NewsIn Depth


The other Arizona election audit

Disaffected Republicans handed Arizona to Joe Biden
Posted: 5:40 PM, Jun 16, 2021
Updated: 2021-06-22 11:37:18-04
Trump Biden Donald Trump Joe Biden

PHOENIX — A bipartisan group’s unofficial audit concludes Republican voters, disenchanted with Donald Trump, were key to Joe Biden’s 2020 presidential win in Maricopa County.

Benny White, a Republican who previously ran for Pima County Recorder, teamed up with Democrat Larry Moore, and Independent Tim Halvorsen, two retired executives from the election company Clear Ballot, to conduct their own unofficial audit into the votes cast during the November 2020 election in Maricopa County.

For several weeks, the trio pored over the cast vote record, a database showing the voting results of each of the 2.1 million ballots cast in Maricopa County.

The data, which is public record, can be used to confirm vote tabulations and better understand voting patterns and behavior.

In addition to finding and identifying disaffected Republican voters, the unofficial auditors looked at other claims about the 2020 presidential race in Maricopa County.

These included theories that thousands of fraudulent ballots were flown in from Asia, that non-citizens skirted Arizona election laws to vote for Biden on federal-only ballots, and that an “error rate” led to inappropriate hand adjudication of ballots.

This team’s audit is independent from the Arizona Senate-led election audit, run by Cyber Ninjas, which has taken place over several weeks at Arizona Veterans Memorial Coliseum. The final report from the Senate's audit is expected in about two months.


White, Moore, and Halvorsen said they noticed in their cast vote record analysis that there was the drop-off of Republican support for former President Trump when compared to Democrats for Biden.

The trio of analysts focused on voters who demonstrated a commitment to candidates from a single party. They only considered those ballots where voters cast a ballot a candidate of one party at least 8 of the 15 partisan races that were up for grabs in 2020.

They found that committed Republicans outnumbered the committed Democrats by 13,000 in Maricopa County, a number that should have benefited the Republican presidential candidate.

However, the analysts found there were more disaffected Republicans than Democrats in Maricopa County.

According to White, 59,800 people did not vote for Trump even though they voted mostly for Republicans down-ballot. Most of those renegade Republicans, 39,102 voters crossed over to vote for Biden.

Far fewer Democratic majority voters, 38,851, did not vote for Joe Biden in the 2020 race. White found that 21,679 of them voted for Donald Trump.

Overall Biden won Maricopa County by 45,109 votes.

This helps to explain why down-ballot countywide races were won mainly by Republicans despite losses at the presidential and senate level, according to the unofficial auditors.

The chart below shows the level of partisanship among Maricopa County voters that did not support the top of their ticket.

Had the drop-off rate of committed Republicans matched the lower rate of drop-off Democrats in Maricopa County, Donald Trump would have been in a position to carry Arizona by over 20,000 votes. Trump ended up losing by more than 10,000 votes.

The chart below shows the difference in Maricopa County between Republican and Democratic candidates in countywide races.


White also said Republicans who failed to vote for Trump came from precincts all over the county. However, there were higher concentrations of these disaffected Republicans in areas that have higher levels of college-educated, wealthy voters such as Paradise Valley, north Scottsdale, and the far southeast suburbs.

“You see a lot of these disaffected Republican voters in Arizona are also the same areas where Democrats really gained a lot of ground in the 2016 election compared to 2012, when Mitt Romney was the Republican nominee, and you really had a night-and-day difference between the type of Republican that voters were choosing between,” said Nathaniel Rakich, a political analyst from FiveThirtyEight.

Maps of Maricopa County’s 751 voting precincts from the 2020 election confirm Rakich’s observations.

High-wealth precincts around Paradise Valley and the Arcadia area of Phoenix collectively had some of the highest percentages of committed Republican voters that did not vote for Donald Trump. Palo Cristi, a precinct adjacent to Camelback Mountain, was carried by Trump 55% to Biden’s 44%, but 6.3% of voters who supported most other Republicans on the ballot did not back Trump.

Darker red areas on the interactive map below show where higher numbers of Republican voters, by Maricopa County precinct, did not choose Trump.

To see more interactive maps breaking down voter counts in Maricopa County, click here.

White, Moore, and Halvorsen were so confident in their findings and their processes that they publicly issued a letter to state Senate President Karen Fann. The letter said they could accurately tally the results of any unopened ballot box and compare their tally from the cast vote record with the hand recount.

Senate liaison Ken Bennett told ABC15 that the offer was not accepted, so they released the results of several batches of 200 votes coming from several randomly chosen boxes. Their hope is that the final hand-count report released by the audit will also include results by batch.

According to Moore, those small batch results helps ensure the integrity of any full hand recount, because otherwise it would be next to impossible to determine if the audit count is accurate.


Using Maricopa County’s cast vote record, a massive database of the 2,089,563 tabulated ballots from November, the trio of analysts also decided to fact-check a series of unsubstantiated fraud claims by Donald Trump supporters. Their goal was to dispel as many myths as possible by simply calculating the data.

There were wild rumors that thousands of counterfeit ballots were flown from China to Arizona to help Biden win, which resulted in the Cyber Ninjas team looking for bamboo fibers in ballots during the official audit commissioned by the state Senate.

White, Moore, and Halvorsen, working independently, took a different approach. They first analyzed whether an injection of fraudulent ballots was not caught by county’s “check-in” system, which verifies if a registered voter has already cast a ballot and prevents them from doing it a second time.

White obtained Maricopa County’s ‘voted' file, which is published shortly after an election is concluded, and found that it contained 2,089,756 names of people who voted. While this number is slightly over the 2,089,563 ballots officially canvassed, a county election spokesperson told ABC15 the two numbers often do not add up because of rare occurrences such as a voter’s early ballot envelope being “checked-in” but there is no ballot in the envelope.

