<p>Sure, the report is complete bull. But I’m not sure it would be possible to get more accurate. For example, where are you going to find the statistically significant population of people who were offered admission to Caltech (or even looked “qualified” to go to Caltech) but who chose not to go to college at all? And why should someone’s million-dollar capital gain be attributed to his college? They are trying to quantify the earning-potential difference a particular college makes. I don’t blame them for trying – it would be interesting to know the answer – but I suspect it would take mounds of data and a zillion simultaneous regressions even to begin to get an accurate answer.</p>