Pennsylvania Election Ballot

Throughout the recent budget process, some lawmakers made unsubstantiated claims that Pennsylvania’s voter rolls are bloated – full of people who have long ago moved or died. This rhetoric, along with proposed legislation to “modernize” voter roll maintenance, paints a picture of hundreds of thousands of ineligible voters submitting illegal ballots and skewing Pennsylvania’s election results. These concerns about the integrity of the Commonwealth’s voter registration data were based in large part on findings that an unaddressed “duplicate voter problem” exists across the country. While the original author later backtracked his initial argument (because duplicate registrations are being systematically identified by ERIC, leading the British author to wonder why every state is not part of the interstate compact), there are still questions about the quality of our voter lists. 

So how do we know if Pennsylvania voter rolls are “clean” or if ineligible voters are committing fraud? A first step is to monitor the public voter file and analyze the data over time, which is exactly what VoteShield has been doing since the end of 2017. 

Want more Democracy Insights?

Subscribe to our weekly newsletter

In order to determine the health of the voter registration list, each week we request a copy of every voter registration record in the state. This information is publicly available. The public version of the file does NOT include sensitive personal information like driver’s license or social security numbers. 
Our software records every change to every field in this file each week, and can flag anomalies that might indicate potential violations to state or federal law, cyber-attack, or otherwise problematic changes to voter registration records. We share our findings with state and local election officials to resolve any potential issues, but we operate fully independently and objectively.

With this unique archive and software, we are able to monitor the changes, additions, and deletions made to the voter rolls in each county across the state, as they happen in near real-time. This allows us to assess whether each county across the state appears to regularly be conducting legally mandated list maintenance and removing ineligible voters from their voter rolls on an annual basis.

About VoteShield and Protect Democracy

VoteShield, a project of Protect Democracy, is a software platform that analyzes publicly available voter databases and tracks changes over time. Our mission is to ensure that U.S. voter rolls are secure, accurate, and complete. We can identify unexpected or improper changes, and we work collaboratively with election officials to resolve potential issues in their voter databases before they might adversely affect an upcoming election. 

Protect Democracy is a cross-ideological nonprofit group dedicated to defeating the authoritarian threat, building more resilient democratic institutions, and protecting our freedom and liberal democracy. Our experts and advocates use litigation, legislative and communications strategies, technology, research, and analysis to stand up for free and fair elections, the rule of law, fact-based debate, and a better democracy for future generations.

Why are we qualified to discuss this topic?

Voteshield has been archiving and analyzing Pennsylvania’s public voter file on a weekly basis since November of 2017, and we have been monitoring Pennsylvania’s public absentee ballot file since 2020. Since then, every public voter record or absentee ballot request in Pennsylvania that has been added, removed, or modified in any way has been recorded and analyzed by our systems. Our software has tracked over 62 million granular field changes across the state’s 67 counties, which when viewed as a time series demonstrates the regular, ongoing voter list maintenance program in Pennsylvania:

Pennsylvania’s voter rolls are regularly cleaned

Pennsylvania’s voter rolls are regularly cleaned

State law requires counties to conduct annual voter registration list maintenance that is compliant with the National Voter Registration Act (NVRA). This includes systematically identifying and inactivating the records of registered electors suspected of having changed addresses based on data provided by USPS. It also includes removing the records of registered electors upon the elector’s request, upon the death of the elector, upon confirmation that the elector moved out of the county, or after the elector’s record has been inactive for two consecutive federal general elections. Inactivations and removals are both required to take place at least once every calendar year.

Our analysis of Pennsylvania’s voter file data since the start of 2020 shows that, consistent with these requirements, every county in Pennsylvania has both inactivated and removed voters at least once every calendar year. In most cases, systematic inactivations have taken place twice a year and removals have taken place multiple times per month. Between the last presidential election on November 3, 2020 and May 1, 2024:

  • Approximately 1.1 million unique active records have been inactivated (11.49% of all unique active records in the voter file during this period).
  • Approximately 1.8 million unique records have been removed (14.17% of all unique records in the voter file during this period).

Moreover, it appears that counties are routinely and systematically removing records that have been inactive for two federal general election cycles. In order to better understand these dynamics, we identified records in our data that appeared to have been inactive for two consecutive federal general elections as “likely eligible for removal”1Note that state and county boards of elections have the most accurate information about eligibility for removal from the voter rolls. It is highly likely that some of the records that we identified as “likely” eligible for removal were not actually eligible for removal. and evaluated whether those records were removed from the voter file in the leadup to the next general election. 

  • Following the 2020 general election, 94% of the records that we identified as “likely eligible for removal” were removed before the 2022 general election. 
  • Since the 2022 general election, 93% of the records that we identified as “likely eligible for removal” have already been removed. 

These removals have been fairly consistent across counties; over both periods, every county removed the majority of records that we identified as likely eligible for removal.

