Some thoughts on polling
By Scott Sumner
Pollsters can try to adjust their sample for gender, race, political party, education, and a dozen other demographic categories. But there’s one category for which it would seem inherently impossible to adjust—differences in willingness to talk to pollsters that cuts across the other demographics. Many Trump voters simply don’t want to respond to pollsters. And you don’t discover the political skew of the non-responders until the election itself.
We’ve already seen that there’s a huge partisan difference in willingness to use mail-in ballots; why should we be surprised that there’s a modest difference in willingness to talk to pollsters?
Perhaps this anti-pollster attitude is more common in places like Wisconsin, with lots of farmers and smaller industrial towns, as compared to Arizona, which fewer farmers and small industrial towns. At least that seems to have been the case in both 2016 and 2020.
On a separate issue, I’ve frequently argued that working class whites that are struggling to get by don’t like being told by Ivy League professors that they benefit from “white privilege”. I don’t even think Hispanics like the concept. (Note to commenters: This point is completely separate from the question of whether working class whites do in fact benefit from white privilege.)
All year long I’ve had a nagging feeling that the “woke” movement could hand the election to Trump. Perhaps it did not, but I suspect it came close to doing so. Perhaps a Trump victory was prevented by something as random as a big October surge in Covid deaths in Wisconsin.
Too soon to say!