I want to continue with talk about my paper on the relationship between African American student enrollment and U.S. News peer assessment scores. The bottom line is that there is little relationship when we look at all schools -- but when one focuses on the top 103 schools and top 26 schools there is a small correlation (around r = .20). The illustration at right, which is taken from my paper, plots the percentage of African American students against schools' U.S. News peer assessment scores of the top 26 schools.
What I want to do now is look at the most elite schools -- those with U.S. News peer assessment scores of 4.0 or better. They are Harvard, Stanford, Yale, Columbia, Chicago, NYU, California Berkeley, Michigan, Penn, Virginia, Georgetown, Duke, Cornell, Texas, Northwestern, UCLA. For those 16 schools there is a substantially (but not statistically significant) higher correlation (r = .43, p = .10). You can see a plot of the peer assessment scores against the percentage African American student enrollment of those 16 schools at right. I'm now wondering if the percentage of African American student enrollment might be most useful for distinguishing between schools in one segment of the market, such as the most elite schools?