained that the problem was flagged by the counties, not the state, and not all counties were able to correct errors.

Previously, the secretary of state had not been providing counties with regular updates of conviction data and, in the past, had been using the wrong source list for data on felony convictions. Once Arkansas switched to the list required by law, the secretary did an overly broad match and provided counties with inflated lists with bad matches. Pulaski County flagged the errors and was able to investigate the list, but some counties with insufficient resources simply sent purge notices to everyone on the list.³⁹

Texas is an example of a bad purge caused by flawed data matching. In 2012, Texas officials conducted a purge of voters presumed to be dead. According to a representative from the Texas secretary of state’s office, the purge was driven by a comparison of Texas voters’ information to the Social Security Administration’s Death Master File — the first time Texas had conducted such an exercise.⁴⁰ Matching to the Death Master File was required under

a then-new Texas law (H.B. 174) mandating election officials to obtain such information about potentially deceased voters quarterly.⁴¹

While the 2008 Brennan Center report on voter purges showed that the Death Master File can contain errors,⁴² the problem in Texas occurred because the state used what are called “weak” matches (meaning that the chances that the person identified was actually deceased were too low to be trusted) to target voters without conducting any further investigation.⁴³ For example, a voter whose date of birth and last four digits of their Social Security number matches a dead person’s record would be a “weak” match.⁴⁴ On these grounds, a living Texas voter (and Air Force veteran) named James Harris, Jr., was flagged for removal because he shared information with an Arkansan, “James Harris,” who had died in 1996.⁴⁵ According to one analysis, more than 68,000 of the 80,000 voters identified as possibly dead were weak matches.⁴⁶ This policy of flagging voters based on a weak match without further investigation was eventually changed when Texas settled litigation that had arisen on account of the bad purge.⁴⁷

States south of the Mason-Dixon Line do not have a monopoly on bad purges. Before the April 2016 primary election, the New York City Board of Elections purged more than 200,000 voters, the majority of whom lived in Brooklyn. In 2014 and 2015, the Brooklyn Borough Office of the Board of Elections targeted for removal people who had not voted since the 2008 election.⁴⁸ New York City officials complied with the portion of federal law requiring them to send notice to affected voters but not with the part that required them to wait two federal elections before purging those who did not respond. Instead, the Board of Elections gave voters 14 days to respond, then purged voters immediately.

In the end, nearly 118,000 registrations were canceled when voters did not respond to these notices.⁴⁹ And through another process, an additional 100,000 voters were removed (also without the required waiting period) because New York City Board of Elections officials believed they had moved.⁵⁰ On Election Day, thousands of voters showed up at the polls only to learn their registrations had been erased. Moreover, these problems were not evenly distributed. One report found that 14 percent of voters in Hispanic-majority election districts were purged compared to 9 percent of voters in other districts.⁵¹

Federal Role in Voter Protection Diminished

The increased purge rates are a cause for concern because there are fewer federal protections against improper purges. The Shelby County decision has halted the preclearance provision, which had previously blocked election changes in certain jurisdictions unless it could be shown that the change would not make minority voters worse off and was not enacted with discriminatory intent.

And at least for now, voters have lost another important protector against improper purges: the Justice Department. Since 1993, the Justice Department has been charged with enforcing the National Voter Registration Act, the primary source of federal protection against inaccurate or overly broad purges. While the Justice Department’s purge history is mixed,⁵² it brought pro-voter NVRA lawsuits during the Obama administration. Enforcement actions for violating the NVRA were undertaken against at least six states. In Florida and New York, the DOJ successfully challenged state purge practices.⁵⁴ In Florida, the Justice Department joined civic groups who successfully challenged the state’s practice of conducting systematic purges just 90 days before an election.⁵⁵

But the Trump administration has reversed course. For instance, in Husted v. A. Philip Randolph Institute, the Obama administration filed a brief in support of plaintiffs challenging an Ohio purging practice in which individuals who failed to vote in a single election received purge notices and were ultimately purged if they did not respond and did not vote in the next two federal elections. Failure to vote in a single election is poor evidence of ineligibility because not voting is common; for example, in the last midterm election, nearly 60 percent of Ohioans did not vote.⁵⁶ But when the case was pending before the U.S. Supreme Court in the summer of 2017, the Justice Department switched sides and supported Ohio.⁵⁷ On June 11, 2018, the Supreme Court ruled in favor of Ohio and the Justice Department’s new position.⁵⁸

Last summer, the Trump Justice Department also sent letters to 44 states demanding information about their voter purge practices.⁵⁹ Although the Justice Department has not taken further action so far, the suspicion is that the inquiries could be a precursor to enforcement actions to force states to purge more aggressively.⁶⁰

New Flaws in Voter Purges

Three new risks have emerged in voter purges in recent years. One is the growth of interstate databases that purport to identify voters who have moved to a new state and are registered in both their current and former state. The two databases primarily used are the Interstate Voter Registration Crosscheck program (Crosscheck) and Electronic Registration Information Center (ERIC).

Launched in 2005 by the Kansas secretary of state, Crosscheck purports to identify voters who may have cast ballots in two different states in the same election. In 2017,28 states participated in Crosscheck by sharing voter data with the system, but not all of those states actively used, or use, Crosscheck to remove voters. The number of participating states in 2018 is still to be determined because a number of states are assessing their participation.⁶¹

Another data-matching initiative, ERIC, began with assistance from the Pew Charitable Trusts in 2012. Twenty-four states and the District of Columbia are or will soon be members of ERIC.⁶²

The second risky development is the increasing number of states scouring their rolls to identify alleged noncitizens registered to vote: The number of states with statutes specifically mandating searching for and removing noncitizens from the rolls has increased from two to six since 2008. Of course, noncitizens are not permitted to vote in federal and state elections, but the sources states rely upon to determine voter citizenship, such as driver’s license lists, are not highly accurate. Moreover, the primary policy justification for aggressive purges aimed at removing noncitizens from the rolls — supposed widespread noncitizen voting — is not supported by the facts, a Brennan Center study of the 2016 election found. The study looked at 42 jurisdictions in 12 states, including eight of the 10 jurisdictions with the nation’s largest noncitizen populations. Out of the 23.5 million votes cast in these jurisdictions, election officials referred only 30 instances of suspected noncitizen voting, or .0001 percent of the total.⁶³

Finally, several conservative activist groups have sued state and local jurisdictions in recent years seeking to force them to purge their rolls more aggressively. For instance, last September the Public Interest Legal Foundation noted that it had brought nine suits in six states in the past two years alleging lax vigilance of voter rolls. That tally was included in a press release announcing that the group had put 248 counties in 24 states “on notice” that they were risking litigation if they could not demonstrate “effective voter roll maintenance.”⁶⁴

Interstate Voter Registration Crosscheck Program (Crosscheck)

Purges based on a change of address have long been complicated and error prone. When the Brennan Center looked at purges a decade ago, it found that states primarily used the National Change of Address database compiled by the U.S. Postal Service to identify movers (as well as driver’s license information).⁶⁵ But states have begun using other databases that go beyond the traditional sources of change-of-address information. Our research shows these new interstate databases have serious weaknesses that can lead to widespread and inaccurate purges.

When it began in 2005, the Kansas-based Crosscheck program had only four members.⁶⁶ In 2017, the most recent year data was shared, 28 states submitted data to the program.⁶⁷ Crosscheck’s purpose is to identify possible “double voters” — an imprecise term that could be used to refer to people who have registrations in two states or who actually voted in an election in multiple states. While it is not uncommon for those who have recently moved to be registered in multiple places, actual double voting is rare. In 2017, Crosscheck examined the records of 98 million voters⁶⁸ and produced 7.2 million “matches” representing 3.6 million voters supposedly registered in two states.⁶⁹

CROSSCHECK IN THE CROSSHAIRS

Crosscheck’s flaws put approximately 100 million voters in its database at potential risk, but some individuals are more vulnerable than others. Because of the loose matching criteria used by the program, parents and children with the same name are at greater risk of being confused with each other. Voters with common names are also more likely to match with different individuals for obvious reasons, but a less-obvious concern is the disproportionate effect this has on minority voters. African-American, Asian-American, and Latino voters are much more likely than Caucasians to have one of the most common 100 last names in the United States.¹

Crosscheck creates matches based on first name, last name, and birthdate. Shared names and birthdates are fairly common. In fact, if you were to gather 23 or more people in the same place, there is a greater than 50 percent chance that two people would share a birthday (day and month).² Even adding in the year doesn’t make an enormous difference: In a group of 180 people, it’s more likely than not that two people will have been born on the exact same day.³

Of course, adding in first and last names substantially decreases the rate at which people look the same on paper. It doesn’t, however, lower that rate sufficiently to make Crosscheck anywhere near accurate. When looking at records of millions of people, matching birthdates and names can still return thousands of inaccurate matches. This is true not only because of the so-called birthday problem but also because of the variation in the popularity of names. Jennifer, for instance, was the most common name for women born in the 1970s⁴ but was the 191st most common name for women born between 2010 and 2017.⁵ On average, 160 Jennifers were born every single day in the U.S. between 1970 and 1979. Among these, there were doubtless many who shared surnames common among Americans.

The program also hurts frequent movers such as college students and military personnel, who are more likely to be wrongly flagged by the database following a recent move. Because Crosscheck’s date of registration data is unreliable, those who move more frequently are more likely to be wrongly identified as having moved out of the state that purges them.⁶

1	Non-white people are more likely to have common shared names. For instance, 16.3 percent of Hispanic people and 13 percent of black people have one of the 10 most common surnames, compared to 4.5 percent of white people. Joshua Comenetz, “Frequently Occurring Surnames in the 2010 Census,” U.S. Census Bureau, October 2016, available at https://www2.census.gov/topics/genealogy/2010surnames/surnames.pdf.
2	Michael P. McDonald and Justin Levitt, “Seeing Double Voting: An Extension of the Birthday Problem,” Election Law Journal 7, (2007): 111–122, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=997888
3	Sharad Goel et al., “One Person, One Vote: Estimating the Prevalence of Double Voting in U.S. Presidential Elections” (working paper, Stanford University, 2017) 3, https://scholar.harvard.edu/files/morse/files/1p1v.pdf.
4	“Top names of the 1970s,” Social Security Administration, accessed June 15, 2018. https://www.ssa.gov/oact/babynames/decades/names1970s.html.
5	5 “Top names of the period 2010 - 2017,” Social Security Administration, accessed June 15, 2018. https://www.ssa.gov/oact/babynames/decades/names2010s.html.
6	Sharad Goel et al., “One Person, One Vote: Estimating the Prevalence of Double Voting in U.S. Presidential Elections” (working paper, Stanford University, 2017) appendix-22, https://scholar.harvard.edu/files/morse/files/1p1v.pdf

Crosscheck compares the voter registration list of each participating state against the voter registration lists of the other participating states and flags all records that have the same first name, last name, and date of birth.⁷⁰ But in groups as large as statewide (or multistate) voter registration lists, the statistical odds of two registrants having the same name and birth date is sufficiently high as to be problematic.⁷¹ A 2017 study led by Stanford professor Sharad Goel found that if applied nationwide, Crosscheck would “impede 300 legal votes for every double vote prevented.”⁷² Moreover, the study found that “there is almost no chance that double votes could affect the outcome of a national election.”⁷³ One of Crosscheck’s problems is that it does not have reliable registration dates, which means that an election official cannot competently determine which of the two places a voter is registered is more recent and therefore which state should remove the voter.

