US Elections: An Embarrassing Failure for Pollsters

OPINION: Electoral polls in the US face yet another reckoning after their uneven performance at best in this year’s vote.

Although the outcome of the 2020 presidential race remained uncertain the following day, it was clear that the polls collectively faltered, overall, providing Americans with clear indications of how the election would turn out.

And that misstep promises to resonate in the field of survey research, which was battered four years ago when Donald Trump led states like Wisconsin, Michigan and Pennsylvania where polls indicated he had almost no chance of winning. Prominent survey-based statistical forecasts also deviated from the target in 2016.

Those failures deepened the shame of a field that has suffered, but has survived, a variety of lapses and surprises since the mid-1930s. Many of those mistakes and failures are described in my latest book, Lost in a Gallup: Poll Failure in the US Presidential Election..

READ MORE:
* Changes, Signs, and Surprises: What to Watch Out For as the US Election Unfolds
* Election in the USA: What it is like to lose a presidential election
* 2020 US Elections: Seven Reasons This Year’s Election Is Not A Replay Of 2016

Criticism was intense in some quarters on Wednesday. Politico’s widely followed “Playbook” newsletter was remarkably scathing. “The survey industry is a disaster,” he declared, “and it should explode.”

Many surprises

While that assessment seems extreme, especially given the resilience of polls over the decades, the poll-driven expectation that former Vice President Joe Biden would lead the Democrats on a radical “blue wave” was not met. Biden may still win the presidency, but it won’t be crushing.

Biden’s overall poll lead, compiled by RealClearPolitics.com, stood at 7.2 percentage points on the morning of Election Day. Just over 24 hours later, his lead in the national popular vote was nearly 3 percentage points.

Pollsters often seek comfort and protection from critics by stating that pre-election polls are not predictions. But the closer they are to the election, the more reliable the polls should be. And a series of individual pre-election polls were embarrassingly out of place.

A notable example was the final Washington Post / ABC News poll in Wisconsin, published last week, which gave Biden a surprising 17-point lead. The result there was still undecided on Wednesday morning, but the margin will surely not come close to 17 points.

Indeed, the surprises in the polls were many and included Senate races like those in Maine, where Republican Susan Collins appears to have fought off a well-funded challenger to win a fifth term, and South Carolina, where Republican Lindsey Graham won easily. election despite polls indicating a much closer race. Graham stated after his victory became clear, “To all the pollsters, they have no idea what they are doing.”

It appears that Republicans will retain control of the US Senate despite expectations, prompted by polls, that control of the upper house will likely shift to Democrats.

Voting problems are not new

The 2020 U.S. elections may represent another chapter in the controversies that have periodically surrounded electoral polls since George Gallup, Elmo Roper, and Archibald Crossley began their sample polls during the 1936 presidential campaign. The most dramatic electoral failure in the presidential election America’s election came in 1948, when President Harry S. Truman challenged pollsters, pundits, and the press to win re-election over highly favored Republican candidate Thomas E. Dewey.

This year’s surprise is nothing like the epic failure of the polls of 1948. But it is surprising to see how the errors in the polls are so varied and almost never the same, as Leo Tolstoy said about unhappy families: each one “is unhappy in his own way. “

The factors that led to this year’s embarrassment may not be clear for weeks or months, but it’s no secret that electoral polls have faced several difficult challenges to solve. Among them is the decrease in response rates to telephone surveys carried out by operators using random dialing techniques.

This technique used to be considered the gold standard of survey research. But response rates to telephone surveys have been in decline for years, forcing survey organizations to seek out and experiment with other sampling methods, including Internet-based techniques. But none of them have become the new gold standard in polls.

One of the most notable innovators of the polls was Warren Mitofsky, who years ago reminded his counterparts that there is “a lot of room for humility in the polls. Every time you get arrogant, you lose. “

Mitofsky died in 2006. His advice rings true today.

W. Joseph Campbell is Professor of Communication Studies at the American University School of Communication.

This article originally appeared on The Conversation. Read the original article