Does anyone know of any studies that take into consideration labour-force dropout as well? It's a bit less enlightening if you compare the 95% of science grads who are employed doing science with the 60% of soft studies grads (both numbers purely hypothetical), to compare the most career-driven X-students with all the Y-students is unfair to Y-studies.

Tenored professorships are nicely paying, hard to get but exist in every discipline (let's pick on Gender Studies) so if 2/100 GS grads become 100K/year tenured profs and 98/100 opt out of the labour force, you're really measuring the wrong thing.