Teachers are central to quality education. Arguably, schools of education are central to quality teachers. The last six months have seen two enormous steps in evaluating schools of education: publication by US News of NCTQ’s rankings of teacher prep programs and release of new accreditation standards by CAEP. For the most part, both NCTQ and CAEP look at inputs in the teacher training process (admission standards, curriculum, etc.) rather than outputs (evidence of how well graduates teach). The issue is that evidence on outputs is very hard to come by. There just ain’t much data.
The obvious first piece of evidence everyone would like to see is measures of value-added of the students of teachers who graduate from each program. Such numbers are generally unavailable because, with a few exceptions, most states don’t collect the necessary data in a usable way. New research by Mihaly, McCaffrey, Sass, and Lockwood suggests that even when the data is collected it’s going to be hard to pin an accurate number to each ed school.
Here’s the gist of the problem. How well teachers perform presumably depends in part on the quality of the school at which they teach. It’s possible to statistically adjust for school differences using what’s called “school fixed effects.” Implementing this kind of statistical control can make a big difference in how an ed school is ranked. For example, looking at rankings for recent graduates from 33 schools of education in Florida, the authors find one school ranked 2nd based on raw scores alone, 1st ranked adjusting for student demographics, and 32nd (out of 33!) after adjusting for school fixed effects.
Life gets further complicated by the fact that schools of education often have pipelines for placing many of their graduates in specific school districts. As an extreme example, suppose an ed school streamed all their teachers into one school. Then there wouldn’t be any to separate the quality of the teacher training from the quality of that one school. Real sorting of graduates into schools is nowhere near that severe, but there is enough sorting that getting accurate rankings of ed schools is tough.
None of this says that VAM scores can’t be used to help measure the output of teacher training programs. But it’s going to be tougher than we’d like.