[verteu]

One of the linked papers [1] describes several approaches to this problem (known as https://en.wikipedia.org/wiki/Causal_inference). They generally agree with the 10-15% ROI result.

For example:

- Regression discontinuity (find a college that accepts people who score >=X on a test, then compare people who scored X vs people who scored X-epsilon)

- Instrumental variables (find something that would nudge people to go to college, like living close to a college)

- Natural experiments (find an event that 'forced' many people to suddenly sign up, like 'avoiding the Vietnam war draft').

[1] Section 4 of https://sci-hub.se/https://doi.org/10.1146/annurev-economics...