Virginia had a major problem with Crosscheck five years ago when it tried to purge nearly 39,000 voters. Crosscheck relies on little information before concluding that registration records in different states belong to the same person. Virginia sent counties the roster of voters for removal without checking its accuracy, and counties were not furnished with any guidance about the data or sufficient time to conduct a thorough review.⁷⁴ Eligible voters were wrongly flagged as having moved from Virginia to another state when they had in fact moved from another state to Virginia.⁷⁵ Error rates in some counties ran as high as 17 percent.⁷⁶ Counties did not begin spotting errors until some had begun removing voters. At the urging of civic groups, the state issued new guidance on the use of Crosscheck data but not until thousands of voters had been purged right before a statewide election.⁷⁷

Especially troubling is that at least four states have policies or regulations on the books providing for the use of Crosscheck in an illegal manner. Alabama,⁷⁸ Indiana,⁷⁹ and Maine⁸⁰ regulations allow counties to use Crosscheck to immediately purge voters from the rolls, without providing these voters notice and a two-election waiting period before deleting them as required by the NVRA.⁸¹ And Arizona regulations permit removing voters based on Crosscheck in some instances within 90 days of a federal election,⁸² which is not allowed under the NVRA for systematic purges such as those using Crosscheck.

Not all participating states are actively using Crosscheck data to identify and remove potentially ineligible voters.

In recent years, at least eight states have left the program altogether and no longer share data with or receive data from Crosscheck.⁸³ Additionally, seven other states have curtailed their use of Crosscheck data by not using it for the purposes of voter-list maintenance.⁸⁴

Instead, these states either do nothing with the data they receive or use it solely to identify people who appear to have voted (not merely registered) in multiple states.

In the midst of publicity around lax security protocols with Crosscheck⁸⁵ and news earlier this year that Crosscheck would review its security protocols and postpone uploading data,⁸⁶ Illinois announced that it would no longer transmit data to Crosscheck.⁸⁷ A state official was quoted as saying, “we will transmit no data to Crosscheck until security issues are addressed to our satisfaction.”⁸⁸ A South Carolina official expressed a similar sentiment, explaining that the state stopped using data “due to issues with verification and concerns about cybersecurity.”⁸⁹ According to an attorney representing the state of Indiana in litigation related to the state’s use of Crosscheck, as of May 2 of this year, Crosscheck was not accepting data from participating states while a review of security processes remained in progress.⁹⁰

Electronic Registration Information Center (ERIC)

The Electronic Registration Information Center is a program that uses voter registration data, motor vehicle licensing information, Social Security Administration data, and National Change of Address information to identify voters who may have moved. Begun six years ago, 24 states plus the District of Columbia are enrolled in the program (or soon will be).⁹¹ To participate in ERIC, states must submit extensive voter data, including full address, driver’s license or state ID number, last four digits of social security number, date of birth, voter registration activity dates, current record status, eligibility documentation, phone number, and email address.⁹² Election officials in ERIC-participating states told us they provide notice and a two-election waiting period before removing voters.⁹³

Election officials reported that ERIC also helps them identify potential voters who have moved into their jurisdictions but have not registered.⁹⁴ And one analysis of ERIC’s first year of operation showed increases in registrations in ERIC states relative to non-ERIC states.⁹⁵

Although most of the election administrators that we interviewed reported positive experiences with ERIC, the new data source has its limits. Administrators from Maryland and Illinois, for example, reported that it could be difficult to determine a voter’s most recent address, which is a problem for frequent movers.⁹⁶ This absence of precise information means that, even though ERIC is generally processed at the state level, it is local officials who must identify errors and determine which registration is more current — the one in the relevant jurisdiction or a registration in another state.⁹⁷ Wisconsin, meanwhile, reported that although ERIC was helpful in updating more than 25,000 registration addresses in 2017 and 2018, it also resulted in more than 1,300 voters signing “supplemental poll lists” at a spring 2018 election, indicating that they had not in fact moved and were wrongly flagged.⁹⁸

Efforts to Purge Noncitizens Are More Frequent and Often Rely on Flawed Data

The Brennan Center’s 2008 study found that attempts to purge noncitizens were rare. Back then only two states, Texas and Virginia, had laws mandating specific procedures for identifying noncitizens.⁹⁹ In the last decade, four more states — Georgia, Iowa, Minnesota, and Tennessee — have passed laws requiring removal of noncitizens.¹⁰⁰ More states are likely to pass such laws because of pressure to aggressively search for and delete noncitizen registrations.

As is true with other purges, the information relied upon to purge alleged noncitizens can be inaccurate. For example, at least 14 states have sought access to the federal Systematic Alien Verification for Entitlements (SAVE) program,¹⁰¹ which checks several databases to ascertain the residence or citizenship status of people who have contacted benefit-granting agencies.¹⁰² Some states, such as Virginia, were granted access. However, states found the database is useful only if an election administrator has someone’s alien identification number, information election officials typically do not possess.¹⁰³

Some states use driver’s license data to purge noncitizens. Minnesota, Tennessee, and Virginia have statutes mandating this approach. Generally, driver’s license data is deployed in one of two ways.¹⁰⁴ One involves review of documents the registrant provided to the driver’s license office when obtaining a license. If a person showed a Permanent Resident Card, the presumption is that the registrant is a noncitizen and should be removed from the rolls. The problem, however, is that a person can lawfully not update their driver’s license information for many years, in which time they may have become a citizen.¹⁰⁵

States may also scour their voter lists for those who did not check the box indicating that they were a citizen on their driver’s license application or renewal. Virginia has a specific statutory provision requiring this; Maryland does not but still engages in the practice.¹⁰⁶ Not surprisingly, election officials told us that sometimes citizens fail to check the citizenship box.¹⁰⁷

In addition, at least three states (Georgia, Louisiana, and Texas) remove voters if they decline jury service on the grounds of noncitizenship.¹⁰⁸ But election officials told the Brennan Center in a 2017 report on noncitizen voting that eligible voters have been known to assert they are noncitizens solely for the purpose of evading jury duty. While illegal, these declarations are not necessarily indicative that a noncitizen has been registered to vote.¹⁰⁹

Activist Groups Pressing for More Aggressive Purges

Another new dynamic is activist groups agitating for election officials to purge the rolls more aggressively. In the past, litigation was often used by groups seeking to protect voters against bad voter purges. For example, civic groups prevented voters from being illegally purged in Michigan in 2008,¹¹⁰ Colorado in 2010,¹¹¹ and Florida in 2012.¹¹²

From 1998 through 2007, most of the litigation seeking purges was brought by the Justice Department — which made voter purges a priority in the midst of a failed nationwide voter fraud hunt¹¹³ — whereas private plaintiffs typically brought suits because they were worried eligible people would be improperly purged. From 2008 to the present however, more than half of the 32 federal purge-related lawsuits brought by private parties have been filed by plaintiffs who believed that jurisdictions are not purging enough names from the rolls.¹¹⁴

In nine cases brought by private parties since 2012, election officials agreed to undertake more aggressive list maintenance.¹¹⁵ One of the defendants in these cases was Noxubee County, a poor, rural, majority-Black county in eastern Mississippi that was sued by the American Civil Rights Union (ACRU, not to be confused with the American Civil Liberties Union).

“They went after minority counties who didn’t have the financial resources to push back,” said Willie M. Miller, the Election Commissioner for Noxubee County’s fourth district.¹¹⁶ As of this writing, the ACRU is suing Starr County and the State of Texas¹¹⁷ for failing to purge aggressively enough, and the like-minded Judicial Watch has brought litigation in California.¹¹⁸

Unfortunately, this litigation has consequences. The ACRU lawsuit against Noxubee County resulted in about 1,500 (more than 12 percent) of its 9,000 voters being made inactive.¹¹⁹ Being designated as inactive is the first stage of the removal process. The waiting period of two federal elections has yet to expire, so it’s unclear at this juncture how many voters will ultimately be removed.¹²⁰ Similarly, Judicial Watch’s 2012 suit against Indiana¹²¹ arguably led to the state undertaking more aggressive list maintenance. Before the suit was dismissed, Indiana announced that it had sent an “address confirmation mailing to all voters” and undertook other purging initiatives that led to more than 480,000 canceled registrations after the 2016 election.¹²² Judicial Watch boasted that their lawsuit “forced” Indiana to undertake additional purge practices;¹²³ Indiana first sent out the required federal notices in 2014, then purged voters who did not respond and did not vote in 2014 or 2016.

Litigation is but one element of a broader strategy by these groups to force purges. In 2016, the Public Interest Legal Foundation published a report entitled “Alien Invasion in Virginia,” complete with a flying saucer on the cover. Extrapolating from a small sample, the missive misleadingly suggested thousands of votes had been cast by noncitizens,¹²⁴ a claim election officials dispute.¹²⁵ The Foundation’s pressure may have had an impact: Six hundred ninety-three alleged noncitizens were purged in the 2016 reporting period, but that number more than doubled to 1,686 in the 2017 period.¹²⁶ The purge has spawned yet more litigation, with several voters complaining that they were wrongly deleted, and the Public Interest Legal Foundation has been sued for defamation and illegal voter intimidation.¹²⁷ Election fraud vigilantes have also brought mass challenges to voters’ registrations, including in North Carolina, where a judge blocked the practice.¹²⁸

CHALLENGES CONTINUE

In at least 15 states, “challenge” laws permit challenges to the validity of a voter’s registration prior to Election Day (additional states allow challenges to eligibility at the time of voting only).¹ These challenge laws, which are designed to allow for questioning the eligibility of registered voters on a case-by-case basis, have been used recently in several states to try to systematically remove voters from the rolls, functioning effectively as a purge that can operate outside the NVRA’s protections. The use of challenge laws as back doors for purging is legally dubious and increases the risk of wrongful removals; precisely what has happened in some states.

