(Note to self - this is what happens when you nerd out and forget the legend!)

What you're looking at is the result of a cluster analysis. The Getis Ord Gi* statistic indicates the presence of hotspots (red) and cold spots (blue).

Basically, the expectation (null hypothesis) is that the phenomenon we're looking at happens at random across the US: equally likely to occur in a county in Washington as in a county in Alabama. However, many phenomena have a certain spatial distribution and this statistic calculates how important location is for the phenomenon to occur. In other words, when you look at the counties I put on the map in my top comment, the hot spots indicate places where these red counties are clustered a lot and the cold spots indicate where the non-red counties are clustered.

edit: To put it even simpler: an increase in diversity alongside an increase in Trump votes is clustered in these hotspots. While this does not prove a causation between diversity and Trump, it does show a correlation between the two.

Maybe I should also look at median income changes in PoC to get to the point raised by that Vox article from a while ago:

kleinbl00, is that why you wanted to know about income deltas?