These are huge sample sizes. How accurate were they in 2016?
they had +3 Clinton as final
the 52-43 is accurate. but the undecideds will break hard Trump (bet on it).
Debatable. Undecideds may be more likely to break against an unpopular incumbent.
They're not actually undecideds though (don't believe any cross-tab racial demographics of them...).
Biden should focus on turning out voters/GOTV. The remaining undecideds are shy Trump voters. Bank on it.
If you don't trust the pollster enough to believe the demographic breakdown, why do you trust it enough to believe the 52-43 margin? This seems like cherry picking your data.
