It looks like I was wrong.
Soon after the election, I speculated that the polls failed to predict the outcome because “unlikely” voters weren’t in the pollsters’ samples after the party conventions but turned out in large numbers on election day. That does not, however, seem to be the case.
Rather, the “likely” voters who were expected to vote for Clinton, notably but not exclusively African-Americans, turned out in much smaller numbers than expected.
That’s only one reason for Clinton’s defeat but I’m not writing to analyze that matter. Rather, I’m just going on record that it looks like I was wrong. Is this an example to be followed by others? It wouldn’t hurt.