I don't think Nate Silver's polling aggregation is the problem, it's (1) the accuracy of the polls themselves, and (2) Nate Silver's idea that he could turn his polling aggregations into an accurate projection of election chances. I said before that he was way off on his understanding of what the polls might be missing (i.e., the systematic bias as opposed to random sample error) and how those errors would correlate from state to state.
In 2016, polls actually did pretty well except for a small number of Midwestern states. It seems like maybe they were off a lot more badly and broadly this time around, although I don't know if that's true.
In 2016, polls actually did pretty well except for a small number of Midwestern states. It seems like maybe they were off a lot more badly and broadly this time around, although I don't know if that's true.
Comment