The Case for Routine Randomization
One of the interesting aspects of the $320,000 kindergarten teacher study is that the study wasn’t originally designed to estimate the value of unusually effective kindergarten teachers. It was designed to study the effects of class size reduction. That was the policy intervention people were interested in back in the 1980s when the well-known Tennessee STAR experiment was created. There’s no way its designers could have known that a series of econometric studies finding huge differences in effectiveness among teachers with similar credentials would subsequently emerge, profoundly changing the way people think about the profession and making class size reduction passe.
Which makes me think that school districts should start randomly assigning some of their students to teachers