An inaccurate registration rarely leads to a fraudulent vote cast

An inaccurate registration rarely leads to a fraudulent vote cast

In Pennsylvania, voter fraud is exceptionally rare, and all known cases have been investigated.

In the 2020 presidential election more than 6.95 million Pennsylvanians voted. Of those nearly seven millions ballots cast, only six specific allegations of voter fraud were reported to district attorneys. In every case, a voter attempted to fraudulently cast a ballot for a family member. All reported incidents were vigorously investigated and prosecuted. 

For example, two women were charged (separately) for obtaining ballots for their deceased mothers in the 2020 election in Bucks County. To illustrate the process of investigating those claims of fraud, the District Attorney’s office commented:

In some of these investigations, the detectives obtained DNA samples from ballots and voters for analysis, took handwriting samples from voters, prepared and served search warrants for IP addresses for electronic mail-in-ballot application submissions, reviewed video surveillance of county government facilities, contacted and spoke to postal carriers and contacted and spoke to FBI agents assigned to voter irregularities.

While not a voter registration issue, a report of fraudulent voter activity in the most recent primary, in which a voter allegedly filled out and returned their spouse’s ballot, was also properly investigated by the Berks County Board of Elections and referred to the District Attorney’s Office. Commissioner Christian Leinbach ensured the public that all allegations of fraud are investigated and commented, “It is not our role to adjudicate, but our role is to investigate to the extent necessary to determine whether or not this should be forwarded to the proper authorities.”

Numerous processes keep ineligible voters from successfully registering, and update or remove inaccurate records before an election.

County election officials verify every voter’s identity before they receive a ballot, whether it is requested in person or by mail. Title 25 of Pennsylvania’s consolidated statutes requires that commissioners, clerks and registrars verify each applicant’s: 

  • Citizenship (voters must be a citizen of the United States for at least one month), 
  • Age (voters must be over 18 years old), and 
  • Residency (voters must be a resident of their voting district for at least 30 days prior to the next election). 

There are criminal penalties, both for potential voters who knowingly submit false information on their applications and for government officials who knowingly process registration applications with unverified eligibility.

There is room for improvement in Pennsylvania

There is room for improvement in Pennsylvania

Given its size and complexity, it is likely that some duplicate records in the Pennsylvania voter database do exist. The existence of duplicate records can undermine public confidence in elections, despite instances of voter fraud in Pennsylvania being extremely rare. Duplicate records also cost taxpayers money in the form of additional mailings and other election resources.

However, it is technically impossible for anyone outside the Department of State (or its legally designated agents, such as ERIC) to give an accurate number of duplicate voter records in Pennsylvania, because outside analysts do not (and should not) have access to the kinds of sensitive personal information – such as social security or drivers’ license numbers – that would make accurate matching and cleaning possible. Even the data ERIC receives is cryptographically hashed before being transmitted, so it cannot compromise the privacy of Pennsylvania voters.

The definition of a duplicate is clear (one or more extra records for the same real voter), but how you identify a duplicate is very difficult, even for election administrators who have access to more data than the public does. As McDonald and Levitt (2008) point out in their analysis of the “birthdate problem,” it is not only possible but expected that some number of individuals will share the exact same name and birthdate on a list of several million voter records. Removing duplicate records is extremely fraught, because each time you make an error, you risk disenfranchising a lawful voter.

Upgrading the SURE Voter Database would help reduce the number of duplicates

Upgrading the Statewide Uniform Registry of Registered Electors (SURE) voter database to a top-down instead of bottom-up system would make de-duplicating records across counties much easier, and would reduce the likelihood of duplicate records being created in the first place. The SURE system not only includes data on each voter’s registration, but also information about how often they voted and by what method, as well as key information related to voting by mail.

Migrating a voter database, especially in a state as large as Pennsylvania, is a very complicated and expensive project, but it should be a key priority after the next general election. A note of caution: even when migrating to a top-down system, it is technically impractical for a voter database that contains 8.7 million to be completely free of duplicates.

About the Authors

Kyle Miller

Pennsylvania Policy Strategist

Kyle Miller leads Protect Democracy’s advocacy efforts in Pennsylvania, including legislative efforts to prevent election subversion and combat the spread of disinformation.

Quinn Raymond

Voteshield Co-Founder & Policy Strategist

Quinn Raymond manages VoteShield, which he co-founded in 2017. He works closely with election administrators, computer scientists, and data analysts to ensure the integrity of voter registration and absentee ballot data across the nation.

Emily Rodriguez

Communications & Advocacy Strategist

Emily Rodriguez is a communications and advocacy strategist, specializing in election security and litigation, as well as state-level strategic plans.

Jiamin Huang

Data Scientist, VoteShield & Elections

Jiamin Huang is a Data Scientist on the VoteShield team primarily working to identify and address anomalous changes to voter registration records.

Madeline McCann

Data Scientist, VoteShield & Elections

Madeline McCann is a Data Scientist on the VoteShield team focused on analyzing anomalous changes to voter registration records.

Related Content