Nope, is there anything you'd suggest conditioning on? Regressions indicate that Latino respondents and non-metro residents are less likely to match; respondents with higher ed. attainment as well as older & female respondents are more likely to match.

Maybe...I just wonder if the first order explanation would be characteristics correlated with both difficulty in matching and Trump voting, rather than unwillingness to report (which might also be correlated with stuff, of course)...so maybe income, education, etc.

I’d bet that there’s some of that, at least.

Might all come down to residential stability...