A misleading tweet uses a faulty interpretation of exit polling in Massachusetts and incomplete reporting on two other races to claim there were irregularities in Super Tuesday elections.
But its main claim is bogus.
It says, “In Massachusetts, Biden’s vote exceeded exit polling by 15%. That’s statistically a huge red flag that fraud occurred.”
The tweet originated on an account that frequently posts in support of Sen. Bernie Sanders, who is currently in second place in the delegate count behind Biden for the Democratic presidential nomination. Asked what the claim was based on, Lisa Pease, who runs the account, sent us a link to a website called TDMS Research.
That site had posted a story on the outcome of the Massachusetts primary that included a table comparing exit poll numbers with the reported vote count for each candidate. Importantly, it used exit poll numbers from 8 p.m. on the day of the primary, according to the table. Those numbers were updated two hours later, though, according to CNN, the outlet cited in the table.
Theodore de Macedo Soares, who runs the site and wrote the post, told us in an email that he intentionally used the numbers from when the polls were closing instead of the final polling numbers.
After the polls close, Edison Research, which conducts the exit polling, uses the election results from the voting precincts it is sampling as one of the factors that informs its forecast.
But thinking that interim exit poll results is a better indicator of voter preference is misguided, said Daron Shaw, a professor at the University of Texas at Austin who has worked on political campaigns and polling.
Exit polls are weighted throughout the day, said Shaw, not just at the end. He called the analysis done by TDMS Research “misleading at best and corrosive at worst.”
Largely speaking, exit polls aren’t meant to validate election results, said Patrick Murray, director of Monmouth University Polling. “Exit polls are not designed to be a check on the vote outcome,” he said. “Period.”
Joe Lenski, executive vice president of Edison Research, described the polling process to the Washington Post in 2016, characterizing the intended use of the results this way: “There are two important uses of the exit poll. One is to project a winner. But the main use of the exit poll that night and historically is to have the most accurate representation of the demographics of voters.”
So, the exit poll numbers used in TDMS Research’s post are misleading. But that’s not all. The way it calculated the difference between those numbers and the election outcome is misleading, too.
TDMS Research included in its table a column labeled, “GAIN/LOSS FROM EXIT POLL,” and it reported that figure to be 15.7% for Biden. That appears to be where Pease got the idea that “Biden’s vote exceeded exit polling by 15%.”
But it’s a misleading calculation of misleading numbers.
It shows the percentage increase from Biden’s share of support in the exit polls at 8 p.m., which was 28.9%, according to TDMS Research, to his share of the final vote total, which was 33.4%, according to TDMS Research.
Calculating the proportional difference between the early exit poll and the final vote yields, roughly, a 15% increase.
Asked why he chose that method for showing the difference, de Macedo Soares told us by email, “Well that’s just a common way of looking at changes of any kind. Say, someone is making $50,000/yr. and then gets a raise of $10,000 she may say ‘wow, that’s pretty good, a 20% raise of my salary! If the person received only a $2,000 raise he may say ‘bummer, that’s only a measly 4% raise.'”
But election experts we talked to didn’t think this was a “common way” of analyzing election results. They said the opposite.
“That is not informative in any way, unless you want to show a bigger number,” said James McCann, a political science professor at Purdue University. He’s never seen a relative difference like this used to analyze election outcomes. “This is needlessly inflammatory,” he said.
William Mayer, a political science professor at Northeastern University, agreed. “Calculating gain/loss as the author does has the potential of making relatively small discrepancies seem much larger,” he wrote in an email to us. “I don’t know of anybody who analyzes poll numbers this way.”
Pease’s post also makes claims about the results of the primaries in Texas and California, but those were more vague. She wrote, “In Texas, a large number of ballots were found that hadn’t been counted. In Los Angeles, which is heavily Bernie, our mail-in ballots have not been counted.”
In Texas, she was referring to a situation in Dallas, where the county election administrator discovered that ballots from 44 machines had been missed and petitioned the court for a recount. The recount didn’t change the outcome of any races. Biden won the county with 41% of the vote, Sanders came in second with 27%, and Michael Bloomberg finished third with 14%. Statewide, Biden received 35%, Sanders 30%, and Bloomberg 14%.
As for the claim about California, that state allows a vote-by-mail option, which means it can take longer to tabulate the results. The secretary of state is due to certify the results of the election by April 10.
A recent poll conducted by the Los Angeles Times and the UC Berkeley Institute of Governmental Studies found that 64% of voters surveyed preferred having more ways to cast a ballot even if it takes longer to get final results. As of March 18, with 1.4 million votes counted, Sanders was winning Los Angeles County with 39% of the vote and Biden was in second place with 29%. The secretary of state’s office estimated that there were 219,500 votes remaining to be counted.
de Macedo Soares, Theodore. “MASSACHUSETTS 2020 DEMOCRATIC PARTY PRIMARY Exit Poll Versus Reported Vote Count.” TDMSresearch.com. 4 Mar 2020.
Massachusetts Exit Polls. CNN. Accessed 11 Mar 2020.
Massachusetts Election Statistics. 2020 Presidential Primary. Secretary of the Commonwealth of Massachusetts. Accessed 11 Mar 2020.
Shaw, Daron. Professor, University of Texas at Austin. Telephone interview 16 Mar 2020.
McCann, James. Political science professor, Purdue University. Telephone interview. 13 Mar 2020.
Murray, Patrick. Director, Monmouth University Polling. Telephone interview. 13 Mar 2020.
Mayer, William. Professor, Northeastern University. Email to FactCheck.org. 13 Mar 2020.
Ura, Alexa. “Dallas County recount turns up 9,149 new ballots but doesn’t swing any races.” Texas Tribune. 11 Mar 2020.
Myers, John. “Taking a long time to count California votes is OK with voters, poll finds.” Los Angeles Times. 3 Mar 2020.