Imagine if you did a study where you got a representative sample of black voters, gave them all a survey and then ranked them from most liberal to most conservative, and then divided that ranking into quartiles. A majority of all four quartiles would be Democrats, including the most conservative.
Now imagine you did the same for a representative group of white voters. The Democrats would only win the two more liberal quartiles, and there'd be virtually none in the most conservative one.
And then if you mix the Democrats from both groups together and ranked them, you'd as a result have disproportionately more black voters on the more conservative end as a result, even if the median black voter is to the left of the median white voter.
I think we really need to dispel this myth that Black voters are largely conservative just because they’re not as left-wing as the median Oberlin undergrad.
Not to mention that there are a non-negligible number of ultra-liberal college student/graduate black 20-somethings, too.
I think this is kind of an implementation of the
prosecutor's fallacy. The percentage of people who go to colleges like Oberlin period, white, black, Asian, etc. is quite low. So of any racial group, the percentage who attended such a college is still going to make a near neglible percentage of voters. Most college educated voters are ones who went to mainstream state universities.