Pollster methodology changes and response patterns change.

This is why I say polling error and bias change across elections.

To add more detail to these posts, a theory for the polling error in 2016 was that pollsters weren't weighting on education properly.

In the last decade, education has been strongly sorted by political ideology with non-college whites becoming Trumpers and college whites becoming Democrats. College whites used to be fairly Republican. And non-college whites used to be somewhat Democratic voting. Non-college whites for example were a part of the Obama coalition and why he did so well in the Midwest. In particular college educated women are one of the most reliable Democratic voting blocs now.

A theory for the polling error in 2020 was that Democrats were much more likely to respond to pollsters during the pandemic. This is called partisan response bias and it is very difficult for pollsters to account for. It has likely been one of the largest sources of error in recent years.

And another theory for both 2016 and 2020 was that Trump brings out voters that are less likely to respond to polls. Not necessarily "shy" Trump voters, just low engagement voters that only come out for him.

It is hard to imagine Trump has improved his standing with 2020 voters such that the 2024 race is significantly different. So it makes a lot of sense that polling error and bias this cycle won't be as bad as in 2016/2020. You have to imagine somehow Biden versus Trump rematch was going to end up a historic blowout as Harris improved significantly from Biden in recontact surveys.

We won't know ahead of time, which is why election forecasts are useful. We shouldn't be counting on polling error or bias in a certain direction as there is no way of knowing. That goes for both people expecting polls to underestimate Trump and people expecting polls to underestimate Harris. We just don't know and historically there is no consistent correlation.

And as noted by Liam Donovan the cross-tabs are still quite unusual this cycle even with Harris as the nominee. They show Trump with quite unusually high support among young voters and Black voters. This isn't an issue for the toplines as the toplines of the polls are adjusted properly. But it suggests we should tread carefully.