Colorado’s former secretary of state, Scott Gessler, matched the voter rolls against driver’s license lists to produce a large (and inflated) list of potential noncitizens. He then attempted to use his state’s challenger laws to remove voters en masse. After much public criticism, Gessler abandoned the effort.²

In Hancock County, Georgia, the majority-white Board of Elections used challenge procedures in the weeks leading up to a 2015 municipal election to challenge 174 voters — nearly 20 percent of the town of Sparta’s electorate. The majority of the challenged voters were Black. Some of the challenges were based on as little evidence as a discrepancy between a voter registration address and an address record in a flawed driver’s license database. Other challenges were based on second-hand claims that a voter had moved out of the county.³ After being sued, the county agreed to reinstate wrongfully challenged voters who had been removed from registration lists.⁴

Iowa’s former secretary of state, Matt Schultz, tried to use challenges to remove suspected noncitizens from the rolls, but he was blocked by a court.⁵

And in North Carolina, a federal court ruled in 2016 that local boards of elections likely violated the NVRA (52 U.S.C. § 20507(c) (2)(A)) when they systematically purged hundreds of voters through citizen-initiated challenge procedures fewer than 90 days before the general election. The judge based her ruling on the systematic purge occurring within the prohibited window, but she also remarked that the challenge process, which allows voters to be removed if they do not show up at a hearing upon being challenged based on second-hand evidence of a move, seemed “insane.”⁶ Nevertheless, state lawmakers expressly rejected legislation that would have made it more difficult to sustain a voter challenge on this basis.⁷

1	Nicholas Riley, Voter Challengers (New York: Brennan Center for Justice, August 2012), https://www.brennancenter.org/sites/default/files/legacy/publications/Voter_Challengers.pdf
2	“Scott Gessler Decides Not To Proceed With Voter Purge After All,” HuffPost, September 12, 2012, https://www.huffingtonpost.com/2012/09/10/scott-gessler-decides-not_n_1871524.html.
3	Complaint, Georgia NAACP et al v. Hancock County Bd. Of Elec. and Registration, No. 5:15-cv-00414 (M.D. Ga. Filed Nov. 3, 2015), https://lawyerscommittee.org/wp-content/
4	Kathleen Foody, “Georgia County Agrees to Restore Black Voters’ Rights,” Associated Press, March 8, 2017, https://www.usnews.com/news/best-states/georgia/articles/2017-03-08/georgia-county-to-restore-black-voters-rights-under-us-law.uploads/2016/01/Hancock-Co-Complaint.pdf.
5	Ruling, Am. Civ. Liberties Union v. Schultz, No. CV00931 (Iowa D. Polk March 5, 2014)
6	“North Carolina Voter Challenge Process Seems ‘Insane,’ Judge Says,” Associated Press, November 2, 2016, https://www.cbsnews.com/news/north-carolina-voter-challenge-process-seems-insane-judge.
7	H. 303, Sess. 2017 (N.C. 2017), https://www2.ncleg.net/BillLookup/2017/H303



Solutions
While no one disputes the rolls should be accurate, voters should be protected from wrongful purges. There are several ways to safeguard voters from overly aggressive list maintenance:
●	Enforce the National Voting Registration Act’s Protections. The NVRA permits an aggrieved voter to sue if a jurisdiction has been informed of a possible violation and does not correct it in a set period of time. Litigation to enforce the NVRA is especially crucial in a time when the Justice Department is unlikely to enforce voter protections and outside groups are agitating for more aggressive purges. Of course, most voters do not have the expertise or resources to bring such litigation. Therefore it is critically important that civil rights and other pro-voter organizations rigorously monitor purge activity and have the wherewithal to sue when necessary.
●	States Should Enact Laws That Provide Even More Protections Than the National Voter Registration Act. While the NVRA includes critical voter protections, states should do more. For example, the NVRA requires that voters suspected of moving from the jurisdiction receive notice of their possible removal. Not surprisingly, most states do not provide notice beyond what is federally required. For example, most states do not provide notice to voters purged based on death or a disenfranchising conviction, and many of those states that do provide notice in these circumstances do so only after the fact. States should surpass these minimal standards. No matter the reason, all voters should be informed in advance of their possible deletion and should be provided easy mechanisms for correcting errors on or before Election Day.
●	Enact Automatic Voter Registration. Automatic voter registration is a popular reform that minimizes errors, saves money, and increases registration of eligible citizens. Automatic voter registration has two key features: (1) eligible citizens are registered unless they affirmatively decline; and (2) voter registration information is electronically transferred from a government office to election officials instead of relying on pen and paper. Currently, 12 states plus the District of Columbia have approved automatic voter registration.129 In addition to adding more voters to the rolls, automatic voter registration also catches more address updates, reducing the need for change-of-address voter purges.

Endnotes

1	In the two-year election cycle ending in 2008, the Brennan Center found the median jurisdiction purged 6.2 percent of voters. For the two years ending in 2016, this study finds that the purge rate of the median jurisdiction had increased to 7.8 percent. We examined 49 states because North Dakota has no advance voter registration requirement and thus does not have required voter registration lists to purge. The state does keep records of individuals who vote, but it is not necessary to be on any registration list at the time of voting to cast ballots. Although there are other impediments to voting in North Dakota, including a strict photo ID law, voters do not face barriers related to voter registration in the state.
2	We assessed 49 states on the following criteria: First, whether the state used the Interstate Voter Registration Crosscheck program in a way that is problematic or not compliant with the NVRA. We found five states deficient in this category. Second, whether the state makes readily available lists of purged voters. We found 49 states deficient in this category (at least 10 states have statutory requirements for making some names of purged voters available, but all fail to do so in practice). Third, whether states provide prior notice to all voters purged on the basis of death, felony conviction, or noncitizenship. We found 49 states deficient in this category (21 states have statutory requirements whereby voters purged on the basis of death or felony conviction receive notice before or after the purge, but no state requires prior notice to voters purged for both categories). For additional recommendations to guard against unlawful or problematic voter purges and why they are important, see Myrna Pérez, Voter Purges (New York: Brennan Center for Justice, September 2008), 25-31, https://www.brennancenter.org/sites/default/files/legacy/publications/Voter.Purges.f.pdf.
3	Calculated from total numbers reported to the Elections Assistance Commission in 2008 and 2016. Compare U.S. Election Assistance Commission, 2008 Election Administration and Voting Survey, https://www.eac.gov/research-and-data/2008-election-administration-voting-survey/ , and U.S. Election Assistance Commission, 2016 Election Administration and Voting Survey, https://www.eac.gov/research-and-data/2016-election-administration-voting-survey/ .
4	These previously covered areas had median purge rates of 9.5 percent, while noncovered jurisdictions had median purge rates of 7.5 percent.
5	The median county purge rate in the 2008-10 election cycle was 8.4 percent. But in the election cycle including the Shelby County decision, 2012-14, the purge rate jumped 26 percent to a median county purge rate of 10.6 percent.
6	Myrna Pérez, Voter Purges (New York: Brennan Center for Justice, September 2008), https://www.brennancenter.org/sites/default/files/legacy/publications/Voter.Purges.f.pdf.
7	Omitting North Dakota, as explained above.
8	We served public records requests on election officials and their offices at the state and local levels in 22 states and sought interviews with election officials in 45. The numbers referenced in the text refer to respondents.
9	U.S. Election Assistance Commission, 2016 Election Administration & Voting Survey, June 2017, https://www.eac.gov/research-and-data/election-administration-voting-survey/.
10	Not all jurisdictions report their data consistently. Whenever we make comparisons across time periods, we restrict our sample to the counties reporting consistently. For instance, 2,394 jurisdictions report removal data for each of the two-year periods ending in 2010, 2012, 2014, and 2016. Our analysis exploring the impact of the end of the preclearance condition of the Voting Rights Act looks only at these counties to ensure an apples-to-apples comparison.
11	U.S. Election Assistance Commission, 2016 Election Administration & Voting Survey, June 2017, https://www.eac.gov/research-and-data/election-administration-voting-survey/ . Sixteen million is in fact a conservative estimate because it includes only voters removed from jurisdictions who reported their data to the EAC in 2016. It therefore does not include voters removed during some problematic purges such as that in Kings County (Brooklyn), NY (discussed above).
12	National Voter Registration Act of 1993, H.R. 2, 103rd Cong. (1993), 52 U.S.C. § 20507, is the main source of federal requirements. For more information on federal law around purges, see Appendix A.

13	Some states are not required to follow the National Voter Registration Act. The NVRA exempts the following states from its purge protocols because those states had Election-Day registration or lacked voter-registration requirements on or after August 1, 1994: Idaho, Minnesota, New Hampshire, North Dakota, Wisconsin, and Wyoming. National Voter Registration Act of 1993, H.R. 2, 103rd Cong. (1993) 52 U.S.C. § 20504(b). This reflects Congress’s assessment that purge consequences are much less grave in a state that permits anyone eligible who is not on the registration rolls to register and vote on Election Day.
14	“Criminal Disenfranchisement Laws Across the United States,” Brennan Center for Justice, last modified April 18, 2018, https://www.brennancenter.org/criminal-disenfranchisement-laws-across-united-states.
15	Ala. Code § 17-4-3(a) (requiring removal “whenever…a person registered to vote in that county has…been declared mentally incompetent”); Ariz. Rev. Stat. Ann. § 16-165(C) (requiring removal “[w]hen proceedings…result in a person being declared incapable of taking care of himself and managing his property, and for whom a guardian of the person and estate is appointed, result in such person being committed as an insane person”); Del. Code Ann. tit. 15, §§ 1701(a), 1702 (requiring removal of “person adjudged mentally incompetent…[which] refers to a specific finding in a judicial guardianship or equivalent proceeding, based on clear and convincing evidence that the individual has a severe cognitive impairment which precludes exercise of basic voting judgment”); Fla. Stat. Ann. § 98.075(4) (requiring removal for “registered voters who have been adjudicated mentally incapacitated with respect to voting and who have not had their voting rights restored”); Ga. Code Ann. § 21-2-231(b) (requiring removal “[of those] who were declared mentally incompetent during the preceding calendar month in the county and whose voting rights were removed”); Haw. Rev. Stat. Ann. § 11-23(a) (requiring removal “[of person] adjudicate[ed] as an incapacitated person under the provisions of chapter 560…[if ] after the investigation the clerk finds that the person…lacks sufficient understanding or capacity to make or communicate responsible decisions concerning voting”); Iowa Code Ann. § 48A.30(1)(e) (requiring removal “[if] [t]he clerk of the district court or the state registrar sends notice that the registered voter has been declared a person who is incompetent to vote under state law”); Ky. Rev. Stat. Ann. § 116.113(2) (requiring removal “[u]pon receipt of notification from the circuit clerk that a person has been declared incompetent”); La. Stat. Ann. § 18:172 (requiring removal “[after] judgment of full interdiction or a limited interdiction for mental incompetence which specifically suspends the right to register and vote and which has become definitive”); Code Me. R. tit. 29-250 Ch. 505, § 1(B) (requiring removal “[if ] the municipality receives notice indicating that a registrant has been placed under guardianship due to mental illness”); Md. Code Ann., Elec. Law §§ 3-102(b)(2), 3-501 (requiring removal “[if person] is under guardianship for mental disability and a court of competent jurisdiction has specifically found by clear and convincing evidence that the individual cannot communicate, with or without accommodations, a desire to participate in the voting process”); Minn. Stat. Ann. § 201.145 (requiring removal “[of persons] under a guardianship in which a court order revokes the ward’s right to vote or where the court has found the individual to be legally incompetent to vote”); Miss. Code. Ann. § 23-15153(1) (requiring removal “[of voters who have] received an adjudication of non compos mentis”); Mo. Ann. Stat. § 115.199 (requiring removal “of voters…adjudged incapacitated”); Mont. Code Ann. § 13-2-402(3) (requiring removal “[if] the elector is of unsound mind as established by a court”); Neb. Rev. Stat. Ann. §§ 32-313(1), 32326 (requiring removal “[of person] who is non compos mentis”); Nev. Rev. Stat. Ann. § 293.540(2)(b) (requiring removal “[if ] the county clerk is provided a certified copy of a court order stating that the court specifically finds by clear and convincing evidence that the person lacks the mental capacity to vote because he or she cannot communicate, with or without accommodations, a specific desire to participate in the voting process”); N.M. Stat. Ann. § 1-4-26 (requiring removal “[w]hen in proceedings held pursuant to law, the district court determines that a mentally ill individual is insane as that term is used in the constitution of New Mexico”); N.Y. Elec. Law § 5-400(1) (c) (requiring removal “[of voter who] has been adjudicated an incompetent”); Ohio Rev. Code Ann. § 3503.18(B) (requiring removal of persons “who have been adjudicated incompetent for the purpose of voting, as provided in section 5122.301 of the Revised Code”); Okla. Stat. Ann. tit. 26, § 4-120.5 (requiring removal “of all persons who have been adjudged incapacitated”); S.C. Code Ann. § 7-5-340(1)(b) (requiring removal “if the elector is adjudicated mentally incompetent by a court of competent jurisdiction”); S.D. Codified Laws § 12-4-18 (requiring removal “of persons declared mentally incompetent”); Tex. Elec. Code Ann. § 16.031(a)(3) (requiring removal “on receipt of…an abstract of a final judgment of the voter’s total mental incapacity, partial mental incapacity without the right to vote…or disqualification under Section 16.002”); Wash. Rev. Code Ann. § 29A.08.515 (requiring removal “[u]pon receiving official notice that a court has imposed a guardianship for an incapacitated person and has determined that the person is incompetent for the purpose of rationally exercising the right to vote, under chapter 11.88 RCW”); W.Va. Code, § 3-2-23(3) (requiring removal “[u]pon receipt of a notice from the appropriate court of competent jurisdiction of a determination of a voter’s mental incompetence”); Wis. Stat. Ann. §§ 6.03, 6.48,6.935 (requiring removal “[through challenge] [of a]ny person who is incapable of understanding the objective of the elective process or who is under guardianship, unless the court has determined that the person is competent to exercise the right to vote”); W.S.1977 §§ 22-3-102(a)(iv), 22-3-115(a)(iv) (requiring removal “[of person] currently adjudicated mentally incompetent”). Additional states provide for loss of eligibility on these grounds but do not specifically describe the manner of removal. See Michelle Bishop, “Disability Is No Reason to Strip a Person’s Voting Rights,” HuffPost, May 12, 2018, https://www.huffingtonpost.com/entry/opinion-bishop-disability-voters_us_5af5b085e4b0e57cd9f9042f.

