Earlier quoted context omitted.
You did not read the article all the way through, nor did the people who voted you up. If you had, you would have encountered references to a variety of studies that attempt to control for going to college as opposed to other predictive factors using things like twin studies. They have found that the impact of college is less than simple analyses show, but the largest contribution to the difference between college gr…
Oh I've read the article, and many like it. I'm not sure you've read my comment carefully though, because you seem to be missing the main point. Such studies are largely invalidated by their very influence on the phenomenon they purport to describe. Do you think the applicants' belief they would earn 45% (or 70%, or 90%) more has no bearing on the labor and loan markets?
E.g. What data do you want? Where can you get it? What tests / analyses do you want them to run? This is a hard bloody problem, so I find "Monday quarterbacking" particularly frustrating -- especially when the priceomics folks are doing a good job of showing / communicating data with documented assumptions.