This is fair. But my issue is that when we do have robust subsample polling and that robust polling contradicts the crosstabs, I'm going to assume the robust polling is correct. Especially if the latter aligns with the priors.
Particularly when the crosstabs don't make a lot of sense. Like we have a lot of data on Generation Z. They haven't really demonstrated much interest in the GOP. So I suppose any sort of realignment is possible, but my priors are going to be stronger on that than on other things.
And what from the Trump campaign would indicate that he would somehow get the largest share of the Black vote since the Civil Rights Act? His "outreach" to Black voters is to talk about how he is a felon.