Americans have been complaining about pollster inaccuracy regarding the last two presidential elections–in 2016 and 2020, that is, the one just held on November 3 this year. But these polls actually were better this year for this recent election than in the past. That seems surprising.
Polling is difficult. The way they make polls is that pollsters cold-call people on their phones and ask them questions. A lot of people refuse to answer, regarding it as spam I guess, and hang up. Maybe this fact is most responsible for the inaccuracies of our U.S. presidential polling. I suspect pollsters know more about reasons than the public does.
Today, The New York Times published the following chart showing the lack of accuracy of the polls for this recent election for Trump and Biden for president. It shows that pollsters got it most wrong regarding the states of Ohio and Iowa, both of which had a differential of seven points. That seems strange to me. In the past, Ohio has been a bell-weather state indicating how the whole nation would vote. And Iowa historically has been the first state to hold its Democratic and Republican primaries and caucasus during a presidential election year. (The backward arrow means “less than,” that is, less than 1%.)