One of my favorite paper presentations ever was by Daniel Parent, who is a good enough reason in himself for pending Labor Economists to apply to HEC. He was trying to present data on income inequalities in the Financial Services industry and was forced to note—all right, I asked—that they didn’t have the data to determine if there was a racial difference in earnings because there wasn’t enough data on high-earning Blacks in the sample to be “statistically significant.” Since the sample used IRS data, among other sources, the answer was clear.
On average, Republican professors gave black students grades that were .2 of a grade point lower than their Democratic colleagues, or about two-thirds of the distance between a B and a B-minus.
(Among eleven black professors in the sample, there were no Republicans, and the Democrats appeared to grade white and black students as their white-Democratic peers did. But there were too few black professors to make that finding statistically significant.)
Again, the finding may not be statistically significant, but the sample, er, complection is.
Their data set is available here.