It’s a frustrating reality that Donald Trump is probably doing better in his reelection campaign than the polls are indicating. Steven Shepard of Politico reports that a lot of the pollsters acknowledge that state polling in particular is a problem.
There’s a fairly simple explanation, although it’s composed of several parts. The most important is that a huge gap has opened up in the political opinions of white people based on their level of educational attainment. Less-educated whites are far more inclined to support Donald Trump and far less likely to respond to a political survey request.
This could lead to an undersampling in the best of worlds, but it’s compounded by the fact that Trump consistently rails against the media and the polls, accusing them of dishonest and unfair practices. Therefore, people who follow Trump’s messaging closely are being prompted not to cooperate with reporters and pollsters.
Finally, there’s an element of shame or judgment involved. Trump and his supporters are frequently and with justification accused of harboring racist beliefs. This adds a potential social cost to openly admitting that you intend to vote for the president. This can be offset in some communities where there’s a high social cost for supporting a liberal or pro-choice candidate, but many people are reluctant to admit to friends and family that they prefer Trump to Biden, let alone admit this to a stranger on the telephone.
This problem was recognized as the reason that state polling was off in 2016 even though the larger sample-size national polls were accurate, and the best pollsters have been working to adjust their models ever since. The question is really whether or not they’ve solved it and whether the average of polls is skewed by polling outfits who haven’t made the proper adjustments.
GOP pollster Glen Bolger said he believes a combination of pollsters’ inability to get the right educational mix and to persuade potential Trump voters to respond and answer truthfully to phone polls is pointing their surveys in a slightly Democratic direction.
“I don’t know how big the effect is. I also don’t know what the ratio is between it being ‘shy Trump’ voters and interviewing too many college graduates and not enough non-college grads,” Bolger said. “But I do think those are factors in some of the polls that show a particularly wide lead for Biden at this point in time. And I do think that things will be closer in the states than the polls indicate right now.”
The truth is that Trump has created a uniquely poll-resistant political coalition. If pollsters don’t have a good plan for this, they’ll underestimate his level of support. Of course, it’s possible to overcompensate and skew your poll in the opposite direction, but it’s significant that the pollsters themselves seem to lack total confidence in their results.
Trump’s measured level of support has been cratering over the last several months, and he’s now clearly losing outside the margin of any error, but there’s still a good possibility that he’s in better shape than the numbers suggest.
Fortunately, the experience of 2016 pretty much assures that the Democrats won’t get complacent. It would be hard to feel confident in any case when the polls can vary so much. For example, there are two recent polls out of Michigan, and one has Biden up by two points while the other has him up by sixteen. The prudent thing to do is to assume the closer result is more accurate.