16	See Doe v. Rowe, 156 F. Supp.2d 35 (D. Me. 2001); Minnesota Voters Alliance v. Ritchie, 890 F.S. 2d 1106 (August 17, 2012); in re Guardianship of Brian W. Erickson, 4th Judicial District, Dist. Ct., Probate/Mental Health Division (October 12, 2012); see also Matt Vasilogambros, “Thousands Lose Right to Vote Under ‘Incompetence’ Laws,” HuffPost, March 21, 2018, https://www.huffingtonpost.com/entry/thousands-lose-right-to-vote-under-incompetence-laws_us_5ab25f7ce4b004fe24699810.
17	.g., Alaska Stat. Ann. § 15.07.130(c) (requiring use of information from bureau of vital statistics); Wash. Rev. Code Ann. § 29A.08.510(2) (permitting use of obituaries); Tex. Elec. Code Ann. § 16.001 (requiring use of Social Security Administration information).
18	Montana Code Ann. § 13-2-220.
19	Conn. Gen. Stat. § 9-32.
20	Ga. Code Ann. § § 21-1-231(a.1)(b) (requiring clerk of superior court to forward noncitizen jury declinations and requiring election officials to remove names from voter list, La. Stat. Ann. § 18:178 (requiring clerk of the court to provide names of individuals who respond to jury notices saying they are noncitizens to Department of State); Minn Stat. Ann. § 201.145 (requiring county auditor to send to county attorney list of names of individuals who are registered to vote and not citizens); Tenn. Code Ann. § 2-2-141 (requiring coordinator of elections to compare registration list with Department of Safety database to ensure non-United States citizens are not registered to vote); Tex. Elec. Code Ann. § 16.0332 (requiring registrar to initiate voter removal process for voters for whom the registrar receives a notice of disqualification or excusal from jury service because of citizenship status); Va. Code Ann. § 24.2-404(A)(4) (requiring registrars to delete record of registered voters known not to be a citizen from reports of Department of Motor Vehicles or Systematic Alien Verification for Entitlements Program).
21	Throughout this document we report median removal rates. The median is the appropriate measure of central tendency because of how the removal rate data are distributed. Because some jurisdictions have very high removal rates, while most are clustered close to the lower bound of zero, using the mean would artificially bias reported numbers upward.
22	“About Section 5 of the Voting Rights Act,” The United States Department of Justice, accessed May 24, 2018, https://www.justice.gov/crt/about-section-5-voting-rights-act.
23	Shelby County v. Holder, 570 U.S. 2 (2013).
24	Between the presidential elections of 2008 and 2012, the median two-year removal rate for both previously covered and noncovered jurisdictions was 7.5 percent. Throughout this section, we limit our analysis to jurisdictions that reported removal rates for each of the two-year periods ending 2010, 2012, 2014, and 2016. Kings County, New York, for instance, did not report removal rates for the two years ending 2016 and thus is excluded from the entire pre/post Shelby analysis. It is important to note that this does not meaningfully impact our analysis: The median removal rate in 2016 for counties that reported their data each year was 7.9 percent compared to 7.6 percent for jurisdictions that reported their data in 2016 but also failed to do so in at least one other year. To maintain consistency with discussions of two-year removal rates elsewhere in this report, we continue to use two-year removal rates here. For instance, Escambia County, Florida, removed 0.42 percent of its voters between 2008 and 2010, and 0.42 percent again between 2010 and 2012. Here we call their median two-year removal rate 0.42 percent. Their four-year removal rate would, of course, be higher. We group the data into four-year buckets because of the natural variation in removal rates between presidential and nonpresidential election cycles.

Formerly covered jurisdictions are disproportionately located in the southeastern part of the country. We considered the possibility that the increased purge rate is attributable to some regional factor or factors aside from the lifting of the preclearance requirements. To control for this, we repeated the above analysis but restricted our sample to just those states in the Southeast (AL, FL, GA, KY, MS, NC, SC, TN, VA, and WV). Among jurisdictions in the Southeast that consistently reported their data, 461 counties were covered under the Voting Rights Act and 388 were not. We found that even within the Southeast, formerly covered jurisdictions increased their purge rates more than their noncovered peers. In fact, noncovered jurisdictions in the Southeast did not increase their removal rates between the two periods. The increase in removal rates in previously covered jurisdictions in this region mirrored those of the group of covered jurisdictions as a whole:

	Federal Election 2008-12	Federal Election 2012-16
Previously Covered	7.2%	9.7%
Not Covered	6.6%	6.6%

Nor can the difference in purge rate be explained by differences in partisan tendency. Formerly covered counties are more Republican-leaning than the nation as a whole. Within counties that reported data consistently to the EAC, President Donald Trump received 51 percent of the ballots cast in counties that required preclearance prior to Shelby, but just 46 percent of the ballots cast in noncovered jurisdictions. To test the possibility that Republican-leaning counties were more likely to increase their removal rates regardless of their status under the Voting Rights Act, we compared the 409 previously covered jurisdictions that Trump received more votes than Hillary Clinton to the 1,594 noncovered jurisdictions in which he did so.

	Federal Election 2008-12	Federal Election 2012-16
Previously Covered	7.3%	9.4%
Not Covered	7.5%	7.4%

Removal rates in noncovered jurisdictions that Trump won did not increase their removal rates at all. Trump-supporting jurisdictions that were previously covered, however, increased their removal rates substantially. Clearly, the increase in removal rates among the jurisdictions that were covered under the VRA was not a function of an electorate likely to support Donald Trump. Sources: Townhall.com, https://townhall.com/election/2016/president; and SouthEastern Division of the Association of American Geographers, http://sedaag.org.

See Appendix B.

See Appendix C. While not a perfect predictor because there are many reasons why a voter might cast a provisional ballot, our finding that high provisional ballot numbers are probative as to the existence of a purge are corroborated by other experts in the field. See, for example, U.S. Commission on Civil Rights, Briefing Report: Department of Justice Voting Rights Enforcement for the 2008 U.S. Presidential Election (Washington: July 2009) (summarizing testimony of Dan Tokaji),
http://www.usccr.gov/pubs/DOJVotingRights2008PresidentialElection.pdf.

Tim Reid and Grant Smith, “Missing Hyphens Will Make It Hard for Some People to Vote in U.S. Election,” Reuters, April 11, 2018. https://www.reuters.com/article/us-usa-election-laws/missing-hyphens-will-make-it-hardfor-some-people-to-vote-in-u-s-election-idUSKBN1HI1PX.
Georgia’s practice of purging voters on the basis of not voting was also challenged. See Georgia State Conf. of the NAACP v. Kemp, No. 2:16-cv-219, filed Sept. 14, 2016 (N.D. Ga.); Common Cause v. Kemp, No. 1:16-cv-00452, filed Feb. 10, 2016 (N.D. Ga.). See also Tony Pugh, “Georgia Secretary of State Fighting Accusations of Disenfranchising Minority Voters,” McClatchy, October 7, 2016, http://www.mcclatchydc.com/news/politics-government/article106692837.html;
Regina Willis, “More Than 380,000 Georgia Voters Receive ‘Purge Notice,’” Rewire.News, July 21, 2017,
https://rewire.news/article/2017/07/21/more-380000-georgia-voters-received-purge-notice/.

Overall, 54% of voters lived in counties in which the removal rate increased. Numbers are drawn from counties that reported data in both 2010 and 2014, a set representing 94% of total Texas voters.

Veasey v. Perry, 71 F.Supp.3d 627 (S.D. Tex. 2014).

