There are also some issues around job preferences not being solely based on pay. The kinds of low-ish pay jobs worked by degree holders aren't identical to the

There are also some issues around job preferences not being solely based on pay. The kinds of low-ish pay jobs worked by degree holders aren't identical to the distribution of low-ish pay jobs worked by non-degree holders, and skew, I believe, more office/white-collar rather than manual labor. Some people implicitly attach a significant value to that, whether due to working conditions or status, enough that they're willing to take that route even if they knew up front that it's not a net-positive financial decision. E.g., if attending college means you effectively make $10k/yr less, taking all costs into consideration, but you have a 60% greater chance of it being at a desk in an air-conditioned office, that might be worth it to some people. In other words, some data on job-type mix rather than solely median salaries would be interesting.