3. Trump might do better with nonwhites than you think.

My operating assumption has long been that Trump would run about as poorly as John McCain and Mitt Romney among African-Americans, and also would run significantly worse among Hispanics. This would require Trump to secure about 64 percent of the white vote in order to win – a tough haul. But Trump’s numbers among nonwhites have actually been relatively decent. If we assume that undecided voters are ultimately representative of decided voters (within groups), Fox News has him winning 7 percent of African-Americans and 27 percent of Hispanics – roughly Romney’s showing in 2012. ABC News has him winning a little bit more than 20 percent of the nonwhite vote – a bit better than Romney (Fox News also has him eventually winning 64 percent of the white vote).

This has led some to question the accuracy of these polls. You can (and should) read a more fulsome response from Jon Cohen and Mark Blumenthal of SurveyMonkey here, but I’ll just fall back upon a saying from one of my favorite law professors: When someone’s argument boils down to “it cannot be,” it means that it probably is.

Furthermore, it’s entirely possible that the media’s single-minded focus upon Hispanics as immigration reform advocates is simply wrongheaded. I’ve written about this at much greater length here and here, but it’s entirely possible that Romney’s showing in 2012 represents a floor, that immigration isn’t as high-salience an issue among Hispanics as many assume, that some of Trump’s appeal to working-class whites translates to working-class Hispanics and African-Americans, and that he will perform surprisingly well (or at least surprisingly not poorly) on Election Day.