31	Veasey v. Abbott, 830 F.3d 216 (5th Cir. 2016) (en banc) cert. denied, 137 S. Ct. 612 (2017).
32	National Voter Registration Act of 1993, H.R. 2, 103rd Cong. (1993), 52 U.S.C. § § 20507(b), (c)(2), (d)(2).
33	Holly Dickson (Legal Director, Arkansas Civil Liberties Union Foundation) to Hon. Mark Martin (Arkansas Secretary of State), October 31, 2016, 3, https://www.acluarkansas.org/sites/default/files/field_documents/369.pdf; John Lyon, “Hutchinson: Clerks Should Lean Toward Letting People Vote,” Arkansas News, August 4, 2016, http://www.arkansasnews.com/news/20160804/hutchinson-clerks-should-lean-toward-letting-people-vote.
34	In Arkansas, those convicted of a felony are ineligible to vote “unless the person’s sentence has been discharged or the person has been pardoned.” Ark. Const. Amend. 51, § 9(a)(1).
35	More than 4,000 people were incorrectly included on the list. See John Lyon, “Hutchinson: Clerks Should Lean Toward Letting People Vote,” Arkansas News, August 4, 2016, http://www.arkansasnews.com/news/20160804/hutchinson-clerks-should-lean-toward-letting-people-vote. Pulaski County found that at least 300 of the 1,800 Pulaski County residents on the list belonged to people who were “completely innocent.” Matthew Mershon, “Pulaski Co. Clerk Says Sec. of State Needs to Take Responsibility in Possible Voter Purge,” KATV, August 13, 2016, http://katv.com/news/local/pulaski-co-clerk-says-sec-of-state-needs-to-take-responsibility-in-possible-voter-purge.
36	See Benjamin Hardy, “Data Mix-Up from Ark. Secretary of State Purges Unknown Number of Eligible Voters,” Arkansas Blog, Arkansas Times, July 25, 2016, https://www.arktimes.com/ArkansasBlog/archives/2016/07/25/datamix-up-from-ark-secretary-of-state-purges-unknown-number-of-eligible-voters; Brenda Blagg, “Taking a Vote: State Botches Inmate Report to County Clerks,” Between the Lines, Northwest Arkansas Democrat-Gazette, July 27, 2016.
37	See Benjamin Hardy, “Data Mix-Up from Ark. Secretary of State Purges Unknown Number of Eligible Voters,” Arkansas Blog, Arkansas Times, July 25, 2016, https://www.arktimes.com/ArkansasBlog/archives/2016/07/25/datamix-up-from-ark-secretary-of-state-purges-unknown-number-of-eligible-voters.
38	See Brian Fanney, “20,000 Cases Erroneously Listed Felonies,” Arkansas Democrat-Gazette, Sep. 3, 2016, https://www.pressreader.com/usa/arkansas-democrat-gazette/20160903/281496455722563.
39	Jason Kennedy (Assistant Chief Deputy Clerk, Pulaski County, Arkansas), interview by Brennan Center for Justice, June 8, 2018.
40	See Julián Aguilar, “Voter Purge Bill Raises Concerns After Living Flagged as Possibly Dead,” The Texas Tribune, September 12, 2012, https://www.texastribune.org/2012/09/12/concerns-raised-after-living-voters-flagged-dead/.
41	2011 Tex. Sess. Law Serv. Ch. 683 (H.B. 174), https://capitol.texas.gov/tlodocs/82R/billtext/pdf/HB00174F.pdf#navpanes=0.
42	See Myrna Pérez, Voter Purges (New York: Brennan Center for Justice, September 2008), 20 https://www.brennancenter.org/sites/default/files/legacy/publications/Voter.Purges.f.pdf.
43	See Defendant Andrade’s Notice Of Withdrawal, Plea To The Jurisdiction, And Motion To Dissolve The Temporary Restraining Order, Moore v. Morton, No. D-1-GN-12-002923 (Dist. Ct. Travis Cnty. Tex. Sept. 21, 2012). See also Chuck Lindell, “State Settles Lawsuit on ‘Dead’ Voter Purge,” American-Statesman, October 3, 2012, https://www.statesman.com/news/state--regional-govt--politics/state-settles-lawsuit-dead-voter-purge/n1zTG10Yiyobma3AlT7QSJ/.
44	Corrie MacLaggan, “Texas Voter Purge Lawsuit Ends with Clarification Memo on Process for Clearing Rolls,” Reuters, October 3, 2012, https://www.huffingtonpost.com/2012/10/03/texas-voter-purge-lawsuit_n_1937564.html.
45	Lise Olsen, “Texas’ voter purge made repeated errors,” Houston Chronicle, November 2, 2012, https://www.chron.com/news/politics/article/Texas-voter-purge-made-repeated-errors-4001767.php.
46	Ibid.
47	See Notice to the Court of Rule 11 Agreement, Moore v. Morton, No. D-1-GN-12-002923 (Dist. Ct. Travis Cnty. Tex. Oct. 3, 2012); see also Chuck Lindell, “State Settles Lawsuit on ‘Dead’ Voter Purge,” American-Statesman, October 3, 2012, https://www.statesman.com/news/state--regional-govt--politics/state-settles-lawsuit-dead-voterpurge/n1zTG10Yiyobma3AlT7QSJ/.

48	See Marjorie Landa, Audit Report on the Board of Elections’ Controls over the Maintenance of Voters’ Records and Poll Access (New York: City of New York Office of the Comptroller, November 2017), 9, https://comptroller.nyc.gov/reports/audit-report-on-the-board-of-elections-controls-over-the-maintenance-of-voters-records-and-poll-access/.
49	Ibid.
50	New York State Office of the Attorney General, “A.G. Schneiderman Moves to Intervene in Lawsuit Against NYC Board of Elections Regarding Voter Registration Purges,” news release, January 27, 2017, https://ag.ny.gov/press-release/ag-schneiderman-moves-intervene-lawsuit-against-nyc-board-elections-regarding-voter.
51	Brigid Bergin, John Keefe, and Jenny Ye, “Brooklyn Voter Purge Hit Hispanics Hardest,” WNYC, June 21, 2016, https://www.wnyc.org/story/brooklyn-voter-purge-hit-hispanics-hardest/.
52	The other major federal statute regulating voter purges is the Help America Vote Act of 2002 (HAVA) 52 U.S.C. § 21083(a). The law reaffirms requirements of the NVRA and contains additional regulations for the maintenance of voter lists, requires states to set up unique identifying numbers for registered voters, requires states to attempt to verify the validity of information submitted by voter registration applicants, and ensures certain voters, including those missing from the voter rolls, can cast provisional ballots.
53	Under the Bush Administration, the DOJ filed several suits against jurisdictions for failing to purge enough voters. Office of the Inspector General, U.S. Department of Justice, A Review of the Operations of the Voting Section of the Civil Rights Division (2013), 97, https://oig.justice.gov/reports/2013/s1303.pdf. See also Steven Rosenfeld, “Voter Purging: A Legal Way for Republicans to Swing Elections?” AlterNet, September 11, 2017, http://web.archive.org/web/20071118161151/http:/www.alternet.org/story/62133/. The Department also pressured a U.S. Attorney to sue Missouri, even though St. Louis had improperly purged 50,000 voters only 4 years earlier. After that U.S. Attorney refused, he was fired. Jonathan Brater, Brennan Center for Justice, “The Purge: Ten Years Later?” June 30, 2017, https://www.brennancenter.org/blog/purge-ten-years-later
54	See, e.g., “Cases Raising Claims Under the National Voter Registration Act,” U.S. Department of Justice, last modified October 16, 2015, https://www.justice.gov/crt/cases-raising-claims-under-national-voter-registration-act; U.S. Attorney’s Office, Eastern District of New York, “United States Announces Settlement with New York City Board of Elections Resolving Improper Removal of Voters from Registration Rolls,” news release, October 31, 2017, https://www.justice.gov/usao-edny/pr/united-states-announces-settlement-new-york-city-board-elections-resolving-improper; U.S. Department of Justice, Office of Public Affairs, “State of Connecticut Agrees to Resolve Claims of National Voter Registration Act Violations,” news release, August 5, 2016, https://www.justice.gov/opa/pr/state-connecticut-agrees-resolve-claims-national-voter-registration-act-violations; U.S. Department of Justice, Office of Public Affairs, “State of Alabama Agrees to Resolve Claims of National Voter Registration Act Violations,” news release, November 13, 2015, https://www.justice.gov/opa/pr/state-alabama-agrees-resolve-claims-national-voter-registration-act-violations; see also Arcia v. Fla. Sec’y of State, 772 F.3d 1335, 1343–48 (11th Cir. 2014) (finding Florida’s purge violated the NVRA’s prohibition on systematic purges within 90 days of a federal election).
55	Ibid.
56	Brief for the League of Women Voters et al as Amicus Curiae supporting Respondents 17, Husted v. A. Philip Randolph Institute, No. 16-980 (2017).
57	Brief for the United States as Amicus Curiae supporting Petitioner, Husted v. A. Philip Randolph Institute, No. 16-980 (2017).
58	See “Husted v. A. Philip Randolph Institute,” Brennan Center for Justice, last modified June 11, 2018, https://www.brennancenter.org/legal-work/husted-v-philip-randolph-institute-0.
59	See Pam Fessler, “Advocates Worry Trump Administration Wants to Revamp Motor Voter Law,” NPR, July 8, 2017, https://www.npr.org/2017/07/08/536006813/advocates-worry-trump-administration-wants-to-revamp-motor-voter-law; Jonathan Brater, Brennan Center for Justice, “The Purge: Ten Years Later?,” June 30, 2017, https://www.brennancenter.org/blog/purge-ten-years-later. For an example of the letters, see https://assets.documentcloud.org/documents/3881818/SOS-Letter.pdf.
60	For example, Vanita Gupta (CEO of the Leadership Conference on Civil and Human Rights and former head of DOJ’s civil rights division under President Barack Obama) said that, “[i]t is not normal for the Department of Justice to ask for voting data from all states covered by the National Voter Registration Act. It’s likely that this is instead the beginning of an effort to force unwarranted voter purges.” Sam Levine, “This DOJ Letter May Be More Alarming Than Trump Commission’s Request For Voter Data,“ HuffPost, July 5, 2017, https://www.huffingtonpost.com/entry/department-of-justice-voter-purge_us_595d22b1e4b0da2c7326c38b. See also Leon Neyfakh, “How Trump’s DOJ Will Try to Purge Voter Rolls,” Slate, July 11, 2017, http://www.slate.com/articles/news_and_politics/jurisprudence/2017/07/ how_trump_s_doj_will_try_to_purge_voter_rolls.html (describing DOJ letters as “a first step toward bringing back a George W. Bush–era strategy of forcing states to aggressively purge their voter rolls under threat of litigation”). The Department did intervene in a lawsuit filed against Kentucky by a private plaintiff organization seeking more aggressive purging. Judicial Watch, Inc. v. Grimes, 3:17-cv-00094, filed November 14, 2017 (E.D. Ky.).

61	Kansas State Rep. Keith Esau, “Interstate Voter Registration Crosscheck Program” (PowerPoint presentation, National Conference of State Legislators, Williamsburg, VA, June 15, 2017), 5,
62	“ERIC,” Electronic Registration Information Center, accessed May 24, 2018, http://www.ericstates.org/.
63	Christopher Famighetti, Douglas Keith, and Myrna Pérez, Noncitizen Voting: The Missing Millions (New York: Brennan Center for Justice, May 2017) .
64	Public Interest Legal Foundation, “248 Counties Have More Registered Voters Than Live Adults,” news release, September 25, 2017, https://publicinterestlegal.org/blog/248-counties-registered-voters-live-adults/. The Brennan Center for Justice and other civil rights groups contacted the same jurisdictions to notify them that some information provided by the organization was misleading. Brennan Center for Justice, “Civil Rights Groups Launch National Effort to Combat Alarming Voter Purge Attempt,” news release, November 22, 2017, https://www.brennancenter.org/press-release/civil-rights-groups-launch-national-effort-combat-alarming-voter-purge-attempt.
65	See Myrna Pérez, Voter Purges (New York: Brennan Center for Justice, September 2008)
66	Missouri, Iowa, Nebraska, and Kansas. See Memorandum of Understanding Between the State of Iowa, Nebraska, and Kansas For the Improvement of Election Administration, December 2005 (on file with the Brennan Center for Justice).
67	These states were Alabama, Arizona, Arkansas, Colorado, Georgia, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Massachusetts, Michigan, Mississippi, Missouri, Nebraska, Nevada, New Hampshire, North Carolina, Ohio, Oklahoma, South Carolina, South Dakota, Tennessee, Virginia, and West Virginia. This information is derived from a spreadsheet obtained from officials in Idaho via public records request (on file with the Brennan Center for Justice)
68	Ibid.
69	Ibid. Some of these matches could also include individuals matched in more than 2 states, so the number of individuals could be lower than 3.6 million.
70	Interstate Voter Registration Crosscheck Program, 2017 Participation Guide, January 2017, goo.gl/zbsygH.
71	See infra text box describing limitations of name and birthdate matching.
72	Sharad Goel et al., “One Person, One Vote: Estimating the Prevalence of Double Voting in U.S. Presidential Elections” (working paper, Stanford University et al., 2017), 3, 26, https://scholar.harvard.edu/files/morse/files/1p1v.pdf.
73	Ibid. 27.
74	See Jonathan Brater, Brennan Center for Justice, “Virginia Offers Lessons for Voter List Maintenance,” November 25, 2013, https://www.brennancenter.org/analysis/virginia-offers-lessons-voter-list-maintenance .
75	See Jim Nolan, “Chesterfield Registrar Delays Purge of Voter Rolls,” Richmond Times-Dispatch, October 9, 2013, http://www.richmond.com/news/local/chesterfield/chesterfield-registrar-delays-purge-of-voter-rolls/article_162e36b5-0be7-5dc8-af9f-48876a167b43.html.

