Commentary on “Great Teaching:Measuring its effects on students’ future earnings” By Raj Chetty, John N. Friedman and Jonah E. Rockoff
Raj Chetty, John Friedman, and Jonah Rockoff have carried out a remarkable study, but I suspect it will be misinterpreted.
The main contribution of their research is quantifying the importance of teaching. Specifically, the authors conclude that students taught by a more effective teacher will collectively earn hundreds of thousands of dollars more over their lifetimes, and that good teachers similarly influence college going and teenage pregnancy. Because each teacher influences thousands of students over a career, this suggests that one excellent teacher could generate enormous social and economic benefits.
I find these results plausible, though there are some real limitations. The researchers present convincing evidence that their estimates of teacher contributions to student achievement are valid and do not simply reflect differences in student background. But this type of “selection bias” could influence effects on earnings and other long-term outcomes. So, the most intriguing findings here are also still somewhat tenuous. Given the small size of the effects for each individual student, even a slight bit of selection bias could dramatically alter the estimated benefits of an individual teacher.
Perhaps the more important question is, what do the results mean for policy? Policymakers had already concluded that we need to do more to improve teaching. As a result, schools and districts around the country are now experimenting with a wide range of policies to improve teacher performance measures and use these to make high-stakes decisions such as dismissing low-performing teachers.
And here is the rub. The authors, recognizing the interest in dismissing low performers, conduct a simulation of such a policy and emphasize these results in their summary. But it would be a mistake to interpret even these careful simulation results as evidence about actual policies. The effects of actual policies never play out the way simulations suggest, because policies are rarely implemented as intended and the inevitable secondary effects are hard to predict.
There are substantial legal, political, and organizational problems associated with dismissing low performers. For example, in a simple system, many teachers would be fired unjustifiably as a result of imprecision in the performance measures—a lawsuit waiting to happen. High stakes associated with the tests will inevitably distort student scores and the assignment of students to teachers, worsening the measurement problem. A more elaborate evaluation system can address this measurement problem, but such systems are costly, and those costs are not considered here. Such an approach could also change the makeup of the profession, in both positive and negative ways.
There is good reason to think that dismissing more low-performing teachers would improve student outcomes, but the Chetty study is not designed to tell us much about that, or about any of the various policy alternatives. What it does provide is the best evidence yet that teachers matter a great deal and that we should continue looking hard for ways to improve teaching and learning in schools.
Douglas Harris is associate professor of educational policy studies at the University of Wisconsin-Madison.
Return to “Great Teaching” (Summer 2012)