The Undecided Verdict on 2020 Polling

Charlie "Chuck" Cook
December 8, 2020

It is going to take some time before it becomes clear what voters were—and were not—intending to say last month, and what meaning we should take from Nov. 3.

Disparities and contradictory findings between the two exit polls create one complication. Another stems from the new streams of data that will be coming in, adding far more detail than was available a few years ago to our collection of voters’ behavior.

Withholding judgement on what did and didn’t happen until more data is available is wise, but political aficionados, along with public officials and candidates and those working for them, want the discussion to begin a bit sooner than next spring. So, the postmortems began within hours of the polls closing and will continue for months.

Many are quick to blame polls for any election surprises. There is no question that the polling industry is going through a challenging time thanks to caller ID. It is not uncommon for a pollster to make 20 calls in order to complete a single interview. Until now, the saving grace for pollsters was that the kind of people who didn’t consent to being interviewed did not have materially different attitudes toward politics as those who did choose to participate in polls. Considerable research is now going on to determine if that is still the case.

The debate over the “shy Trump voter” will continue, with some arguing that there were voters who were reluctant to tell pollsters of their intention to vote for the president. Others, including most neutral observers who don’t see much evidence of the phenomenon, say, “not so fast.”

While I have long been skeptical of how pervasive it is, and to what extent it might be offset by any “shy Clinton/Biden voters” who might live in heavily Republican areas or have a hyper-partisan spouse, an election-week survey conducted by the Republican polling firm Public Opinion Strategies found there might be some evidence for the phenomenon. Pollsters asked, “During this campaign, did you keep your support for Donald Trump a secret from most of your friends?” Nineteen percent said they had, while 8 percent of Biden backers confessed to having kept their support of Biden to themselves. My own confession is that my mother kept her frequent votes for Democrats to herself. I don’t think my Republican voting father ever knew how often his votes were offset by those cast by his wife of over 68 years.

My own guess is that while there probably were more “shy voters” supporting Trump than there were for Hillary Clinton or Joe Biden, that partial offset doesn’t explain why there were some surprising election results.

Some of the blaming of the polls goes back to 2016, when those who saw Clinton ahead in the national surveys erroneously predicted a Clinton Electoral College win as well. The national polling that year was hardly off. Clinton’s RealClearPolitics average on Election Day was 47 percent and she received 48 percent of the final vote; Trump’s polling average was 44 percent and he received 46 percent of the vote. The polling in basically 47 states (all but Michigan, Pennsylvania, and Wisconsin), were pretty close as well.

As more research comes out, my bet is that many will see that the polls pretty much nailed the 51.3 percent Biden national share of the vote. In fact, the final RealClearPolitics average was a tenth of a point off, giving Biden a 51.2 percent share. Of the major polls released in the last five days before the election, Quinnipiac put Biden at 50 percent, while NBC News/Wall Street Journal and Fox News had him at 52 percent. Those taken a bit earlier tended to show Biden with a larger lead, indicating some closure in the last two weeks.

Many of the poll-blamers ignore the fact that undecided voters are, well, undecided. As this column has noted before, pollsters are not mind-readers and polls are not crystal balls that can see the future. If someone is asked who they plan to support and they say they haven’t decided, that means that they haven’t decided.

For virtually the entire campaign, most Americans believed that in the end, Trump would win reelection, even many who planned to vote against him. After all, incumbent presidents usually do win. As it became increasingly clear that Biden was actually the favorite, did that narrow slice of undecided voters, most of whom are independents, who by definition are skeptical about both political parties, get skittish about giving Democrats control of everything? After all, it sure sounded like things were headed that way. Did these undecided voters decide that if the country was going to give Biden the keys to the car, they didn’t really want to give the full tank of gas and a credit card to his party?

The 2020 campaign began a year and a half ago with Marianne Williamson’s call for reparations for slavery, followed by various Democratic presidential contenders and progressives calling for Medicare-for-all, open borders, abolishing ICE, and a Green New Deal. It ended with calls for packing the Supreme Court.

Maybe those polls weren’t so wrong after all.

This article was originally published for the National Journal on December 4, 2020.