Voters fill out their ballots on January 23, 2024 in Loudoun, New Hampshire. Florida Gov. Ron DeSantis dropped out of the race two days ago as Republican presidential candidates former President Donald Trump and former United Nations Ambassador Nikki Haley are competing in the nation’s first primary.
Thassos Katopodis | Getty Images
Lonna Atkeson remembers 2014 being the first year she received hate mail.
Atkeson is a political scientist who studies election surveys and public opinion and has been conducting voter polls since 2004.
But she said a decade into her polling days, the angry messages started rolling in.
“I started getting letters from people saying, ‘You are part of the problem. You are not part of the solution. I will no longer answer your surveys. You are an evil scholar trying to brainwash our children. ‘ ” Atkeson recalled in an interview with CNBC.
For Atkeson, the notes signal a shift: A more polarized electorate has begun to lose faith in institutions like polls, and voters may no longer be willing to talk to her.
At the same time, technology has advanced, and landlines or mail are no longer foolproof ways to contact survey respondents.
“People are not answering the phone,” Rachael Cobb, a political science professor at Suffolk University, told CNBC. “Even in the past 10 years, you might have tried 20 callers to get the information you needed. Now, it’s doubly so: It takes 40 callers to get the information you need, so each poll takes longer and costs more.
Pollsters cite polarization and technology as among the obstacles complicating the task of accurate voter surveys.
As a result, pollsters have made some major mistakes over the past few election cycles.
“If you look at some of the big mistakes, I mean, they were pretty big,” Atkeson said.
voting blind spot
The 2016 presidential election scarred the polling industry, and one of its major missteps was Different kinds headline News scattered this messageClaiming that Democratic candidate Hillary Clinton has about a 90% chance of defeating Republican candidate Donald Trump.
An industry-wide postmortem identified several key reasons for the 2016 poll failure.
Some factors are beyond the control of pollsters.
For example, according to American Association for Public Opinion ResearchSome voters waited until the last minute to decide whose names to put on their ballots, making them difficult to count.
Because of Trump’s controversial comments during the 2016 campaign, some voters were shy about expressing support for him. As a result, they don’t always admit their voting intentions to pollsters.
But other factors are a direct result of methodological oversight.
“People don’t think about educational representation,” said Matin Mirramezani, chief operating officer of Generation Lab, a polling group that specializes in young voters. “Education is the lesson of 2016.”
A large portion of Trump’s supporters are non-college-educated white voters, who were underrepresented in 2016 polls, in part because those with higher education were “more likely to vote” than those with less education. ” responded to the investigation. AAPOR.
Despite identifying these problems, as the 2020 election approached, the poll produced the highest error rate in 40 years, once again underestimating Trump’s support, Asia Pacific Research Association established.
and during 2022 midterm electionsThe media believed a “red wave” of voters would overwhelmingly push Republicans back into control of Congress, but that wave never came. Democrats retained their Senate majority and narrowly ceded the House of Representatives.
2024 course correction
Heading into the 2024 rematch between Trump and President Joe Biden, pollsters are trying a variety of strategies to avoid repeating history and accurately capture the elusive Trump vote.
First, pollsters tweaked their “weighting” method, which assigns a multiplier to each respondent to change how much their answers affect the overall poll results.
Pollsters have long used weighting to build survey samples that accurately reflect voters’ gender, age, race or income. But after 2016, they paid special attention to weight education.
Atkeson recommended that pollsters consider not only the weight of education in 2024 but also variables such as how someone voted in 2020 or even whether they rented or owned a home or whether they were a blood donor.
“You just start labeling everything you can,” Atkeson said. “Anyone who can tell us, ‘Well, what does the population actually look like?'”
In addition to weighting, pollsters are also paying more attention to survey respondents they once underestimated.
“Some people will start a poll, and they’ll tell you who they’re voting for, and then they say, ‘I’m done. I don’t want to talk to you anymore. Goodbye,'” Don Levy, for The New York Times The director of the Siena College Institute that conducted the poll told CNBC. “In 2020 and 2022, we are not counting these people.”
But this time, Levy said they are counting “churn.”
Levy said they found that if they counted impatient respondents in 2020 and 2022, their poll results would “shift about a quarter of a point in Trump’s direction,” eliminating about 40% of the error.
Levy added that SCRI has also taken additional steps to target Trump voters, modeling the sample to give higher survey quotas to those deemed “more likely to be Trump voters in rural areas.”
“If you think of them as M&M’s, then we could say the Trump M&M’s vote is red,” Levy said. “We also have some extra red chocolate chips in the jar.”
About The Author
