The NYT has a nice puff piece on the computer science department at the University of Washington, from which I received one of my two undergraduate degrees.
Some things have changed a lot since then, like the building, the department ranking, and being in the engineering school rather than arts and sciences, but some things remain, like there being "too few" slots.
Hank Levy and Ed Lazowska, mentioned in the article, were both around in my undergraduate days, and I took a very enjoyable class from Ed on operating systems. It was probably the most challenging course I took at Washington other than the first quarter of the graduate micro sequence from Gene Silverberg. At the end of it, Ed gave all the students buttons that said "I Survived Computer Science 451".
This seems like a perfect opportunity for both an experiment and the potential use of non-experimental techniques. First, let's start with a potential "queasy"-experiment - wherein we take the incoming GPA, GRE score, undergraduate major, PhD program and of course a 'humanities error term' (since going into a PhD program in the humanities is obviously an error - kidding) and compare the students who select and don't select into the program using a matching technique on their time to degree (we could also think about other outcome variables such as rank of placement in first position, but let's keep it simple). And then, UM could take the program going forward and randomly assign admissions to the program for future cohorts of grad students. Point being, I think you're right Jeff, that we should be cautious of the misuse of causality in place like the Record, but also that the (perhaps more pernicious?) misuse also occurs within the academy. How many PhD program requirements or decisions are also made with similarly shoddy causal reasoning? Before we start casting aspersions broadly, maybe those of us who are more causally-attuned should make sure our own house is in order?
ReplyDelete