> If you want to criticize colleges on their cost, the data to use is expenditure data, which is much cleaner.

As if the opaque pricing of colleges isn't just as baffling as the healthcare system's resulting cost situation.

Another function they have is to exclude the poor and naive. Richer and more worldly students probably have higher completion rates.

Median is more important than average. A 2 or 3 variable analysis would probably be even better. I'd like to see how costs actually distribute for all students.