@josh brought up some stats about race mixing in the last stream, but they're a bit malinformed.
For the past few generations, the white population of the US has been between 90% and 60%. So even if white women picked a completely random man, between 90% and 60% of white women would end up with a white man. When you consider self segregation between races, that would be even higher since they live in white communities with other whites.
By contrast, if a black women picked a completely random man, only 12% of black women would be with black men and 88% with with non-black men.
I don't have the stats Null used on hand, but once you take this into consideration, white woman "racially loyalty" isn't as strong as the raw data implies.