Skip to main content
Politics

Polling error reached a record high in 2020. Experts point to a Trump factor.

Share

From a voter’s perspective, polling in the days leading up to elections may seem significantly flawed as of late. Experts say the numbers support this hypothesis. During the 2016 and 2020 presidential races, state-level polling was inaccurate, according to the American Association for Public Opinion Research. National polling also featured greater error than usual.

In fact, according to the AAPOR evaluation of 2020 general election polls, error “was the highest in 40 years for the national popular vote and the highest in at least 20 years for state-level estimates of the vote in presidential, senatorial, and gubernatorial contests.”

Courtney Kennedy, the Pew Research Center Director of Survey Research, said there are several reasons for polling error, but one factor was unique to these two races: former President Donald Trump.

“Trump provided explicit cues to his supporters that polls were ‘fake’ and intended to suppress votes,” the AAPOR report said. “These statements by Trump could have transformed survey participation into a political act whereby his strongest supporters chose not to respond to polls.”

PRE-ELECTION POLLS FAILED US IN THE LAST TWO PRESIDENTIAL RACES.

LET ME CLARIFY. IN 2016, POLL RESULTS FOR STATES WERE INACCURATE.

IN 2020? POLLS SHOWED A DISTORTED REALITY BOTH IN INDIVIDUAL STATES—AND THE  COUNTRY AS A WHOLE.

SO THE QUESTION IS: ARE POLLS GETTING WORSE AND WORSE? THAT IS, ARE THEY BECOMING LESS ACCURATE?

LET’S GET THIS STRAIGHT.

THINKING BACK TO 2016, IT’S HARD TO FORGET ALL THE OUTLETS PROJECTING A HILLARY CLINTON VICTORY.

SPOILER ALERT: THAT DIDN’T HAPPEN.

BUT THE NATIONAL POLLS WERE STILL CORRECT IN PREDICTING CLINTON WOULD WIN THE POPULAR VOTE. SHE SIMPLY LOST IN THE ELECTORAL COLLEGE.

WHICH MEANS ONLY STATE POLLS WERE REALLY WRONG THAT YEAR.

BUT THEN, IN 2020, POLLING WAS OFF AT BOTH THE NATIONAL AND STATE LEVEL BY GENERALLY OVERESTIMATING DEMOCRATS’ ADVANTAGE.

ACCORDING TO THE AMERICAN ASSOCIATION FOR PUBLIC OPINION RESEARCH EVALUATION OF 2020 GENERAL ELECTION POLLS, ERROR “WAS THE HIGHEST IN 40 YEARS FOR THE NATIONAL POPULAR VOTE AND THE HIGHEST IN AT LEAST 20 YEARS FOR STATE-LEVEL ESTIMATES OF THE VOTE IN PRESIDENTIAL, SENATORIAL, AND GUBERNATORIAL CONTESTS.”

THE STATE WITH THE GREATEST ERROR IN THE LAST TWO WEEKS PRE-ELECTION? NORTH DAKOTA.

THE AAPOR REPORT SAYS POLLS WERE OFF BY 14 POINTS.

COURTNEY KENNEDY, THE PEW RESEARCH CENTER DIRECTOR OF SURVEY RESEARCH, HAS SOME INSIGHT.

COURTNEY KENNEDY: “IF THE QUESTION IS, IS THE TRAJECTORY SUCH THAT POLLS ARE ONLY GOING TO GET WORSE AND WORSE, I DO NOT NECESSARILY THINK THAT THAT’S THE CASE. ONE, YOU KNOW, BIG QUESTION FOR THE POLLING COMMUNITY, IS WHETHER THE 2016 AND 2020 ERRORS WERE LARGELY, YOU KNOW, KIND OF STEMMING FROM DONALD TRUMP BEING ON THE BALLOT.”

LONGWORTH: NOW, THAT’S NOT TO SAY THE POLLING ERRORS ARE THE FORMER PRESIDENT’S FAULT.

IT’S JUST THAT THERE MIGHT BE A RELATIONSHIP BETWEEN HIS CANDIDACY AND POLLING RESPONSES.

KENNEDY: “HE WOULD, YOU KNOW, REPEAT THAT MESSAGE THAT A LOT OF POLLS ARE FAKE, AND HOAXES, AND THAT KIND OF THING. AND SO, YOU KNOW, IT’S FAIR TO WONDER IF THAT MESSAGE FROM THAT PARTICULAR CANDIDATE MADE HIS SUPPORTERS DISTRUSTFUL OF POLLS, MAYBE ANTAGONISTIC TOWARDS THE POLLING INDUSTRY, AND NOT PARTICIPATE IN POLLING.”

LONGWORTH: IT’S A THEORY, BUT THEORIES ARE ALL WE TRULY HAVE, CONSIDERING THE AAPOR REPORT SAYS, IDENTIFYING CONCLUSIVELY WHY POLLS WERE SO INACCURATE IN 2020 “APPEARS TO BE IMPOSSIBLE WITH THE AVAILABLE DATA.”

KENNEDY: “YOU KNOW, WE RECENTLY HAD AN ELECTION LIKE THE GUBERNATORIAL RACE IN VIRGINIA, POLLS WERE PRETTY DARN GOOD. AND DONALD TRUMP WAS NOT ON THE BALLOT, HE WAS ALSO NOT ON THE BALLOT, DIRECTLY IN THE 2018 MIDTERM ELECTION, AND IN THE POLLS WERE REASONABLY GOOD THERE AS WELL.”

LONGWORTH: SO, THAT TRUMP FACTOR IS WHEN PEOPLE WON’T PARTICIPATE IN POLLS—PER THE FORMER PRESIDENT’S URGING.

AND WHEN YOU TAKE HIM OUT OF THE EQUATION, YOU’RE LEFT WITH THE USUAL OPPORTUNITY FOR ERROR THAT COMES FROM SAMPLING, MEASUREMENT, COVERAGE AND NON-RESPONSE ISSUES.

AND, AT THE END OF THE DAY, KENNEDY SAYS PEOPLE REALLY NEED TO KEEP IN MIND THAT, WHILE POLLS ARE PRETTY PRECISE FOR GAGING HOW PEOPLE FEEL ABOUT ISSUES.

KENNEDY: “THEY’RE JUST NOT PRECISE ENOUGH TO CALL AN ELECTION.”

LONGWORTH: AND THAT’S WHY PEW SAYS THERE’S GENERALLY ABOUT A 3-POINT ACCEPTED MARGIN OF ERROR.

GOT IT STRAIGHT?

LET ME KNOW IN THE COMMENTS AND MAKE SURE YOU HIT THE STRAIGHT ARROW BIAS METER.