Co-Principal Investigator: Dinerstein, Michael

Purpose: The purpose of this project is to explore how the malleable incentives implicit in teacher tenure policy are related to teacher practice and student achievement by estimating whether the predictiveness of the incentivized measures changes after they become incentivized.

Project Activities: Using a difference-in-difference quasi-experimental design, the researchers will complete an analysis of changes in student outcomes after the New York City Department of Education (NYCDOE) instituted a change to the teacher tenure policy in 2009-2010.

Products: Products will include information about the relationship between the teacher tenure policy and student outcomes. The team will produce peer-reviewed publications.

Structured Abstract

Setting: Researchers will examine data from all non-charter NYCDOE schools serving general education students.

Sample: The project sample consists of 19,806 teachers teaching grades 4-8 and mathematics, ELA, or both and 537,534 students taught by these teachers. Teachers and students are observed up to 4 different years. Because the sample includes all NYC teachers and students meeting these criteria, we have broad representation across many race/ethnicity categories and SES levels. The sample is intended to represent the population of urban students in the U.S.

Malleable Factors: The researchers are focusing on a specific 2009-10 change to the teacher tenure policy in NYC that increased teachers' incentives to produce high student test scores. Prior to the change, tenure rates were high and had little relation to measured teacher value-added. NYCDOE then changed the policy such that tenure decisions would depend closely on measured teacher value-added and tenure rates fell.

Research Design and Methods: The researchers are using a quasi-experimental design: difference-in-difference to compare changes in student outcomes before and after NYC changed their teacher tenure policy. The control group is the set of teachers with tenure prior to 2009-10. These teachers are unaffected by the change in tenure policy because it only affects yet-to-be-tenured teachers, and thus their condition is the same before and after the policy change. The treatment group is the set of teachers with fewer than 4 years of experience in NYC. These teachers are largely untenured and thus the set of teachers prior to 2009-10 face very different incentives than the set of teachers 2009-10 and later.

Control Condition: The control group is the set of teachers with tenure prior to 2009-10.

Key Measures: The measures will include current-year test scores as well as the unincentivized measures: future test scores, current and future attendance rates, current and future grades on both tested and untested subjects.

Data Analytic Strategy: To see how the intervention affects the predictiveness of value-added the team will compare the actual treatment effects on unincentivized measures to the treatment effects that would be predicted by the underlying (untreated) cross-sectional relationship between the incentivized measure (current-year test scores) and other measures multiplied by the actual treatment effect on the incentivized measure. They will use cross-sectional regressions to estimate the underlying relationships and difference-in-difference regressions to estimate the policy impacts on students with different teacher types (treatment or control).