I'm in a swing state with an abortion measure on the ballot, and while all the polls claim it's close, I'm not really sure they are properly accounting for the number of voters that have been activated by the possibility of enshrining pro-choice into the state constitution.
These polling strategies are complex and a lot of thought goes into them, but they rarely can account for uncommon circumstances that increase voter turnout in local or state elections and how that will effect the national election.
While this is entirely personal reexperience bias, I also wonder how effective these polls are at reaching a representative survey group. I know at least on my phone basically all survey calls and texts go to spam and I wonder if older, more conservative voters are getting overrepresented due to their likelihood of not having those kinds of spam filters in place.