I've discussed this with people on HN before, and I really wish I could find better stats on MS attrition rates, especially at good schools.

I was once a PhD student at UC Berkeley, and I felt it was pretty much a horror show of failure and attrition (the experience caused me to view claims of a "shortage" of US grad students in a very different light). It's tough to get into med school (or a top law school), but looking at the numbers, it doesn't appear to be any worse than top engineering grad schools (where GPAs and GRE scores are sky high). But in the professional schools, once you're in, attrition rates are very low. Yale law school, for instance, has no 1L attrition. Columbia's 1L attrition rate is 0.3%. Attrition rates for elite med schools tend to be well below 1%. Something odd is going on when congress claims that there is such a severe shortage that we need to start awarding green cards to anyone who gets a grad degree in STEM from a US university [1], yet attrition rates start at 35% (in engineering) and only get worse from there.

I think that engineering, math, physical sciences are a pretty grueling path compared to the professions - a point of view supported (at least at the PhD level) by a RAND position paper:

http://www.rand.org/pubs/issue_papers/IP241.html

Anyway, none of this is meant to refute your point - you are right about the MS issue. Though I really do wish we had better info on MS attrition rates. When I was at UCB, there was some attrition from the MS program as well, though it was not anywhere close to as high as the PhD program.

[1] I support this, however, I don't support the reasoning behind it - that a poor educational system is the cause of the non-participation of US students in STEM grad programs. I believe the aversion is rational and market driven, and any policy to expand the engineering workforce at the graduate level should consider this carefully.

http://www.cgsnet.org/study-finds-completion-rates-stem-mast...

This was part of a study done by the council for graduate schools that recognized that attrition rates from MS programs were understudied compared to undergrad or PhD rates.

The full study is behind a pay wall, but still some interesting data.

STEM MS degrees appear to be completed within 4 years about 2/3 of the time - though that number drops to the low 40s within 2 years.

Unfortunately, there's only so much you can draw from this. MS programs aren't nearly as regulated as Law or Med school or PhD programs which tend to be offered only by very academically serious schools. Also, one of the big reasons for failing to finish a MS, according to the article, was work. So many students pursuing a MS are already employed in the field - something you can't do without a professional degree. Still, dropout rates were much higher than for MBA programs, so this data clearly means something.

It would be more interesting to know the attrition rates for MS students in STEM who are more like professional students at top colleges. In other words, what is the attrition rate for MS students who are enrolled full time at a top college.

Maybe these answers are behind the pay wall :(