Rethinking Value-Added: Short and Longer-Term Teacher Effects on Student Achievement
I examine the extent to which teachers have longer-term effects on student achievement using administrative data from Miami-Dade County Public Schools (MDCPS). I define longer-term effects as the effects teachers have on student achievement in the years after they teach their students. Using a flexible statistical model, I obtain teacher-specific short-term and longer-term effects. I also estimate the extent to which short-term and longer-term effects relate to one another, on average. Results suggest that there is meaningful variation in teacher longer-term effects and that short-term and longer-term effects are not perfectly correlated; traditional value-added models assume short-term and longer-term effects are perfectly correlated. Finally, I find that having a master’s degree or Ph.D. is associated with higher longer-term effects when estimating the model for math teachers; I do not find an association between graduate-level degree and longer-term effects for ELA teachers.
Assessing the Validity and Stability of Short and Longer-Term Teacher Effects
In this paper, I assess the validity and stability of short-term and longer-term teacher effect estimates. I assess validity by examining whether future teachers predict the past test score gains of students they have not yet taught. This particular test is designed to provide evidence of student sorting bias that could potentially invalidate the teacher effect estimates. I then assess the stability of teacher effect estimates by considering the stability of teacher effects across different cohorts of students and the stability of teacher effects across math and English language arts within a given cohort of students. Results show that teacher effect estimates suffer from sorting bias. Although this is problematic, it suggests the need to understand whether the sorting bias is large enough to invalidate teacher effect estimates; this is an area of future research. With respect to stability, there is substantive overlap of teacher effects—both short-term and longer-term—across student cohorts, which suggests that the estimates carry some true signal of teacher quality and are reliable. Overlap of teacher effects across subjects is also non-trivial, but it is less stable than the across-cohort stability. These results suggest that teachers have different strengths in different subjects.
Court-Ordered Finance Reforms in the Adequacy Era: Heterogeneous Causal Effects and Sensitivity (with Kenneth A. Shores)
We provide new evidence about the effect of court-ordered finance reforms that took place from 1990 to 2010 on per-pupil revenues and graduation rates. We account for heterogeneity in the treated and counterfactual groups to estimate the effect of overturning a state’s finance system. Seven years after reform, the highest poverty quartile in a treated state experienced an 11.5 to 12.1 percent increase in per-pupil spending and a 6.8 to 11.5 percentage point increase in graduation rates. We subject the model to various sensitivity tests, which provide upper and lower bounds on the estimates. Estimates range, in most cases, from 6 to 12 percentage points for graduation rates.
The Effects of No Child Left Behind on Children’s Socio-Emotional Outcomes (with Camille Whitney)
Many people have worried about possible adverse effects of high-stakes testing on socio-emotional outcomes. This paper uses a difference-in-differences approach to investigate the effects of the introduction of high-stakes testing via the No Child Left Behind Act on socio-emotional outcomes. The outcomes are from the Early Childhood Longitudinal Survey – Kindergarten Cohort of 1998-99, a nationally representative longitudinal survey. Outcomes include student-rated externalizing, internalizing, interest and competence in tested subjects and in school in general, and teacher-rated approaches to learning. Analyses find no statistically significant effects on the socio-emotional outcomes included in this analysis for the average student. However, there are differences in treatment effects by race, gender, and socio-economic status. Hispanic students have particularly positive treatment effects. Schools and policy-makers need to consider how to promote socio-emotional wellbeing among all student subgroups as they revise standardized testing regimes