Eric.ed.gov – Missing Data and Mixed Results: The Effects of Teach For America on Student Achievement Revisited

This paper revisits existing experimental work on Teach For America (TFA) and extends it by examining treatment effects across the distribution of student achievement. TFA is a rapidly expanding teacher preparation program that currently serves over half a million students in low-income districts across the country. Previous research results did not have notable variation by subgroup. Estimates were inaccurate due to the treatment of a non-response code as a valid response value. Revised estimates confirm positive effects for math and not reading, but show that TFA teachers were especially effective for African American students, but not Hispanics, and for females, but not males. In addition to examining differences across subgroup, others have argued that a distributional approach is important for thoroughly investigating policy interventions because examinations focused solely on mean impacts might obscure large countervailing differences in program impacts that offset one another. To deepen the understanding of the effect of TFA on student achievement, this study investigates distributional as well as mean impacts. New distributional results reveal TFA teachers are especially beneficial for the upper-middle of the distribution of math achievement, but not the upper and lower tails. Tables and figures are appended.