In politics, we’re all ‘poll dancers’

Bill Knight column for 3-5, 6 or 7, 2020                   

After Super Tuesday’s primaries in 14 states, Democrats abroad and the American Samoa territory, some may think it’s all over but the cheating.

One wonders, however, whether innumerable polls thus far have made a difference – or will in the next eight months.

In 1984, remember, polls said Walter Mondale would defeat Ronald Reagan, but the Minnesota Democrat lost 49 states. And in 2008, Barack Obama was forecast to win the New Hampshire primary, where Hillary Clinton won.

Meanwhile, before this year’s early primaries, polls said Joe Biden led Democratic candidates with 26%, followed by Bernie Sanders’ 21, Elizabeth Warren’s 15 and Pete Buttigieg’s 6. That was followed by a Quinnipiac polling of head-to-head showdowns between Democratic candidates and President Trump that showed Mike Bloomberg beating Trump 51-42, barely ahead of Sanders’ 51-43 and the rest far behind.

What the…?

Polls aspire to measure public opinion but they’re mere snapshots of slices of time, and it’s a long time until some Democrat attains 1,991 pledged convention delegates (or 2,375.5 delegates if a second ballot is needed – when party leaders can put their thumbs on the scale), much less November 3.

“There is still a lot of good polling in news media reports,” said the late Andrew Kohut of Pew Research Center, “but its value is diminished by trends.

“Polling is being used by the press less as a check on the conventional wisdom about voter opinion, and more as a way of underscoring the running story line with numbers,” he continued. “Polls, and some pollsters, have become part of the tabloidization of political reporting. And more and more polls are being conducted on a judgmental rather than on a systematic basis.”

Why should polls have such an outsized influence on the Republic? And why hasn’t the press abandoned the horse-race approach to offer voters substance on issues and character?

“News is no longer what the candidates are doing, but what other people think of candidates and the candidates’ reactions to the polls,” commented Bloomsburg University professor and columnist Walt Brasch.  “The media then devote more of their news coverage to people who are ahead in the polls.”

Nate Silver, whose FiveThirtyEight predictions are widely respected, criticizes lousy media coverage and bad analysis. Silver focused on combining various solid pollsters to more accurately forecast the 2008 and 2012 elections. (Silver concedes he underestimated Donald Trump in 2016, but he still gave him a 29% chance, much more than most prognosticators. Silver was on the mark for the 2018 mid-term elections.)

He complains that most media anointed Biden as front-runner and recognized Kamala Harris as a formidable challenger, before his disappointing performances and her leaving the race.

“There’s a bias toward believing in explanations that involve secrecy or things happening that are hidden,” Silver said in The Atlantic this month.

Also, some polling comes from partisan forces: candidates or consultants, industry or other interest groups, etc., making results biased by what questions are asked and who’s asked.

“The polling agencies with the least interest in the outcome of the election always perform the best,” said Virginia Miori, of the Business Intelligence & Analytics program at Saint Joseph’s University in Philadelphia.

Further, good polls can be pricey and face new problems, such as fewer land lines and more robocalls that people ignore, plus difficulties in detecting late surges at the grassroots and a “social desirability” tendency for respondents to say what they think is socially acceptable instead of their real opinion.

“Consider blue-collar workers versus white-collar workers,” Miori added. “They’re available at different times of the day.”

Pew recently reported that pols’ response rates have gone from 36% in 1997 to 6% today.

Still, there are reliable, qualified pollsters, including Associated Press/GfK, Gallup, Harris, Pew and Roper, plus the pollingreport.com compilation.

Meanwhile, here are 10 questions voters (and reporters) should ask: Who conducted the poll? Who paid for it? How was it conducted? Were participants chosen randomly or another way? Who answered the questions? How many responded? Were they compensated? What questions were asked? Can conclusions be generalized to the whole population? What’s the margin of error?

Answers could help inform voters avoid an error at the ballot box.

Recommended reading: “Polling and the Public: What Every Citizen Should Know,” by Herbert Asher; “The Voter's Guide to Election Polls” by Paul Lavrakas and Michael Traugott; and “A Mathematician Reads the Newspaper” by John Allen Paulos.