Concerning the relative importance of the diploma and what you learn in college, following experiment design comes to mind. Measure wage premium for (1) not en

Concerning the relative importance of the diploma and what you learn in college, following experiment design comes to mind. Measure wage premium for (1) not enrolled to college (2) enrolled to college for 1 year and dropped out (3) 2 years and dropped out (4) 3 years and dropped out (5) graduated. One hypothesis suggests linear improvement from 1 to 5, the other hypothesis suggests discontinuity between 1~4 and 5. I don't have this data, but this seems very doable, so it probably was already done. Can you help with literature search?