New Yorker: “On October 18, 2016, the New York Times gave Hillary Clinton a ninety-one-per-cent chance of beating Donald Trump. Five days later, ABC News released a tracking poll showing her ahead of Trump by twelve points. Buoyed by the polls, Democrats—especially Democratic women—approached November 8th with a joyful sense of inevitability. The collective disbelief when Clinton lost was tinged with confusion: How could the pollsters have been so wrong? Now, with Joe Biden leading Trump by double digits in the lead-up to Election Day, according to the latest NPR/PBS NewsHour/Marist survey, the question has to be asked: Are voters hoping for a Biden victory about to fall in the same trap?”

“Polls are not predictive. They are snapshots taken at a particular moment in time. By the time the public sees a poll, that moment will have passed. Andrew Weissert, who runs a Republican polling firm outside of Chicago, told me, ‘I remember, in 2012, when Mitt Romney had a stellar debate performance, and, all of a sudden, people were writing President Obama’s political obituary. Then the next debate happened. Obama had a strong debate. And suddenly the polls were showing something completely different.’ Polls are sensitive to the news because people are sensitive to the news. They are inconsistent because people are inconsistent. According to Andrew Mercer, a senior research methodologist at the Pew Research Center, polls also tend to vary because ‘there are a lot of subjective decisions about how to go about things that different pollsters do differently for a variety of reasons.'”