Well, people can say what they like about the NRC rankings, but just looking at the data for the CS departments, I see some pretty graphs here, and some potentially promising correlations.
(See that outlier on the right? Three cheers for Tufts! 56% female faculty, baby.)
(I'm not pleased with that huge cluster of 0% non-Asian minority faculty. But three cheers for FIU, Iowa, and Auburn!)
I haven't yet started removing outliers or doing anything fancy, but for kicks I ran some correlations on these four variables of interest:
For anyone not used to reading ugly SPSS outputs, we find a significant correlation between non-Asian minority CS faculty and non-Asian minority CS PhD students, r = .308, p < .01. We also find a sig. correlation between female CS faculty and female CS PhD students, r = .281, p < .01. And, it turns out if there are non-Asian minority PhD CS students there is also a high likelihood of female PhD CS students, r = .247, p < .01.
These are not super-huge correlations, and it's likely the outliers are conflating things, but it's still promising just from the graphs. If I have time over the weekend I'll play with the data a bit more formally and see what turns up.