“Right off the bat that tells me that there were no 30,000 or 40,000 Korean, Chinese, or Afghan ballots - or whatever - injected because I have a voter for every ballot,” White said.

The database White used included the ballot results and precinct information, but no voter names.


The three retirees then determined that Arizona’s ‘federal-only’ voters did heavily favor Biden, but the total number of federal-only ballots in Maricopa County was not enough to cover the vote spread between Biden and Trump.

“There was a small number of federal-only ballots that really came into play here,” Halvorsen said. “It’s not a significant percentage.”

Arizona has a unique rule that requires two kinds of ballots to be printed. Regular ballots contain all federal, state and local races. Federal-only ballots only have presidential and congressional races.

In 2004, Arizona voters passed Proposition 200, which required proof of citizenship to register to vote. The law was intended to prevent non-citizens and other unqualified people from illegally casting ballots.

In 2013, the United States Supreme Court upheld a ruling by the Ninth Circuit that found Arizona could not preempt federal laws that only require an attestation of citizenship. To address this, former Secretary of State Ken Bennett, now audit liaison for the State Senate, created Arizona’s bifurcated registration system that allowed for people who registered with a federal form to vote in federal races.

“If you want to be a full ballot voter you must provide proof of citizenship,” said Gina Roberts with the Citizens Clean Election Commission. “If you do not provide that proof and the county cannot verify citizenship for you through existing voter registration or MVD records, then you are what is considered a federal-only voter, meaning you can only vote in federal races.”

One claim by some Trump supporters is that federal only voters are likely "illegal immigrants," but that claim is unfounded as no such cases have come to light in Arizona.

The analyst trio found that 8,114 federal-only ballots were cast in Maricopa County, that’s 0.4% of all votes cast.

Biden carried 5,781 of these ballots to Donald Trump’s 2,134. The difference is 3,647 votes. Even if every one of these votes came from non-citizens, it’s still not enough to offset Biden’s 10,000 vote margin of victory statewide.


White, Moore, and Halvorsen also wanted to know if there was a significant number of tabulated ballots that were only marked for the presidential contest.

They, again, found a small number. Across Maricopa County, 3,024 ballots were only marked for Joe Biden. Trump received 3,474 of these. It’s a difference of 450 in favor of Trump.

To read more on data tracing Biden-only and Trump-only ballots in 2020 Arizona election, click here.


Of all the unsubstantiated fraud claims surrounding the 2020 general election in Maricopa County, one of the most enduring is suspicion about an 11% “error” rate.

Some of Donald Trump’s staunchest backers, such as MyPillow CEO Mike Lindell, have cited this as a primary reason to question the former president’s loss to Joe Biden. The claims have been boosted by far right-wing, pro-Trump media organizations such as OANN and Gateway Pundit.

The phrase “error rate” implies that the tabulation equipment made a mistake, which is not the case since election equipment is specifically designed to kick out ballots it cannot scan. Instead, what the "error rate" refers to is the county’s adjudication rate.

“When an early ballot is tabulated and cannot be read for some reason, whether it be damage or stain, it is reviewed by a bipartisan team that looks at each contest for voter intent,” said Kathren Coleman, who was a deputy Maricopa County recorder during Adrian Fontes’s term.

One Republican and one Democrat are always paired together when reviewing voters' choices that require adjudication, according to Coleman. After a decision is made by the bipartisan adjudicators, the original markings and the final outcome are recorded in the cast vote record.

A 2019 Arizona law dropped a requirement that adjudicated ballots had to be reviewed in all contests. Instead, the election management software indicates which race on the ballot was unreadable, and just that race is reviewed.

In these cast vote records, the orange writing on the top image indicates a ballot that has been adjudicated. The image on the bottom shows a non-adjudicated ballot.

In November, 235,392 Maricopa County ballots were adjudicated, which accounts for 11.3% of all ballots cast.

A spokesperson for the Maricopa County Elections Department told ABC15 that percentage is not unusual. The August primary had a similar adjudication rate of 11.1%, although fewer overall ballots were cast.

“The ‘disinformationists’ are pointing to the fact that there were a quarter-of-a-million ballots adjudicated by hand,” said Moore.

What Moore's independent audit team found in the cast vote record was only 11,954 of the adjudicated ballots involved the presidential contest. That’s a small percentage of the total.

Moore also said 55% of the adjudicated presidential votes involved unqualified write-ins. This is when a voter writes in a name of a fictitious character, such as Mickey Mouse, or the name of a real person who did not file the necessary paperwork to officially become a write-in candidate. Those votes don’t count.

When the remainder of the presidential adjudicated ballots were reviewed, 2,069 were Biden votes and 1,516 were Trump votes, a difference of 553 in favor of Biden.

Biden won Maricopa County by 45,109 total votes.

The remaining 223,438 adjudicated ballots involved other contests. The cast vote record showed no changes made at the top of the ticket.

Halvorsen, Moore, and White hope their analysis will put an end to the fraud claims involving adjudicated ballots.

They also hope the final report from the Cyber Ninjas’ audit at the coliseum will include a cast vote record review and address the topic of adjudicated ballots. That report is expected to be sent to Arizona Senate President Karen Fann in about 6 weeks.

The map below shows how many ballots were electronically adjudicated in the Presidential race. A lower number indicates there were more Trump ballots adjudicated than Biden. Precincts with no adjudicated ballots have been removed.

White, Moore, and Halvorson believe the methods they are using to analyze the election are better than what is being done at the Coliseum because they said hand counts can be prone to human error.

Neither of these auditing group’s results will make a difference to the outcome of the 2020 presidential election. November’s results were canvassed last fall, and Congress approved the Electoral College totals in January.