Thursday Feb 25

Winter 2020

Voter Purges—A Key Element of Suppression


In the United States the all-important voter lists are maintained by the states and counties and supervised by either secretaries of state or election commissions.  In order to secure orderly elections from the local to federal level, these bodies must assure that the lists are accurate. 

Under the US Constitution, once a citizen is registered, her voting privileges cannot be taken away. There are only two exceptions:  death or changing address.  Nothing can cure the loss of voting privileges by death, but if a voter is changing address within the state, their right to vote is restored by notifying the authorities of their new address or, if they have moved to another state, by re-registering after having met the residency requirement of that state.

 The United States Supreme Court in the Husted v. A. Philip Randolph Institute decision in June, 2018, brought by the state of Ohio, issued a critical ruling.  They determined in a 5-4 vote that a voter could be taken off the rolls, if the individual had not voted in two election cycles and the state made an effort to contact the voter to verify their address and was unsuccessful in doing so.  This could not be a “caged” effort targeting only a segment of voters, for example African-Americans or women.  It needed to be premised on repeated elections where the voter did not exercise her franchise for an additional two federal elections or four years in the Ohio situation. If non-voting and failure to respond were coupled, then the state in seeking to maintain the accuracy of the voter file could purge such questionable voters from the list. 



This decision and the general requirements of the list-cleaning process rely on two Congressional statutes, the National Voter Registration Act (NVRA) of 1990 and the Help America Vote Action (HAVA) of 2002, one of which is now thirty years old and the other eighteen years old.  Postcards play a central role in the decision and the trigger to purge a voter. 

Postcards are increasingly an archaic communication method.  Five years ago in 2015, a story in the Washington Post wondered if they were already obsolete, The U.S. Postal Service processed 770 million stamped postcards in fiscal 2014, down from 1.2 billion in 2010, according to figures provided by agency spokesperson Sue Brennan. The most recent statistics from 2019 complied by the USPS indicate that the number has now fallen to a bit over 550 million and still dropping like a rock.  Yet, the decision from 2018 hinges on receipt and return of a postcard to verify a continuing or changing address in a time when mass communication has migrated to other platforms.

Mail in general is just not working for a lot of people.  The Upshot department at the New York Times surveyed 4400 adults of various ways and means on the question of how we sometimes mess up even on these relatively simple tasks and interactions with business and government.  The Times was most interested in the draconian way that some states use mail response to determine continued eligibility for healthcare protection under the Affordable Care Act based on the simple receipt and return of a letter from the government.  They asked the survey participants if they ever flubbed up, and then they calculated the percentages based on income.  Left mail unopened, 25% under $20,000 and 26% over $100,000; forgotten a bill, 30% on the low end and 26% on the high end; and, missed an appointment 39% for lower income and 26% for higher. The rich, the poor, and regular working people also fail to bring the right documents to the Motor Vehicle office between 23% and 19%, let their car registration expire between 14% for the poorer and 15% of the richer, or let their health insurance expire, 7% of the rich and 16% of the poor.  In point of fact, 25% of Americans have received a government form in the mail, according to this survey, that they did not understand. 

On this fragile foundation, rests American’s continued ability to exercise their right to vote!

The Problem with Purges: Cases in Point

The problem in a nutshell is that purges are not accurate enough, and states are not doing enough to protect voters right to vote, while they are too quick to purge. 

The Voter Purge Project is now processing the pure voter files on a regular (weekly, monthly, or quarterly basis) for almost twenty states.  We collect these reports as regularly as they are available or in some cases (WI, TX, etc) as often as we can afford in order to compare the voter lists and determine who is being dropped or purged, and whether or not that is accurate.  We’ve learned a lot about the way voter files are kept and maintained by the states, and we are concerned about the methodology, the lack of transparency, and the disenfranchisement of so many voters. 


In a well-reported incident involving our partner, the Ohio Voter Project, we were able to restore the voting privileges of 40,000 voters in a pre-announced purge of 235,000 by the secretary of state in Ohio. 

The purge list was posted on the internet before being implemented.  We had been processing the lists for several years so we were almost immediately able to review the list and find the errors.  In one case, it was clear that about 25,000 had voted recently.  In the other instance about 15,000 had correct addresses. 


The state of Georgia has become notorious for its controversial purges before the last gubernatorial race when a last-minute purge by the secretary of state may have been the difference in allowing him to win the election.  A purge was announced in the spring of 2020 of more than 100,000 voters.  Observers were quickly able to find an error involving 10,000 voters where the voting date period had been incorrectly used. 

