NC has a lot of close 2024 races. But are those polls accurate? | Opinion

It’s said that the only poll that matters is the one on Election Day, but there will still be intense interest over the next eight months in what polls show in tightly contested North Carolina.

But while poll results make headlines, there are questions about their reliability as the decline in telephone landlines and the rise of caller ID have made it harder to reach a true random sampling of likely voters.

“A lot of pollsters are trying to find ways to get a more reliable sample because that is the secret to polling,” said David McLennan, a political scientist who directs the Meredith Poll, a leading survey of North Carolina.

Random dialing was once the gold standard for an accurate sampling, but pollsters have mostly given up on that approach as fewer people have listed numbers and more decline to answer calls from an unknown number.

McLennan said it would once take six calls to reach a person who would answer a survey. Now it takes 125. “It’s really not very productive and you tend to get a lot of older people, more rural people,” he said. “You don’t get a lot of young people.”

Instead of random calls, pollsters now are turning to large groups of people who have agreed to be surveyed. Martin Kifer, director of High Point University Survey Research Center’s HPU Poll., is one of those pollsters.

The HPU poll shifted to surveying selected groups a few years ago. For this year’s primary, the poll surveyed more than 700 people who were likely to vote or had already voted early.

“It’s tough,” Kifer said. ”We’re trying our best to find a breakdown of respondents who are going to reflect as closely as possible North Carolina.”

Brooks Anderson, an operations manager with Eastcoast Research, a North Carolina-based firm that develops survey panels, said polling a selected group can reflect popular opinion, but it loses “a level of randomness that really helps to drive the credibility higher.”

Pollsters and those who study them say polls face an array of challenges to their accuracy. Those include computer ‘bots” filling out surveys and a reluctance of some respondents to say they will vote for former President Donald Trump.

Polls can predict outcomes with surprising accuracy, as some did in the March 5 North Carolina primary, other results are puzzling. A recent WRAL-TV poll, for instance, found Trump favored by 52% of North Carolina voters ages 18 to 34. But in 2020, North Carolina voters ages 18 to 29 favored Biden over Trump by 57% to 40%.

Not only do poll results vary widely, so do opinions about the polls themselves.

Steven Greene, an N.C. State political science professor who studies public opinion, said polls are remarkably accurate despite the latest obstacles. He said polls are working “about as well as ever, which is pretty astounding when you look at the changes in technology and the behavior of people.”

W. Joseph Campbell, a professor emeritus of communication at American University in Washington, D.C. and the author of Lost in a Gallup: Polling Failure in U.S. Presidential Elections, said polls are struggling with accuracy. He noted that five major polls predicted Biden would win in 2020 by 10 or more points. Biden won by 4.5 points.

”The polls are in trouble in presidential elections,” he said.

One consequence of inaccurate polling, Campbell said, is that errant assessments skew the coverage of elections.

Campbell added, “Polls don’t go wrong the same every time, but they go wrong often enough for us to be wary of them. We should keep in mind what they are saying, but we can’t treat them as prophecies.”

Polls feed the horse race aspect of elections, but they may not closely call a race, particularly in down-ballot races in which many voters don’t know the candidates. And all polls are speculative when conducted long before an election. Much can happen and opinions can change.

As this year’s heated presidential, gubernatorial and other races get going in North Carolina, voters are better off listening to what candidates are saying instead of what early polls say voters are thinking.