Mike Titelbaum (University of Wisconsin-Madison) wrote in by email:
I am a big fan of The Philosophers' Cocoon, and occasionally jump into the comments section...[But] There's something that occasionally comes up on Cocoon threads and elsewhere that bothers me...
Often when we counsel students comparing graduate schools, we suggest that they look (among other things) at the percentage of admits who complete the program. The underlying implication seems to be that higher completion rates indicate a better program. But this seems to me out of step with another discourse floating around the internet in general and the Cocoon in particular—that deciding as a PhD student that you don't want to become an academic isn't a failure in any sense.
Sometimes a department has a low completion rate because they're unsupportive, are chasing out certain students, or have a toxic environment. But PhD students leave programs for many reasons, including that the program has done a good job of revealing to them what the academic life is like, and they've decided it isn't the life for them (even though it might be available to them). If a department was pressuring people to stay and finish their degree despite having had such a realization, that would enhance their completion rate but it seems to me would be the kind of department that I would discourage a student from joining.
There's also presumably a big selection effect here. Different departments wind up with different kinds of students. Some populations might be more inclined to decide that academia isn't for them and leave their program before completing. That would hurt a department's completion rate, but perhaps indicate nothing bad about how the department treats students once they arrive. In fact, if a department is more willing to take a risk on certain types of students who might not finish, that might be a good sign about that department's atmosphere and goals.
As I said, this is just something that's been bothering me. I have absolutely no empirical data on any of this, and not even much anecdotal data. But if you're measuring a department by its completion rate, then you're assuming the sole goal of a PhD program is to create PhDs (and possibly then academics). Perhaps instead the first goal of a PhD program should be to help students figure out whether writing a dissertation, completing a PhD, and then becoming an academic is the right path for them?
I think Mike raises really good points here. In email discussion, I suggested that perhaps the best thing to do is to compile and make more transparent both qualitative and quantitative data regarding attrition and job-placemtn.