The Voter Purge Project looked at the proposed purge list on two levels.  Georgia, as a pre-clearance state under the Voting Rights Act, is still required maintain racial data within its voter file. Despite public claims to the contrary, our analysis of the file found that the percentages of purges were aligned with the racial demographics of the state, almost precisely with African-American voters being purged one-tenth of a percentage point less than other voters. 

On the other hand, we worked with volunteers connected to Georgia State University who canvassed the purge lists we provided and completed almost one-hundred visits with people on the list.  16.3% of the visits established that voters were wrongly purged.  Four were former felons, who had been purged despite having satisfied their sentences.  One had registered, but had never voted.  Nine others still lived at the correct address and therefore should not have been disenfranchised, and four that the VPP team talked to were shocked and had no idea that they had been purged. 

These stories run the gamut of our experience in determining the accuracy of the list.  83.7% of the list were wrong address, abandoned houses, or deceased, but that is not good enough when it comes to a citizen’s right to vote.  Extrapolating against the entire purge list of 100,000 that would mean that 16,300 voters were being wrongly taken off the rolls. 


The VPP team tried another type of test to determine the accuracy of the purge list in Florida.  The state allows voters to include their phone numbers with their other information.  Not all voters do so, but a significant percentage do.  We picked up the phone and called 143 of the numbers around Miami, Clearview, and Tampa/St. Pete in random zip codes to see what we could determine. 

Of the 143, calls we made to test the list, 89 were wrong numbers or out of service and 5 of the 80 had deceased, which is actually a lower number than we would have expected in Florida.  54 had active numbers where we were able to talk to people, leave messages, and verify the numbers, establishing that with a minimal effort, the State of Florida could have made the effort to verify their information as well in 37.8% of the cases.

We did talk to two people who told us that they were wrongly purged.  One, Fermin Nuñez’, told us he received a piece of mail that looked sort of like generic junk mail, informing him that his name was slated to be purged, if he didn’t provide gov’t issued current ID.  Then he received a follow-up mailing a week or less later, informing him that his name had already been purged.  There are so many things wrong with the Florida practice, based on Mr. Nunez’s statement.  The proverbial postage-paid, correctly addressed postcard was not provided, but instead he was supposed to somehow find the wherewithal to provide his ID in order to prevent being purged.  Underlining how pro forma the process is in Florida, had he been able to do this in a lickety-split, it seems like Florida already had his purge letter in the mail before he could have done. 


In Michigan, the VPP team undertook a different kind of test than on purges there.  We scoured the internet and social media to determine whether we could locate these “purged” voters with modern, 21st century tools.  A volunteer using a Facebook, Google, LinkedIn, and a contact search tool called “iwave”, worked through a random list of 139 voters who had been purged.  61.9% were still living in Michigan, many in the same towns as on the voter list.  38.1% were unable to be contacted or were identified as having moved to other parts of the country. 

It took an average of two-minutes to identify someone on the list, one way or another.  Given the level of Michigan’s drops and purges during a year, using only this methodology to verify the accuracy of purges would entail some increase in staffing to produce the solution (less than a half-dozen additional people), but everything being equal, if that meant reducing the inaccurate purges drastically, we believe it would be worth it. 

Additionally, as the Voter Purge Project is doing now, assisting people who have moved within Michigan to register at their new addresses, if that is their situation, would be a vital and invaluable democratic service.

North Carolina

Action NC, our partner in North Carolina, visited roughly 100 addresses to determine the accuracy in field of voters listed by the state as purged in random zip codes in Charlotte.  Forty-seven visits were successfully completed with direct conversations and data verification.  Of these forty-seven, fourteen were purged inaccurately, having voted previously and continuing to live at the recorded address of their registration.  The error rate on this test of the purge list was 29.78 or approximately 30%, meaning roughly one of three on the purge list were incorrect. 


Voter purges have been a partisan donnybrook over the last year.  The list is maintained and overseen by a bipartisan Election Commission.  The commission believed the purge list was so faulty that they unanimously voted to not purge any voters until after the election in November.  Lawsuits have been joined to force the purge and have gone back and forth through the court system in Wisconsin to achieve the bitter stalemate that there will be no action until after the election.  

