Wednesday, January 16, 2013

We ignore causality when it suits us ...

From today's University of Michigan Record Update:
The Sweetland Center for Writing's Dissertation Writing Institute, now in its 10th year, is helping to boost doctoral program completion rates of graduate students. Compared with national rates of 50-60 percent, institute students complete at a rate of 88 percent.
To interpret this causally requires that students who select into the program are a random sample of all doctoral students, as is required for the causal interpretation the university gives to this difference in completion rates. That seems pretty darn unlikely, and the university makes no case for it in the RU write-up.


NB: This is not a diss on the Sweetland program. One of my students went through it and I think gained a lot from it in terms of the quality of his writing (though it had no effect on his probability of degree completion). My complaint is with the university pushing casual causal claims as part of its marketing efforts that faculty members would mercilessly mock in a seminar or refereeing context.

No comments: