Do run those numbers on a per-major basis, though. And try (though is is hard to get data) to separate the cohort who went to study and the cohort who went to party for 4 years.
In other words, some data on job-type mix rather than solely median salaries would be interesting.
Unless you mean against minimum wage, but I'm not sure how telling that is. Certainly those with CS degrees will appear to be faring rather well, but that ignores that people without any degrees at all are doing just as well in the industry, and therefore are actually doing significantly better once you factor in the aforementioned costs.