The merits and politics of the lawsuits are past our scope, but if the Election Commission, that manages the voter list, argues the purges are flawed, we find their position credible.  The list is exorbitant to process, as we have noted earlier ($12,500 each pull), but having examined the full list and the proposed purge list, as provided to us, we can easily see the problem.  First, no birth dates or years are included, making it hard to separate many individuals with similar names.  Wisconsin is the only one of the twenty states we process where no dates are included.  On the other hand, Wisconsin does furnish the phone numbers for a relatively high percentage of people on the list, making it possible to verify at some level the accuracy of the list. 

Without being privy to all of the deliberations of the Election Commission, we have to believe lacking the dates and being able to make some phone calls triggered their conclusion that the purge list was too damaged to fix without disenfranchising tens of thousands of voters.

Going Out the Back Door

Voter suppression is one edge of the sword that concerns the Voter Purge Project, but the other is a recognition that if the voter registration lists are only being monitored for new registrations and not also monitored for purges and drops, then it is impossible to determine whether or not the voter pool is being expanded or shrunk.  Here is a sample of seven of the many states being monitored by the Project, many of which are in the short list of states receiving so much attention as battlegrounds for the coming election.


CO:             338,603 new, 66,267 dropped

FL:              629,362 new, 639,140 dropped

GA:             541,146 new, 368,556 dropped (since 1 December 2019)

MI:             549,989 new, 131,300 dropped

NC:             579,078 new, 288,764 dropped

OH:            695,697 new, 295,872 dropped

PA:             160,175 new, 198,339 dropped (through June 2020)


Except as noted, these figures are all contemporaneous, following calendar 2020 through August 2020, and don’t necessarily reflect registration surges in the last several months of registration.  At that point, Florida has lost almost 10,000 net voters from its list, despite the fact that felons who were allowed to register for the first time in that state based on the expansion of the franchise directed by Florida voters in the 2016 initiative.  Court decisions have gone back and forth to dilute the impact of that measure which would have been the greatest voter expansion since the voting edge was raised in the 1970s.  Nonetheless, it is astounding that the net voter number is down in the state since general registration drives have occurred throughout the year and since some number of felons who have satisfied their sentences and any outstanding fees would have still been expected to yield a major increase.  Pennsylvania also dropped more than 38,000 voters through the first six months of the year. 

Relatively speaking, Colorado and Michigan dropped a smaller percentage compared to total registration increases.  North Carolina only added one new voter for every two that were newly registered, once drops were counted.  Ohio was slightly better gaining 57 net new voters for every 100 new registrants.  On the other hand, Georgia netted only one new voter for every three that were registered after factoring in purges.  It’s a puzzle to understand why equivalent states in population to these, like Colorado and Michigan, with electorate sizes that are comparable, purged so many few voters.

Does it matter?  Absolutely!  Partisan strategists are constantly making their cases for superiority based on their claims of greater or lesser new registrants compared to their opponent in this state or that.  Without careful calculations on how many voters are leaving through the back door, any claims about the number of new votes coming in through the front door are specious.

Data & Algorithm Transparency

How do the states handle these millions of pieces of data that are so critical to creating a seamless electoral system in our democracy that would allow all eligible voters to practice their franchise?  These are public officials.  The states all have open records or freedom of information laws.  This shouldn’t be hard to figure out and then fix.


The Voter Purge Project enlisted one of our partners, the Labor Neighbor Research & Training Center (LNRTC) to make freedom of information requests to all fifty states with just those questions.  What software and algorithm is each state using?  How do they make the determinations on the lists? We made the requests in the spring of 2020. We are still waiting for most of the responses.  When we have received replies, the answers were almost as opaque as the process itself for the most part.  At this point there has only been actual information of any kind provided by less than one-third of the states that have responded.  The other two-thirds are engaging us in most cases in a button-button-find-the-button exercise, often arguing that we need to route to a different contact or address.  Even the states that have replied thus far are not doing so with any specificity or clarity.  As we continue to gather responses and leap tall hurdles in one stride, we will report in further detail.

Do Election Officials Want the Voter List to Be Accessible?

A similar inexplicable phenomenon exists in the handling of the voter lists themselves by elected officials.  The NVRA is clear that the list has to be accessible and reasonable.

Following that dicta, many states make the voter files accessible without cost through a click on a website.  Other affix minimal handling costs.  And, then there is Alabama that charges $33,000 for each list, Wisconsin $12,500, and a number of other states that charge anywhere between $1500 and $6000 per list.  In the modern, digital age there is really no excuse for such exorbitant charges that on their face seem to contradict the letter and spirit of the NVRA.

