A better way to evaluate teachers?
January 15, 2017 Leave a comment
Well, I can tell you for sure that simply holding them responsible for student test scores is decidedly not what we should be doing. It is not a coincidence that of all the nations that out-perform us in education, none of them take this approach (like the way that none of the nations that out-perform us in health care do it by leaving more to the marketplace). Anyway, I know I read this excellent Paul Tough piece before on teaching resilience– and hopefully recommended it– but I just came across it again and was drawn to the portion on teacher evaluation:
A few years ago, a young economist at Northwestern University named C. Kirabo Jackson began investigating how to measure educators’ effectiveness. In many school systems these days, teachers are assessed based primarily on one data point: the standardized-test scores of their students. Jackson suspected that the true impact teachers had on their students was more complicated than a single test score could reveal. So he found and analyzed a detailed database in North Carolina that tracked the performance of every single ninth-grade student in the state from 2005 to 2011—a total of 464,502 students. His data followed their progress not only in ninth grade but throughout high school.
Jackson had access to students’ scores on the statewide standardized test, and he used that as a rough measure of their cognitive ability. This is the number that education officials generally look at when trying to assess teachers’ impact. But then Jackson did something new. He created a proxy measure for students’ noncognitive ability, using just four pieces of existing administrative data: attendance, suspensions, on-time grade progression, and overall GPA. Jackson’s new index measured, in a fairly crude way, how engaged students were in school—whether they showed up, whether they misbehaved, and how hard they worked in their classes. Jackson found that this simple noncognitive proxy was, remarkably, a better predictor than students’ test scores of whether the students would go on to attend college, a better predictor of adult wages, and a better predictor of future arrests.
Jackson’s proxy measure allowed him to do some intriguing analysis of teachers’ effectiveness. He subjected every ninth-grade English and algebra teacher in North Carolina to what economists call a value-added assessment. First he calculated whether and how being a student in a particular teacher’s class affected that student’s standardized-test score. Then, separately, he calculated the effect that teachers had on their students’ noncognitive proxy measure: on their attendance, suspensions, timely progression from one grade to the next, and overall GPA.
Jackson found that some teachers were reliably able to raise their students’ standardized-test scores year after year. These are the teachers, in every teacher-evaluation system in the country, who are the most valued and most rewarded. But he also found that there was another distinct cohort of teachers who were reliably able to raise their students’ performance on his noncognitive measure. If you were assigned to the class of a teacher in this cohort, you were more likely to show up to school, more likely to avoid suspension, more likely to move on to the next grade. And your overall GPA went up—not just your grades in that particular teacher’s class, but your grades in your other classes, too.
Jackson found that these two groups of successful teachers did not necessarily overlap much; in every school, it seemed, there were certain teachers who were especially good at developing cognitive skills in their students and other teachers who excelled at developing noncognitive skills. But the teachers in the second cohort were not being rewarded for their success with their students—indeed, it seemed likely that no one but Jackson even realized that they were successful. And yet those teachers, according to Jackson’s calculations, were doing more to get their students to college and raise their future wages than were the much-celebrated teachers who boosted students’ test scores. [emphasis mine]
I’m not particularly enthusiastic about rewarding teachers based on any particular measures of student performance as I think the evidence is fairly clear that there’s far better ways to improve the quality of teaching and education. But, as long as we’re going to stick with it, it would be great to incorporate this kind of data, rather than just test scores.