Some of the state-level polling was off, but the national polling data was spot on. RCP on the day of the election had Clinton over Trump by 3.2 points nationally. Clinton ended up winning the national popular vote by 2.1 points, with that 1.1 point difference well within the margin of error of the polling data. So the national polling data was correct.
But there's really no point making comparisons to 2016. Trump wasn't the incumbent then [[he is now), he had no record to run on [[he most certainly does now). Biden is not Hillary. And we weren't in a viral pandemic then causing a full-on economic depression. It's not even remotely the same environment as 2016.
As far as the polls, after 2016, methodology got revamped to address the failures with some of the state-level polling. If you want to make comparisons, look at the accuracy of polling data for the 2018 and 2019 elections for a barometer of how accurate they are now.
Bookmarks