Pollsters to examine 2020 predictions

So, what happened to that widely predicted Biden landslide, that Democratic takeover of the United States Senate, the burgeoning majority in the U. S. House of Representatives?

It didn’t happen. Well, it’s likely that Trump supporters in blue states like Connecticut didn’t want to come out and play with pollsters when they called.

And nationally, the big turnout for Joe Biden was nearly equaled by enthusiastic Trump supporters who were willing to give him four more years, and remain shell- shocked that he lost.

The week before the election, the survey released by the Sacred Heart University Instuitute for Public Policy showed Biden winning Connecticut 51.4 percent to President Donald Trump’s 26.4 percent, with 20 percent undecided. When the cloud of statistics cleared in the days after the election, Biden took 59.26 percent of the vote, and the president scoring 39.21, with a nearly 80- percent turnout, according to unofficial totals of the secretary of the state.

The Quinnipiac University Polling Institute, the self- proclaimed “gold standard” survey, found Biden with a nationwide lead throughout the campaign, culminating in a day- before- the- election margin of 50 percent to 39 percent among likely voters, with Florida and Ohio in the bag for the former vice president. A subset of 35- to-49 year- olds in the Quinnipiac poll might have been more accurate: Biden 47 percent and Trump 46 percent. Trump easily won Florida and Ohio.

Predicting Biden’s victory this year still worked out better for pollsters than the 2016 surveys that promised an easy Hillary Clinton win over Trump.

But polling professionals and political scientists are still scratching their heads over how the scientific art of measuring public opinion, with campaign season reports detailing the horse races seized upon by news outlets, seemed to be missing so much, again.

“The big factor in the pre- election was the number of folks who were undecided,” said Michael Vigeant, chief executive officer of GreatBlue Research Inc., which polls for Sacred Heart University. He said this week that many Connecticut Trump supporters were surveyshy, knowing that their state was on- track to hand its seven electoral votes to Biden.

“That was the biggest thing that happened,” Vigeant said. “They basically said ‘ I am so sick and tired of being ostracized for my view and I’m not going to tell you.’ Connecticut is a blue state so they said they were undecided.”

Vigeant believes that in red states that stunningly supported Biden, like Georgia and Arizona, the former vice president’s supporters told pollsters they were undecided.

Another factor was the truly undecided voters, who finally made a decision in the polling booth. “The variable of this was the person who said ‘ I like what Trump did for the economy but I’m sick and tired of listening to him, and Biden is more- moderate.’ That fared well for him,” Vigeant said.

A third possible cause would be the definition of a likely voter. Some polls do not include the opinions of people who did not participate in the previous presidential election. Also, in the 2020 COVID era of Zoom calls, people might have had “phone fatigue” said Vigeant, and declined accepting calls for unfamiliar numbers.

Douglas Schwartz, a Ph. D. who is associate vice president and director of the Quinnipiac University Poll, and responsible for the survey’s methodology, declined to release more than a brief statement, pending a further study that won’t be completed until the spring of 2021.

“A full examination of what went wrong with polls this year is going to take a while,” Schwartz wrote. “At the moment, I still need to see the final election results and final exit poll results, and without those I’m not able to make even preliminary hypotheses about what exactly the issues are. After the 2016 election, it took 6 months for the American Association of Public Opinion Research ( AAPOR) to release their findings about polling errors; I would expect a full evaluation of 2020 to take at least as long, though we might have some idea of the situation before then.”

The AAPOR report on the 2016 election cited a variety of possible reasons for the errors, including the infamous pre- election bombshell that the FBI was again investigating Clinton’s email records. Also the polling postmortem found that many surveys did not adjust for the over- representation of college graduates surveyed, while non- collegeeducated whites were not counted enough.

Even placement on the ballot, in which a certain proportion of voters check off the person on the top of the ticket, no matter what the party, might not have been properly considered, because Clinton and Trump’s names appeared in different locations depending on the state.

Gayle Alberda, associate professor of politics at Fairfield University, said the biggest thing for pollsters is to figure out just who makes up the pool of voters. “Election polling is much more of an art,” she said in an interview. “It’s still a science, but there is an art to it. If pollsters do not accurately tease out who the electorate is, it can lead to errors. In 2016 the under- sampling of lower- educated whites led to a skew.”

She said that one early theory on the closeness of the 2020 race is that Democrats might have been quarantined at a higher level and would pick up the phone to talk with surveyors, while workingclass Republicans might not have been available as much.

“Also, college students and others might be more apt to take surveys, whereas with the election so divisive, it was hard for a Trump supporter to have enough trust in the institutions to participate, when their standard bearer was talking about fake news sources,” said Alberda who teaches a class in polls at Fairfield U. “Nonresponses should have been weighted toward Trump.”

Alberda believes that polls, while anticipating a high turnout for Biden, missed the intentions of Trump supporters to also come out in record numbers, although they may end up 6 million fewer than Biden’s vote total of 80 million.

“It could also be just a fundamental flaw in the way we’re doing surveys,” Alberda said. “There might have to be some really careful actions that need to be taken to adjust some of the things that we measure to determine the likely voter. She said that state polls such as the Sacred Heart survey, tend to not be as well- funded as the national polls such as Quinnipiac.

“This year was really unique,” Vigeant said. “You had the COVID impact, and ‘maybe I don’t go to the election.’ But the reality was the turnout was much- greater.”

A preliminary study by the Pew Research Institute in Washington indicates that the 2020 polling improved on four years earlier, but with Democrats embracing mail- in voting during the pandemic, it has taken more time to see final totals.

“It’s also important to recognize that not all states suffered a polling misfire,” said the Pew report, written by Scott Keeter, Courtney Kennedy and Claudia Deane. “In many important states that Biden won ( at least based on current vote totals), including Arizona, Colorado, Georgia, Minnesota, New Mexico, Nevada and Virginia, polls gave a solid read of the contest.”

Overall, the mistakes in the national polls seem to be about average for the last 12 presidential contests, Pew found. “The fact that the polling errors were not random, and that they almost uniformly involved underestimates of Republican rather than Democratic performance, points to a systematic cause or set of causes,” they reported, including the unresponsive voters, reticent Trump supporters, turnout prediction underestimates, and the pandemic.

“As we begin to study the performance of 2020 election polling in more detail, it’s also entirely possible that all of these factors contributed in some way – a “perfect storm” that blew the polls off course,” the Pew researchers concluded.

For leaders of the Democratic and Republican parties in Connecticut, it’s academic, at this point.

“I feel badly for the industry, but the reality is that it’s more and more difficult to talk with people,” said J. R. Romano, the state GOP chairman.

“Even though Democrats made gains, they are not where near where they were predicted.” He questioned the need for “horse race” campaign surveys that serve as fodder for TV news.

“The polling industry is like an aging athlete,” Romano said. “It might be time to retire. I don’t think the polling industry is adapting fast enough. It’s too hard to get people on the phone.”

Democratic State Chairwoman Nancy DiNardo said that the focus on polling misses a larger point.

“Not a single reputable poll predicted Trump would win, and they were correct,” DiNardo said. “They predicted Biden winning with a sizable national margin, and they were correct. The polls said Biden was ahead in some battleground states and in play in others, and again, correct. Polls aren’t predictors. They are a measure of preference at the time they are conducted, and overall, they weren’t wrong. It might be more productive to focus on why voters made the choices they did on their ballots. It’s the only poll that really matters.”