We examine behavioral responses to an incentive program that offers high-performing teachers in ten school districts across the country $20,000 to transfer into the district's hardest-to-staff schools. We discuss behavioral responses to the program on high-performing teachers’ willingness to transfer (supply) and the effect of the transfer offer on the internal dynamics of the receiving schools (demand). We found low take-up rates among the 1,514 high-performing teachers who were offered the incentive, with minimal sorting on observable characteristics. Within the new schools, transfer teachers were less likely than their counterparts in a randomized control group to require mentoring and more likely to provide mentoring themselves. No significant differences occurred in school climate, collegiality, or the way in which students were assigned to teachers, but evidence indicates that principals may have strategically assigned existing teachers to grades in both treatment and control schools in response to the quality of the incoming teachers.

Randomized trials are a common way to provide rigorous evidence on the impacts of education programs. This article discusses the trade-offs associated with study designs that involve random assignment of students within schools and describes the experience from one such study of Teach for America (TFA). The TFA experiment faced challenges with recruitment, randomization of students, and analysis. The solutions to those challenges may be instructive for experimenters who wish to study future interventions at the student or classroom level. The article concludes that within-school random assignment studies such as the TFA evaluation are challenging but, under the right conditions, are also feasible and potentially very rewarding in terms of generating useful evidence for policy.