NYU is building an accountability system to measure its teachers

The former testing czar at the old Board of Education, Robert Tobias, sometimes offers criticism of the accountability programs being produced these days at Tweed Courthouse. He’s also been hatching an accountability system of his own — this one to study the effectiveness of teachers produced by New York University’s school of education, where he now works.

Preliminary results suggest that teachers trained at NYU are getting above-average results in English, but they give students no extra boost on math tests, Tobias said last week at the educational research conference Philissa and I attended in San Diego. He also found that NYU-trained elementary-school teachers produced significantly greater results for students than middle-school teachers, and that the teachers get better as they become more experienced. The effect tapers off at between five and nine years into the job, he said.

Tobias’s results could provide one clue about what’s being found in an ongoing research project about teacher training programs in New York City. So far, that project has found that different programs produced different student results but has not named the programs that had the largest effects.

The results could also be important as alternative teacher training programs like Teach For America increasingly bring into question the need for traditional programs based entirely at universities. “As a dean I want to say I want to steal these three and have them do it at my school,” said Rick Ginsberg, who runs the education school at the University of Kansas, referring to the professors working with Tobias. “We’re fighting this battle all the time.”

Concern about defending programs housed at universities was on display all week in San Diego, where most attendees were affiliated with schools of education. One well-attended lecture devoted itself entirely to the future of graduate schools of education. The dean of the University of Michigan education school, Deborah Ball, delivered the talk and said a chief role for education schools is to explain why, rather than whether, some teachers do better than others.

Tobias said he intends for NYU to use his model to compare the different programs that the education school trains teachers to use in the classroom.

Tobias’s system, which he is developing with a set of NYU professors, employs the same model the Bloomberg administration is using to measure teachers’ quality, a so-called “value-added” formula that compares a prediction of how well students would be expected to score on English and math tests to how well they actually score. The model gives high marks to teachers whose students do better on tests than certain background factors — things like class size, family income, and special education status — would predict.

The model has attracted criticism because it draws on a small and narrow set of information, and Tobias said during his presentation that New York’s state tests are not perfectly suited to this kind of analysis. He called that a “limitation” of his study.