Quote (Thor123422 @ 3 Sep 2020 04:03)
Yeah, the data last time wasn't actually bad, it's just that the polls swung so fast with the Comey letter nobody believed them.
That's only half the story of 2016 polling. Yes, the national polls were pretty good, and they are surely forgiven for missing the late movement caused by the Comey letter. But the state level polls were far worse, and that's the real story of 2016.
In reality, pollsters missed two trends moving in opposite directions: disaffected white non-college people who hadnt voted in a long time turning out specifically for Trump, and disaffected low turnout minority voters turning out specifically to vote against Trump. The former group is plentiful in the Rust Belt and got Trump across the finish line there, while the latter group lead to his smaller than expected margins in states like AZ, TX or GA. On a national level, these trends roughly cancelled each other out, but on the state level, they led to big polling errors in both directions.