We did our analysis on a faculty-by-faculty basis. We were comparing computer scientists with computer scientists, and folks in the faculty of management with their true peer group. The discrepancies varied faculty by faculty.
We also found that because we had made our regression line, it was very easy to say that this faculty member was far below where they should be relative to another faculty member who was close to the regression line.
We really did a deep dive into the data and were able to make tailored adjustments based on the regression line.