76	Ibid.
77	See Matthew Barakat, “Va. Removes 40K From Voter Rolls Over Democrats’ Objections,” Richmond Times-Dispatch, October 17, 2013, http://www.richmond.com/news/state-regional/va-removes-k-from-voter-rolls-over-democrats-objections/article_2d111de4-49de-523b-bd9c-5d93b7c0a00e.html. Virginia subsequently began to release reports explaining, among other things, the way Crosscheck data was used. They found that only 10 percent of Crosscheck matches were usable for list maintenance. Virginia Department of Elections, Annual List Maintenance Report, 2017, 5, https://www.elections.virginia.gov/Files/maintenance-reports/2017SBEListMaintenancereport.pdf.
78	Alabama law exempts county boards from the requirement that they contact voters to verify suspected address changes when another state provides notice that “the elector registered to vote in another jurisdiction, within or without the State of Alabama, at a date subsequent to the date the elector registered to vote in the jurisdiction of the county board of registrars.” Ibid. at § 17-4-38.1(c). An Election Handbook provided by the Alabama Secretary of State’s office indicates that such notice is sufficient to disqualify and remove the voter when a “registration official from another state notifies registrars in writing that the voter has registered elsewhere.” Alabama Law Institute, Alabama Election Handbook: Eighteenth Edition (2017), 262. But in a December 1, 2016 email obtained through a public records request, the Secretary of State’s Supervisor of Voter Registration provided county registrars a list of voters that Crosscheck suggested had “registered to vote in another state more recently” than in Alabama and directed the registrars to review the list and “take the action you would normally take as if you received notice directly from another state.” Clay Helms (Supervisor of Voter Registration, Office of Alabama Secretary of State), email to local registrars, December 1, 2016, on file with authors. In an interview, Alabama confirmed that the state has considered Crosscheck data as information provided directly from another state; although the state does filter the data to rule out some mismatches, it does not require a notice and waiting-period process. Alabama, which uses ERIC, has not determined whether it will use Crosscheck in future years.John Bennett (Deputy Chief of Staff/Communications Director, Alabama Secretary of State’s Office), interview by Brennan Center for Justice, June 15, 2018.
79	Indiana Code § 3-7-38.2-5(d). The Brennan Center is suing Indiana over this matter. Indiana NAACP & League of Women Voters of Indiana v. Lawson, No. 1:17-cv-2897 (S.D. Ind.). Indiana’s law does not provide notice as required by the NVRA. See Indiana Code § 3-7-38.2-5(d). On June 8, 2018, a federal judge issued a preliminary injunction against the law, meaning it is temporarily blocked. Order Granting Plaintiffs’ Motion for Preliminary Injunction, Indiana NAACP & League of Women Voters of Indiana v. Lawson, No. 1:17-cv-2897 (S.D. Ind.). Available at https://www.brennancenter.org/sites/default/files/legal-work/2018-06-18_Order_Granting_Plaintiffs %27_Motion_for_Preliminary_Injunction.PDF.
80	“2017 Maine Crosscheck Data Review Plan” (providing, “If the matched data shows that the Maine voter record is older than the other state’s voting record, then the Maine record will be cancelled. No notice to the voter is required”). Document produced in response to public records request issued by Brennan Center for Justice and on file with authors.
81	Idaho also removes voters immediately, but its practice permitting immediate removal of individuals flagged by Crosscheck without notice or a waiting period does not violate federal law because Idaho is exempt from the NVRA, and therefore does not have to abide by the NVRA notice and waiting period requirements.
82	State of Arizona, Elections Procedures Manual, 2018 Edition, 104, http://live-az-sos.pantheonsite.io/sites/default/ files/2018%200330%20State%20of%20Arizona%20Elections%20Procedures%20Manual.pdf (allowing removal within 90 days in the case of a “true match,” which could, but need not, involve direct confirmation from the voter). Arizona declined to be interviewed for this report but informed the Brennan Center that it no longer participates in Crosscheck. Eric Spencer (State Election Director), email to Brennan Center for Justice, June 14, 2018, on file with the Brennan Center. The state did participate in the last Crosscheck match in 2017, and its 2018 election manual provides for removal through Crosscheck.
83	These states are Alaska, Florida, Kentucky, Massachusetts, New York, Oregon, Pennsylvania, and Washington. See Jonathan Brater, Brennan Center for Justice, “The Purge: Ten Years Later?,” June 30, 2017, https://www.brennancenter.org/blog/purge-ten-years-later. In addition to those states, Arizona informed the Brennan Center that it is no longer participating in Crosscheck. Eric Spencer (State Election Director, Arizona), email to Brennan Center for Justice, June 14, 2018, on file with the Brennan Center.
84	According to the Center for Investigative Reporting, those 7 states are: Colorado, Georgia, Louisiana, Nevada, North Carolina, South Carolina, and West Virginia. See Aaron Sankin, “Crosscheck is ineffective and insecure. But states aren’t withdrawing,” Reveal, March 26, 2018, https://www.revealnews.org/blog/crosscheck-is-ineffective-and-insecure-but-states-arent-withdrawing/.

85	For a discussion of some of these security lapses, see, for example, Jonathan Brater, Brennan Center for Justice, “The Purge: Ten Years Later?,” June 30, 2017, https://www.brennancenter.org/blog/purge-ten-years-later; Dell Cameron, “As Crosscheck Moves to Secure Voter Data, Hacking Fears Grow Among Experts and Politicians,” Gizmodo, January 24, 2018, https://gizmodo.com/as-crosscheck-moves-to-secure-voter-data-hacking-fears-1822344007
86	See Testimony on the Interstate Crosscheck Program, Kan. H. Comm. on Elections, January 17, 2018 (testimony of Bryan A. Caskey, Director of Elections), http://www.kslegislature.org/li/b2017_18/committees/ctte_h_electns_1/ documents/testimony/20180117_01.pdf; Allison Kite, “Kobach’s Office Will Delay Data Uploads for Crosscheck Voter System to Accommodate Security Review,” The Topeka Capital-Journal, January 17, 2018, http://www.cjonline.com/news/20180117/kobachs-office-will-delay-data-uploads-for-crosscheck-voter-system-to-accommodate-security-review .
87	Sophia Tareen, “Illinois Delays Sending Voter Data to Multi-State Program,” Associated Press, January 16, 2018, https://www.usnews.com/news/best-states/illinois/articles/2018-01-16/illinois-delays-sending-voter-data-to-multistate-program .
88	Ibid.
89	Aaron Sankin, “Crosscheck Is Ineffective and Insecure. But States Aren’t Withdrawing,” Reveal, March 26, 2018, https://www.revealnews.org/blog/crosscheck-is-ineffective-and-insecure-but-states-arent-withdrawing/.
90	See Transcript of Oral Argument at 49:7-50:13, In. NAACP & League of Women Voters of In. v. Lawson, No. 1:17-cv-2897 (S.D. Ind., May 2, 2018).
91	“ERIC,” Electronic Registration Information Center, accessed May 24, 2018, http://www.ericstates.org/ .
92	Electronic Registration Information Center, Inc., Bylaws, December 2016, 21, http://www.ericstates.org/images/documents/ERIC_Bylaws_12-16-2016.pdf ; Tim Harper, Bipartisan Policy Center, “Florida Joins the ERIC Club – and Brings 14 Million New Eligible Voters,” March 20, 2018, https://bipartisanpolicy.org/blog/florida-joins-theeric-club-and-brings-14-million-new-eligible-voters/ . Kentucky informed the Brennan Center in June 2018 that the state would soon be joining ERIC, and agreed to use ERIC data for voter list maintenance in court settlement. Jared Dearing (Executive Director, State Board of Elections), interview by Brennan Center for Justice, June 14, 2018; Judicial Watch, Inc. v. Grimes, 3:17-cv-00094, filed November 14, 2017 (E.D. Ky).
93	See, e.g., Matt Dietrich (Public Information Officer, Illinois State Board of Elections), interview by Brennan Center for Justice, May 8, 2018; Wayne Thorley (Deputy Secretary of State for Elections, Nevada Secretary of State) and Justus Wendland (HAVA Administrator, Nevada Secretary of State), interview by Brennan Center for Justice, May 18, 2018; see also Colo. Rev. Stat. § 1-2-605(7).
94	For example, Alabama credited ERIC with helping to increase voter registration in the state. John Bennett (Deputy Chief of Staff/Communications Director, Alabama Secretary of State), interview by Brennan Center for Justice, June 15, 2018.
95	Gary Bland and Barry C. Burden, Electronic Registration Information Center (ERIC) Stage 1 Evaluation (RTI International, December 2013), 1, https://www.rti.org/sites/default/files/resources/eric_stage1report_pewfinal_12-3-13.pdf. In ERIC’s first year of operation, “ERIC states showed a net improvement in new registration of 0.87 percentage points over non-ERIC states.”
96	Roger Stitt (Voter Registration Manager of Operations, Maryland State Board of Elections), interview by Brennan Center for Justice, May 8, 2018; Kyle Thomas (Director, Voting and Registration Systems, Illinois State Board of Elections), interview by Brennan Center for Justice, May 10, 2018.
97	Kyle Thomas (Director, Voting and Registration Systems, Illinois State Board of Elections), interview by Brennan Center for Justice, May 10, 2018.
98	Memorandum from Meagan Wolfe, Interim Administrator (Prepared by Sarah Whitt, WisVote IT Lead, and Jodi Kitts, WisVote Specialist) to Wisconsin Election Commission Members, May 24, 2018, provided to Brennan Center by Wisconsin Elections Commission (on file with Brennan Center). Wisconsin implemented the supplemental poll lists after some voters experienced problems at a February 2018 election. Through the use of the supplemental poll lists, these voters were able to reactivate their registrations at the polls and vote, rather than having to re-register. Sarah Whitt (WisVote Functional Lead, Wisconsin Elections Commission), interview by Brennan Center for Justice, June 4, 2018.

