US pollsters come under fire after recent missteps

TO Ulysses BellierAFP

The US political world was stunned in 2016 when Donald Trump won despite pre-election polls showing him trailing, and in 2020 Joe Biden’s lead was smaller than polls had predicted.

Have pollsters learned enough from their mistakes to be more accurate this year?

just days leftDemocrat Kamala Harris and Trump appear to be standing shoulder to shoulder. But if polls again underestimate the Republican vote, the former president could well be ahead.

Experts say the core of the problem has not changed since Trump’s dramatic disappointment in 2016, when part of his electorate refused to respond to polls.

“We haven’t found a silver bullet,” Courtney Kennedy, vice president for methods and innovation at the Pew Research Center, told AFP.

In the meantime, each polling firm is taking its own steps to try to correct the problem.

Calling Trump’s Friend

In 2020, many people, when contacted by phone, “would just scream at us, ‘Trump!’ and hang up,” said Don Levy, director of the Siena College Research Institute, which conducts highly regarded public opinion polls. New York Times.

In an effort to better account for such voters, Siena now records responses from people who hang up, even if they don’t answer all their questions.

Levy said he also tries several times to contact people who don’t answer the phone the first time.

“If we call them a third or fourth time, we’ll get more of them, which means more callbacks, more potential Trump voters in the sample.”

Pew, on the other hand, now allows respondents to respond online or by phone, Kennedy said.

“We have very different people participating in different ways,” she said.

Data Weighting

Once the answers are collected, social scientists must decide how to weigh the data.

If a particular group is underrepresented in the data relative to its share of the overall population (such as rural Republicans), these responses are usually given more weight in the final calculations.

Levy said the New York Times/Siena poll weights based on what they think the 2024 electorate will look like, using previous elections as a guide, as well as a number of other factors.

Joshua Clinton, a political science professor at Vanderbilt University, expressed caution about weighing based solely on previous elections.

“Suppose there is more enthusiasm among Democrats than among Republicans… then the increase for 2020 will actually reduce support among Democrats in your poll,” he told AFP.

It’s a “no-win situation,” Clinton said, because pollsters can only know whether their weights are correct until actual voter data comes in.

“We’re targeting a universe that doesn’t exist yet,” Levy said.

‘Who knows?’

With so much focus on Trump voters who may not be showing up in the polls, Clinton warns that, on the other hand, Democratic support may also be underestimated.

After 2016 and 2020, “you might be tempted to conclude that polls will always underestimate Republicans … but that’s not true,” he said.

In Michigan, a key state in this year’s presidential election, polls underestimated Democratic support in the 2022 midterm elections, he said.

“Who knows what will happen in 2024?”

Siena College’s Levy also raised the possibility of a “shy Harris voter,” someone who is reluctant to publicly reveal their support for the vice president, perhaps because they live among die-hard Trump supporters.

When asked whether polls might be too biased toward absentee Trump supporters, Pew’s Kennedy expressed skepticism.

“I’ve seen enough data that it’s clear that polls are having a really hard time reaching enough Trump supporters to really do the grunt work of getting it right,” she said.

Because the competition is so tight and within the margin of error, “we should really just tell ourselves… it’s all the same and not pretend we can get it to the decimal place,” she said.

“It’s just a fantasy land.”

– AFP