We took a poll of pollsters (unscientifically) to see what changes they’ve made since 2016 and what they’re worried about in 2020.

“One key takeaway from 2016 was just how important someone’s level of educational attainment was in predicting their vote.”

Excellent overview of the complexities and perils of the polling process, given the pandemic and the experience in 2016.

Only 20 days to go until the U.S. election. Given that every time I tweet a poll people complain about the polls here is a helpful primer on how those polls have changed since 2016 and what is still keeping the pollsters up at night ...

Most pollsters now weight by education, some by density. More are using registration lists. But we’re still having similar debates about phone vs online surveys.

10.17.16. @FiveThirtyEight: @HillaryClinton has 88% chance of beating @realDonaldTrump. 10.16. 20. @NateSilver538: @JoeBiden has 87% chance. In this article, #pollsters explain how #polls have changed since 2016. Are they more reliable? I have doubts. fivethirtyeight.com/features/what-… pic.twitter.com/YFrLkQjsaT

10.16.20. @NateSilver538: #Biden has 87% chance of beating #Trump. 10.17.16. @FiveThirtyEight: #Hillary had an 88% chance! In link below, Silver explains why we shouldn't be worried. Does it make you feel better? fivethirtyeight.com/features/what-… pic.twitter.com/vsnAKw4NDX

While encouraging in many ways, I find the lack of attention to gender in this piece a bit surprising. It seems that there is decent evidence that women are more likely to respond to surveys and to dislike Trump. Weighting could help.

10.16. 20. @NateSilver538: #Biden has 87% chance of beating #Trump. 10.17.16. @FiveThirtyEight: #Hillary had an 88% chance! In the link below, #pollsters explain what they have changed since 2016. What do you think? fivethirtyeight.com/features/what-… pic.twitter.com/9qMu5nTVo7