99	Act of May 18, 2006, 2006 Va. Laws Ch. 940 (S.B. 313), § 1 (codified as amended at Va. Code Ann. §§ 24.2-404 & 24.2-427); Act of June 11, 1997, 1997 Tex. Sess. Law. Serv. Ch. 640 (H.B. 1645), § 1 (codified as amended at Tex. Elec. Code Ann. § 16.0332).
100	Act of Apr. 30, 2009, 2009 Georgia Laws Act 86 (H.B. 549), § 1 (amending Ga. Code Ann. § 21-2-231, to require removal of registration of every person that declined jury duty based on noncitizenship); Act of May 5, 2017, Act 2017 (87 G.A.) ch. 110, H.F. 516, § 4 (amending Iowa Code Ann. § 48A.30 to require cancelation of voter registration of every person that submits documentation to prove noncitizenship for purpose of disqualifying themselves from jury duty); Act of Apr. 1, 2010, 2010 Minn. Sess. Law Serv. Ch. 201 (H.F. 3108), § 12 (adding Minn. Stat. Ann. § 201.158, which requires county auditors to challenge the registration of every registered voter who is identified as a noncitizen by the commissioner of public safety) (later repealed and re-codified as amended at Minn. Stat. Ann. § 201.145); Act of May 23, 2011, 2011 Tennessee Laws Pub. Ch. 235 (S.B. 352), § 1 (adding Tenn. Code. Ann. § 2-2-141, requiring coordinator of elections to begin removal proceedings for registered voters identified as noncitizens in the Department of Safety database).
101	Janelle Ross, “Voter Roll Purges Could Spread to At Least 12 States,” HuffPost, July 31, 2012, https://www.huffingtonpost.com/2012/07/31/voter-roll-purge_n_1721192.html.
102	“About SAVE,” U.S. Citizenship and Immigration Services, last modified September 8, 2016, https://www.uscis.gov/save/about-save.
103	“SAVE Verification Process,” U.S. Citizenship and Immigration Services, last modified June 15, 2016, https://www.uscis.gov/save/about-save/verification-process; Edgardo Cortes and Liz Howard (Virginia State Board of Elections), interview by Brennan Center for Justice, August 9, 2017.
104	Minn Stat. Ann. § 201.145; Tenn. Code Ann. § 2-2-141; Va. Code Ann. § 24.2-427.
105	See Marc Levy, “State Disputes Claim 100K Noncitizens Registered to Vote,” AP News, March 1, 2018, https://www.apnews.com/033c89a4d0d646d386a63117c0c72a11. Relatedly, a Wyoming official told us that when the state investigated a list of potential noncitizens produced from state Department of Transportation records, the state did not determine that there were any noncitizens on the rolls and found that many purported noncitizens had subsequently naturalized and were thus eligible to vote. Jennifer Trabing (Election Policy and Planning Analyst, Elections Division, Wyoming Secretary of State’s Office), interview by Brennan Center for Justice, May 9, 2018.
106	Va. Code Ann. § 24.2-410.1; Roger Stitt (Voter Registration Manager of Operations, Maryland State Board of Elections), interview by Brennan Center for Justice, May 8, 2018.
107	Christopher Famighetti, Douglas Keith, and Myrna Pérez, Noncitizen Voting: The Missing Millions (New York: Brennan Center for Justice, May 2017), (“Other times, noted one administrator, a citizen will forget to check the ‘citizen’ box when filling out a driver’s license form and that will trigger a process which could end in a citizen’s registration being canceled, and also artificially inflate the number of alleged noncitizens who are on the registration rolls.”).
108	Ga. Code Ann. § 21-2-231; La. Stat. Ann. § 18:178; Tex. Elec. Code Ann. § 16.0332.
109	Christopher Famighetti, Douglas Keith, and Myrna Pérez, Noncitizen Voting: The Missing Millions (New York: Brennan Center for Justice, May 2017), (“Several interviewees described how eligible Americans sometimes check a box on a jury service form claiming not to be citizens because they do not want to serve on the jury. ‘One way for people to get out of jury duty is they can say they’re a noncitizen and fill out a card saying they’re not a citizen,’ explained Jacquelyn Callanen, Elections Administrator in Bexar County, Texas.”)
110	U.S. Student Ass’n Found. v. Land, 2:08-CV-14019, filed September 17, 2008 (E.D. Mich)

111	“Common Cause of Colorado, et al. v. Buescher,” Brennan Center for Justice, last modified January 22, 2010, https://www.brennancenter.org/legal-work/common-cause-colorado-et-al-v-buescher
112	Arcia v. Fla. Sec’y of State, 772 F.3d 1335, 1343–48 (11th Cir. 2014).
113	Adam Gitlin and Wendy Wesier, The Justice Department’s Voter Fraud Scandal: Lessons (New York: Brennan Center for Justice, January 2017).
114	Compare Judicial Watch, Inc. v. Logan, 2:17-cv-08948, filed December 13, 2017 (C.D. Cal.), Judicial Watch, Inc. v. Grimes, 3:17-cv-00094, filed November 14, 2017 (E.D. Ky.); Voter Integrity Project NC, Inc. v. Wake Cty. Bd. of Elections, 5:16-cv-00683, filed July 18, 2016 (E.D.N.C.); Bellitto v. Snipes, 0:16-cv-61474, filed June 27, 2016 (S.D. Fla.); Va. Voter’s Alliance, Inc. v. Leider, 1:16-cv-00394, filed April 7, 2016 (E.D. Va.); American Civil Rights Union v. Philadelphia City Commissioners, 2:16-cv-01507, filed April 4, 2016 (E.D. Pa.); American Civil Rights Union v. Rodriguez, 7:16-cv-00103, filed March 4, 2016 (S.D. Tex.); American Civil Rights Union v. Noxubee Cty., 3:15-cv-00815, filed November 12, 2015 (S.D. Miss.); American Civil Rights Union v. Clarke Cty., 2:15-cv-00101, filed July 27, 2015 (S.D. Miss.); American Civil Rights Union v. Martinez-Rivera, 2:14-cv-00026, filed March 27, 2014 (W.D. Tex.); American Civil Rights Union v. McDonald, 2:14-cv-00012, filed Jan. 27, 2014 (W.D. Tex.); True the Vote v. Stewart, 1:13-cv-03369, filed December 12, 2013 (D. Colo) (voluntarily dismissed); True the Vote v. Wintz 1:13-cv-03368, filed December 12, 2013 (D. Colo) (voluntarily dismissed); American Civil Rights Union v. Jefferson Davis Cty., 2:13-cv-00087, filed April 26, 2013 (S.D. Miss.); American Civil Rights Union v. Walthall Cty., 2:13-cv-00086, filed April 26, 2013 (S.D. Miss.); Judicial Watch, Inc. v. Husted, 2:12cv-00792, filed August 30, 2012 (S.D. Ohio); Judicial Watch, Inc. v. King, 1:12-cv-00800, filed June 11, 2012 (S.D. Ind.), with Common Cause Indiana v. Lawson, 1:17-cv-03963, filed October 27, 2017 (S.D. Ind.); Indiana NAACP v. Lawson, 1:17-cv-02897, filed August 23, 2017 (S.D. Ind.); N.C. State Conf. of the NAACP v. N.C. State Bd. of Elections, 1:16-cv-1274, filed October 31, 2016 (M.D.N.C.); Husted v. A. Philip Randolph Institute, 2:16-cv-00303, filed April 6, 2016 (S.D. Ohio); Common Cause v. Kemp, 1:16-cv-00452, filed February 10, 2016 (N.D. Ga.); Ga. State Conf. of the NAACP v. Hancock Cty. Bd. of Elections and Registration, 5:15-cv-00414, filed November 3, 2015 (M.D. Ga.); The Democratic Party of Va. v. Va. State Bd. of Elections, 1:13-cv-1218, filed October 1, 2013 (E.D. Va.); LULAC v. Harris Cty., 4:12-cv-03035, filed October 11, 2012 (S.D. Tex.); Colón-Marrero v. Conty-Pérez, 12-cv-1749, filed September 12, 2012 (D.C.P.R.); Arcia v. Detzner, 1:12-cv-22282, filed June 19, 2012 (S.D. Fla.); Mi Familia Vota Education Fund v. Detzner, 8:12-cv-1294, filed June 8, 2012 (M.D. Fla.); Janis v. Nelson, 5:09-cv-05019, filed on February 18, 2009 (D.S.D.); Common Cause of Colorado v. Buescher, 1:08-cv-02321, filed October 25, 2008 (D. Colo.); Mont. Democratic Party v. Eaton, 9:08-cv-00141, filed October 6, 2008 (D. Mont.); U.S. Student Ass’n Found. v. Land, 2:08-cv-14019, filed September 17, 2008 (E.D. Mich).
115	Judicial Watch, Inc. v. Grimes, 3:17-cv-00094, filed November 14, 2017 (E.D. Ky.); Voter Integrity Project NC, Inc. v. Wake County Board of Elections, 5:16-cv-00683, filed July 18, 2016 (E.D.N.C.); American Civil Rights Union v. Noxubee County, 3:15-cv-00815, filed November 12, 2015 (S.D. Miss.); American Civil Rights Union v. Clarke County, 2:15-cv-00101, filed July 27, 2015 (S.D. Miss.); American Civil Rights Union v. Martinez-Rivera, 2:14-cv-00026, filed March 27, 2014 (W.D. Tex.); American Civil Rights Union v. McDonald, 2:14-cv-00012, filed Jan. 27, 2014 (W.D. Tex.); American Civil Rights Union v. Jefferson Davis County, 2:13-cv-00087, filed April 26, 2013 (S.D. Miss.); American Civil Rights Union v. Walthall County, 2:13-cv-00086, filed April 26, 2013 (S.D. Miss.); Judicial Watch, Inc. v. Husted, 2:12-cv-00792, filed August 30, 2012 (S.D. Ohio).
116	Willie M. Miller (Noxubee County District 4 Election Commissioner), interview by Brennan Center for Justice, May 29, 2018.
117	American Civil Rights Union v. Rodriguez, 7:16-cv-00103, filed Mar. 4, 2016 (S.D. Tex.).
118	Judicial Watch, Inc. v. Logan, 2:17-cv-08948, filed December 13, 2017 (C.D. Cal.).
119	American Civil Rights Union v. Noxubee County, 3:15-cv-00815, filed November 12, 2015 (S.D. Miss.).
120	Sylvester Tate (Noxubee County District 1 Election Commissioner) and Willie Miller (Noxubee County District 4 Election Commissioner), interview by Brennan Center for Justice, May 29, 2018.
121	Judicial Watch, Inc. v. King, 1:12-CV-00800, filed June 11, 2012 (S.D. Ind.)

