Do you believe there is a nonresponse bias?

[RI]

Something which may be an interesting look at the issue: https://www.nytimes.com/2022/11/03/upshot/midterms-wisconsin-polling-experiment.html

[Del Tachi]

I expect Republican non-response will be less of an issue this year because Republicans are more enthused than in 2020.  Enthused voters are more likely to answer polls.

If Republicans overperform next week, it will be hard to disentangle whether it was from polling error or the GOP gaining momentum and winning over independents in the closing stretch.  That so many Democrats in competitive races never reliably broke 50% may suggest the latter.

[T0rM3nTeD]

Seems like the best polling prediction would be to have every 5th grade class (select any grade) in the US do a mock election. They won't understand anything about any candidates and will just ask their parents who they should vote for. Obviously a flawed system, but it seemed to be fairly accurate in Iowa.