Retaining election records from the 2020 election

Did the number of people who voted in the November 3, 2020 general election equal the number of ballots cast?  A recent research study by the America First Policy Institute (AFPI) says the records that would answer the question do not exist.

Records are supposed to be maintained to determine how many people have voted in an election and how many ballots have been cast.  Section 20701 of the Civil Rights Act of 1960 provides that election officials have to keep all records of federal primary, special, or general elections for 22 months after the date of the election.  Any willful violation can be punished by a fine of up to $1,000 or imprisonment for one year, or both.  Section 20702 provides similar penalties for the willful theft, destruction, concealment, mutilation, or alteration of any such record.  The purpose of these sections is to make it more difficult to hide election fraud.

How good of a job are election officials doing in retaining these records?

In 2021, the Voter Reference Foundation (VRF) began a study "to analyze the difference between total ballots cast in various states in the 2020 general election and the total number of voters who cast ballots in that election.  Listed below are the discrepancies between the number of ballots cast and voters found from VRF's data from secretaries of state, which were obtained through public records requests.  For the states marked with 'precinct-level data,' this indicates the statistics AFPI obtained from our public records requests to the precinct level, which the VRF did not have at the time of this review."

State | Discrepancy Between the Number of Ballots Cast and the Number of Voters ̶

Alaska | 3,326

Connecticut | 37,256

Colorado | 439

Florida | 158,319

Georgia | 52,703 Precinct level data

Idaho | 11,147

Michigan | 280,605

Montana | 1,896

Minnesota | 48,328

Nevada | 14,738 Precinct level data

New Jersey | 56,563

New Mexico | 3,844

North Carolina | 21,04

Ohio | 55,330 Precinct level data

Pennsylvania | 91,086 Precinct level data

Virginia | 63,607

There are some precincts where the total number of ballots cast is greater than the total number of people who actually voted, and other precincts where the total number of voters recorded as voting is greater than the number of votes. There could be one precinct with 100 more votes than voters and another in the same county with 100 more voters than votes. If they were lumped together at the county level, it would look like there were an equal number of voters and votes, but that would hide the problem. Thus, it appears the county-level data underestimates vote discrepancies. The findings for Pennsylvania revealed that there were three precincts where the number of ballots cast in the November 2020 general election was more than double the number of voters recorded as voting. In those precincts, there was 1,985 more votes than voters recorded as voting.

Because the information received from the respective secretaries of state in the VFR study was so inadequate, the AFPI decided to conduct a more thorough investigation beginning in March 2022.

For subsequent analyses, three main datasets were used: the state dataset we received from the Secretary of State (SoS)/state's election authority, the county datasets, and the canvasses.

Public record requests to county officials occurred over several months, beginning in March 2022. This examination sought to obtain time-stamped voter record data from the top 100 most populated counties of the 14 swing states: Arizona, Colorado, Florida, Georgia, Iowa, Michigan, Minnesota, North Carolina, New Hampshire, New Mexico, Nevada, Pennsylvania, Virginia, and Wisconsin.

Incredibly, only six of these top 100 county election officers responded with data of voters who voted in the November 3, 2020 general election without disclaimers (meaning some claim that the data were not exactly retained).  Nonetheless, discrepancies still existed for these counties.

Miami-Dade County, FL: 1.6% (12% of the precincts are missing)

Orange County, FL: 3.82%

Cobb County, GA: 8.8% (Secretary of State data 0.68%)

Woodbury County, IA: 3.06%

Buncombe County, NC: 0.14%

Johnston County, NC: 0.07%

For Miami-Dade, Florida, for just the available precincts, the discrepancy between the number of registered voters who voted and the number of ballots cast is 1.6%. This translates into a difference of 16,617 votes. To give perspective on the size of that gap, in 2018, Senator Rick Scott won Florida's U.S. Senate seat by 10,033 votes. Nikki Fried became Florida's Agriculture Commissioner by just 6,753 votes. It is more than half of Governor Ron DeSantis' victory margin in the governor's race of 32,463 votes. Cobb County, Georgia, had an 8.8% discrepancy, amounting to 34,893 votes. That gap, in a county that President Joe Biden carried by 14 percentage points, was three times President Biden's winning margin in Georgia in 2020.

For the other 94 counties, the report "confirm[s] that virtually no one (only 6% of county election officials and two [Secretaries of State] of the states and counties reviewed) on the state or county level are correctly retaining the data they are required to preserve regarding voter rolls on the day of each election."

In the vast majority of cases, the information necessary to compare the number of ballots cast with the number of people who voted does not exist. In other cases, when that data is available, the numbers don't match. That is not to say there was incompetence, voter fraud, or stolen elections. It is a question of transparency and accurate records.

The study makes the following recommendation:

Transparency is essential if we are to restore confidence, considering there are roughly 29% of Americans who did not believe the proper winner was declared in either the 2016 or 2020 general elections and that only 59% of Americans feel confident that their votes will be accurately counted. Having these federally (and in some cases state) mandated records retained and timestamped, preferably digitally, would significantly reduce doubt in our democratic process and immediately restore confidence. This should be easily achievable in the digital age, and Americans should expect this sort of retention to happen. It's not only a civil rights issue but a transparency issue that has eroded confidence in our electoral system.

Image: Tom Arthur via Wikimedia Commons, CC BY 2.0.

If you experience technical problems, please write to