122	Indiana Secretary of State, “Hoosier Voters to Receive Postcard with Election Information in the Mail,” news release, June 19, 2014, https://goo.gl/Xy2cbp; see also Max Greenwood, “Indiana Purges Nearly 500,000 from Voter Rolls,” The Hill, April 20, 2017, http://thehill.com/blogs/ballot-box/329659-indiana-purges-nearly-half-a-million-from-voterrolls ; Indiana Secretary of State, “Indiana Prepares for Future Elections by Cleaning Up Its Voter Roll,” news release, April 18, 2017, https://goo.gl/9dFzps.
123	Judicial Watch, “Judicial Watch, True the Vote Historic Indiana Lawsuit Forces Statewide Clean-Up of Voter Registration Lists, Permanent Changes in Election Law Procedures,” news release, August 7, 2014, https://www.judicialwatch.org/press-room/press-releases/judicial-watch-true-vote-historic-indiana-lawsuit-forces-statewide-clean-voter-registration-lists-permanent-changes-election-law-procedures/.
124	Public Interest Legal Foundation, Alien Invasion in Virginia, September 2016, https://publicinterestlegal.org/files/Report_Alien-Invasion-in-Virginia.pdf; see also Public Interest Legal Foundation, Alien Invasion II, May 2017, https://publicinterestlegal.org/files/Alien-Invasion-II-FINAL.pdf.
125	Christopher Famighetti, Douglas Keith, and Myrna Pérez, Noncitizen Voting: The Missing Millions (New York: Brennan Center for Justice, May 2017), .
126	These noncitizen removal figures are derived from Virginia’s annual list maintenance reports. Virginia Department of Elections, Annual List Maintenance Report, 2015, https://www.elections.virginia.gov/Files/maintenance-reports/2015SBEListMaintenancereport.pdf; Virginia Department of Elections, Annual List Maintenance Report, 2016, https://www.elections.virginia.gov/Files/maintenance-reports/2016SBEListMaintenancereport.pdf; Virginia Department of Elections, Annual List Maintenance Report, 2017, https://www.elections.virginia.gov/Files/maintenance-reports/2017SBEListMaintenancereport.pdf.
127	League of United Latin American Citizens of Richmond v. Public Interest Legal Foundation 1:18-cv-00432-LOIDD V.A.E.D, https://www.southerncoalition.org/wp-content/uploads/2017/01/LULAC-of-Richmond-v.-PILF.pdf. See also Pema Levy, “Member of Trump’s Voter Fraud Commission Sued for Voter Intimidation,” Mother Jones, April 12, 2018, https://www.motherjones.com/politics/2018/04/member-of-trumps-voter-fraud-commission-sued-for-voter-intimidation/; Pema Levy, “Trump Election Commissioner Used Dubious Data to Allege an ‘Alien Invasion,’” Mother Jones, July 18, 2017, https://www.motherjones.com/politics/2017/07/trump-election-commissioner-used-dubious-data-to-allege-an-alien-invasion/; Stephen Dinan, “American Kicked Off Virginia Voter Rolls as ‘Declared Non-Citizen,’” The Washington Times, June 18, 2017
128	See supra text box on challenges.
129	Alaska, California, Colorado, District of Columbia, Georgia, Illinois, Maryland, New Jersey, Oregon, Rhode Island, Vermont, Washington, West Virginia. For a detailed description of the status of implementation of automatic voter registration in these states see “Automatic Voter Registration,” Brennan Center for Justice, last modified April 17, 2018, https://www.brennancenter.org/analysis/automatic-voter-registration.

Appendix A: Federal Statutory Regulation of Voter Purge Practices

Purge practices are regulated by a combination of federal and state law. Below is a summary of federal statutes:

VOTING RIGHTS ACT

As a general matter, the Voting Rights Act (VRA), 52 U.S.C. § 10301 et seq, prohibits discrimination in voting. The Supreme Court has held that this prohibition applies to purges.¹ Prior to 2013, certain jurisdictions were required to seek federal preclearance of purge practices before they were implemented.² However, the formula by which these jurisdictions were covered was invalidated in Shelby County v. Holder,³ effectively ending preclearance until Congress issues a new formula. Purge practices must still comply with Section 2 of the VRA, which bans discriminatory voting practices.⁴

NATIONAL VOTER REGISTRATION ACT

The National Voter Registration Act (NVRA) is the most comprehensive federal law regulating voter purges and applies to 44 states. Six states (Idaho, Minnesota, New Hampshire, North Dakota, Wisconsin, and Wyoming) are exempt because they had election day registration or no voter registration as of the date provided by the NVRA. These exemptions make sense because purge consequences are much less grave in a state that permits anyone eligible who is not on the registration rolls to register and to vote on Election Day (or does not require them to register in order to vote).

The law discusses five categories of removal from voter rolls: (1) request of the registrant; (2) disenfranchising criminal conviction; (3) mental incapacity; (4) death; and (5) change in residence.⁵ The NVRA sets forth a series of specific requirements that apply to purges of registrants believed to have changed residence.⁶

The law also contains a series of additional proscriptions on state practices. For example, it provides that list maintenance must be uniform, nondiscriminatory, and in accordance with the Voting Rights Act.⁷ It also prohibits systematic voter purges (those programs that remove groups of voters at once) within 90 days of a federal election.⁸ The Act also has provisions that apply on Election Day if a voter has changed address. Voters who have moved within a jurisdiction are permitted to vote at either their new or old polling place (states get to choose), while purged voters — mistakenly believed to have moved — who show up on Election Day have the right to correct the error and cast a ballot that will count.⁹

HELP AMERICA VOTE ACT

The Help America Vote Act of 2002 (HAVA) reaffirms the requirements of the NVRA and contains additional regulations for voter list maintenance.¹⁰ For example, HAVA requires states to create statewide voter registration databases with unique identifiers for registered voters.¹¹ The law also requires states to attempt to verify the validity of information submitted by voter registration applicants.¹² HAVA also ensures that certain voters, including those who do not appear on poll books, are permitted to vote provisional ballots at minimum.¹³

1	Young v. Fordice, 520 U.S. 273 (1997).
2	52 U.S.C. § 10304.
3	570 U.S.C. 2 (2013)
4	52 U.S.C. § 10301(a).
5	52 U.S.C. § 20507(a)
6	See 52 U.S.C. § 20507(d)(1).
7	52 U.S.C. § 20507(b)(1)
8	52 U.S.C. § 20507(c)(2)(A).
9	52 U.S.C. § 20507(e)
10	52 U.S.C. § 21083(a).
11	52 U.S.C. § 21083(a)(5)(A).
12	52 U.S.C. § 21083(a)(5)(B).
13	52 U.S.C. § 21082

Appendix B: What Explains a Jurisdiction's Purge Rate?

Removal Rate

D (Preclearance Condition Lifted)*D (2014)

0.0150***
(0.00166)

D (Preclearance Condition Lifted)*D (2016)

0.0240***
(0.00207)

D (Preclearance Condition Lifted)

0.00605***
(0.00193)

Median Age

-0.000601***
(0.000168)

-0.000601***
(0.000169)

Percent of Residents Who Moved in Past Year

0.0582***
(0.0124)

0.0578***
(0.0124)

Log (Median income)

0.00639**
(0.00283)

0.00625**
(0.00283)

Log (Voting Age Population)

-0.000184***
(0.000608)

-0.000182***
(0.000608)

Log (Percent Black)

-0.00124***
(0.000362)

0.00125***
(0.000362)

D (Secretary of State Appointed by Governor)

0.00634***
(0.00187)

0.00636***
(0.00187)

D (Secretary of State Appointed by Legislature)

0.0168***
(0.00202)

D (State Legislature Controlled by Republicans)

0.0138***
(0.00122)

Constant

0.0339
(0.0293)

0.0353
(0.0293)

Observations R-squared

9,057
0.069

9,057
0.073

Robust standard errors in parentheses, clustered by county.
Year dummies not shown.

*** p<0.01, ** p<0.05, * p<0.1

Notes: Data are from the 2010, 2012, 2014, and 2016
reporting periods. Includes jurisdictions that reported in
each time period.

Sources: U.S. Election Assistance Commission, U.S. Census Bureau: American Community Survey 5-Year Estimates, National Conference of State Legislatures

Appendix C: Relationship Between Purge Rates and Provisional Ballot Rates

Provisional Ballot Rate

Removal Rate

0.0177**
(0.00697)

Turnout Rate

-0.00553***
(0.00164)

Log (Medium Income)

0.00189***
(0.000504)

Log (Percent Black)

-0.000554*
(0.000308)

Log (Percent White)

-0.00453***
(0.00132)

D (Implemented Strict Voter ID Requirement)

-0.00314
(0.000406)

Constant

-0.0185***
(0.00523)

Observations
R-squared

1,854
0.741

Robust standard errors in parentheses, clustered by county. Year and state-level dummies not shown.

*** p<0.01, ** p<0.05, * p<0.1

Notes: Data are from the 2010, 2012, 2014, and 2016 reporting periods. Includes jurisdictions covered under Section V of the Voting Rights Act at the time of the Shelby County decision in 2013 that reported in each time period.
Sources: U.S. Election Assistance Commission, U.S. Census Bureau: American Community Survey 5-Year Estimates, National Conference of State Legislatures.

Regression analysis shows that the higher a covered county's purge rate the higher their provisional ballot rate. Each 1 percent increase in removal rates was associated with an additional 1.8 provisional ballots for every 10,000 ballots cast. Although this number is small, the median for these jurisdictions in the 2012 presidential election was fewer than 1 provisional ballot per 10,000 cast. Importantly, this statistically significant relationship holds even after controlling for other sociodemographic factors such as population, turnout rate, racial composition, political orientation, and implementation of strict voter ID requirements.

As with any statistical study of this sort, it is impossible to determine whether the increase in purge rates in any particular county is responsible for an increase in provisional ballots. However, a closer look at the numbers in a few jurisdictions suggests how this relationship might work.

Shelby County, Alabama, the jurisdiction at issue in Shelby County v. Holder, is illustrative. After preclearance ended in 2013, the county’s removal rate more than doubled, from 5.0 percent to 10.4 percent. In 2014, more than 18 percent of the county’s voters were purged. In 2012, the provisional ballot rate was 0.15 percent, virtually identical to the national average of 0.16 percent. Following years in which the county purged an average of 10 percent of voters, the provisional ballot rate tripled to 0.45 percent.

Montgomery County, Alabama, also had to seek federal preclearance for purges in the past. From 2009 to 2012, when preclearance was required, the average two-year removal rate was 4.7 percent, well below the national average. But after Shelby County effectively ended preclearance, the removal rates increased dramatically, nearly tripling to 12.0 percent. Montgomery County’s numbers are similar to Shelby County’s. In the two years ending in 2014, a period covering the cessation of preclearance, Montgomery County had a massive purge in which 21 percent of voters were removed. Subsequently, the provisional ballot rate shot up from 0.31 percent in the 2012 presidential election to more than 1 percent in the 2016 election.

Stay Connected to the Brennan Center

Visit our website at www.brennancenter.org
Sign up for our electronic newsletters at www.brennancenter.org/signup

Insider | Up-to-the-minute info on our work, publications, events, and more.

Justice Update | Snapshot of our justice work and latest developments in the field.

Money in Politics | Latest state and national developments and original analysis
.
Redistricting Round-Up | Analysis of current legal battles and legislative efforts.

Fair Courts | Comprehensive news roundup spotlighting judges and the courts.

Liberty & National Security | Updates on privacy, government oversight, and accountability.

Twitter | www.twitter.com/BrennanCenter
Facebook | www.facebook.com/BrennanCenter
Instagram | www.instagram.com/brennancenter

New and Forthcoming Brennan Center Publications

The State of Voting 2018
Wendy R. Weiser and Max Feldman

Election Integrity: A Pro-Voter Agenda
Myrna Pérez

The Justice Department’s Voter Fraud Scandal: Lessons
Adam Gitlin and Wendy R. Weiser

Florida: An Outlier in Denying Voting Rights
Erika Wood

Democracy & Justice: Collected Writings, vol. X
Brennan Center for Justice

The Fight to Vote
Michael Waldman

A Federal Agenda to Reduce Mass Incarceration
Ames Grawert, Natasha Camhi, and Inimai M. Chettiar

Extreme Vetting and the Muslim Ban
Harsha Panduranga, Faiza Patel, and Michael W. Price

For more information, please visit www.brennancenter.org

Converted to HTML by Al Swilling, SENAA International