To monitor purges, as the Voter Purge Project does, requires processing lists on a regular and systematic basis as often as they are updated or at the least once per month.  In some cases, we process and analyze new lists on a weekly basis, if the states update on that basis.  To do so in every state, given some of the excessive and unreasonable costs established for the lists, would cost more than a million dollars annually. 

It is hard to avoid the conclusion that some states simply do not want their lists to be available for access and inspection.  In point of fact, some states make the lists almost inaccessible even when they are acquired.  Kansas when it provided the list also provided a separate file with the categories explaining how the list is organized.  That’s a good start, except that the categories did not match how they actually organize the list.  Arizona provides the list, due to litigation that reduced the price brought forward by Project Vote some years ago, but initially didn’t share the identifier that allows the list to be processed easily. 

What are We Doing About It?

Given the impending election, the Voter Purge Project has engaged in a thorough and aggressive program to reach out to as many of the individuals on the purge list as we are able, even though the full scope of the task is past our capacity.  We contracted for a phone match on more than 1.5 million purged voters in 15 states, and another half-million as a test of the accuracy of the lists and to outreach voters where purges are disguised more effectively.  The net result harvested 260,000 valid cell phone numbers.  We have now texted these individuals twice.  First, alerting them that they have been duly registered voters, but now may find themselves disenfranchised for some reason or error by their state.  We included a link to the local registration source in their state and the texts all derive from an area code in state.  The second text also alerted these purged voters that they can text back to us with any questions or for assistance in re-registering.

Once we have received the responses, we are activated the next stage of the enfranchisement program.  Volunteers are responded to those citizens who have asked for help in re-registering.  We are also called to check on people who have said that they were left off the list incorrectly or have indicated to us that they are in the process of re-registering.  We wanted to make sure that those individuals with early registration deadlines had done what they need to do to get back on the rolls, and, where allowed, to request a mail ballot, if needed.  The last stage of our effort focused on the states where we have been processing the full voter list who are among those more than twenty states that had same day registration right up to November 3rd, so that we could encourage them to complete the registration process.  Finally, we sent a “reminder to vote” message to all of those with whom we had contact.

There Must Be A Better Way

Voter purges are not a trivial matter.  They are a big deal!

If the numbers on our field tests held up throughout all of the voters dropped through the lists in North Carolina this year, 86,629 people would be pushed off the rolls in error.  The same numbers across the Georgia list would see 60,443 losing their right to vote.  From our experience in Ohio, we would fear that 59,174 might be wrongly disenfranchised.  These numbers are hugely significant.  Elections from the bottom to the top of the ballot are determined by far fewer numbers than are being ejected from the rolls.  This system is broken and must be fixed.

The American people need to have confidence in the election system.  To achieve such confidence, they need to have complete surety on the accuracy of the voter list.  They can’t wonder if their names might have disappeared when they show up to vote on election day.  They need a transparent system that allows purges to be understood, tested, and corrected, rather than shrouded in mystery, and, worse, politics.  Notification cannot rest on the fragile reliance on postcards, quickly disappearing from regular use.  The new tools of texting, email, and social media are viable ways to provide additional alerts to problems on voting status as well as the ways and means to correct errors. Election officials and secretaries of state need to be as aggressive in the reregistration of voters as they move within cities in a state to keep voters eligible and on the rolls, as some of them have been at purging voters from the list. Many tools are available, as we have indicated, but the resources have to be devoted to developing more and to using efficiently and effectively what exists. 

In a democracy, such as we claim in America, we have to not just believe in universal suffrage, but we need to guarantee it exists and make the effort to assure all voters that they have not only the right to vote, but the protected ability to exercise that right.  Without understanding and correcting the problems with voter purges, now seen as a key weapon in voter suppression, no such assurances can be made.

About The Voter Purge Project:

The Voter Purge Project protects eligible voters against disenfranchisement by monitoring, reporting on, and organizing against wrongful voter purging. It is a collaboration among three non-profit, 501c3 organizations: The Ohio Voter Project, ACORN International, and the Labor Neighbor Research & Training Project.

Steve Tingle-Hock is a data engineer and the founder and director of the Ohio Voter Project, as well as a co-founder of the Voter Purge Project with his partner and co-founder, Wade Rathke, who is also the Chief Organizer of ACORN International and Executive Director of Labor Neighbor Research & Training Center.

Subscriber Login

Latest Issue

Read The New Issue


Joomla! Debug Console


Profile Information

Memory Usage

Database Queries