Thus far, the two best predictors of GOP voter preferences have been white socioeconomic status and an area’s partisanship. Trump has performed best among non-Hispanic whites with low socioeconomic status, especially in blue areas. Meanwhile, Cruz has fared best in more conservative areas, Kasich has run best in blue areas, and both have run slightly better in areas with high white socioeconomic status.

To illustrate this phenomenon, let’s divide the counties that have voted into four quadrants.1 First, let’s divide the GOP electorate by white socioeconomic status, based on a composite index we created, 2 so that 50 percent of Republican voters that have cast a ballot so far fall in each each half (High, Low). Then let’s cut the GOP electorate into two halves by partisanship (Blue, Red): About half of the 23 million GOP primary votes have been cast in counties where President Obama won at least 44.6 percent of the two-party vote in 2012; the other half have been cast in counties where Mitt Romney took at least 55.4 percent.

And voila: We have four quadrants that represent four unique segments of the GOP electorate: Let’s call them HighBlue, LowBlue, HighRed and LowRed.