It frustrates me that these ROI calculations are done as comparisons between two separate populations, each with different characteristics like raw intelligence or work ethic. Return on investment should be looking at treatment effects only.

For example, it may be that science minded people have higher salaries in general, and those people also tend to apply to technical schools. In that case, the real ROI is from being technically minded, not from attending a school that focuses on that.

A better test would be looking at kids that got into both Harvey Mudd and MIT and separating them into populations based on where each kid decided to matriculate. That way you're factoring out the admissions bias.

Malcolm Gladwell wrote an article in the New Yorker about the difference between state schools and Ivy League schools that actually looked at treatment effects. And he found that there were none for the majority of the population.

I haven't seen one of these done for college vs. no college, but that would be much more valuable than misleading shit like this.

edit: summary of the Gladwell article here http://notesandrestsmakemusic.wordpress.com/2011/09/14/the-v...