Panel Paper: How Did It Get This Way? Disentangling the Sources of Teacher Quality Gaps Across Two States

Saturday, November 4, 2017
Water Tower (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Dan Goldhaber1, Vanessa Quince1 and Roddy Theobald2, (1)University of Washington, (2)American Institutes for Research

There is mounting descriptive evidence of substantial “teacher quality gaps” (TQGs) between advantaged and disadvantaged students in U.S. public schools. Recent evidence from longitudinal data on public schools from North Carolina and Washington (Goldhaber, Quince, & Theobald, 2016) demonstrates that TQGs have existed in every available year of data in each state and for each observable measure of student disadvantage (i.e., race/ethnicity and poverty level) and teacher quality (i.e., experience, licensure test scores, and value added).

In this follow-up study, we study four different processes—the hiring of new teachers into their first jobs, the movement of teachers between schools and districts, the attrition of teachers from different types of schools, and changing student demographics—that may be contributing to these TQGs, and attempt to disentangle the extent to which each process contributes to overall TQGs. The first three of these processes have all been well-studied in the teacher labor market literature, but we do not know how important each process is in contributing to TQGs. Beyond these teacher labor market processes, changing student demographics across different classrooms, schools, and districts could also contribute to TQGs, particularly given recent evidence of increased income and racial segregation across schools and school districts.

 To disentangle the contribution of each of these processes to overall TQGs, we adopt a three-stage empirical approach. First, we leverage longitudinal data in both North Carolina and Washington and estimate two sets of relationships: a) the relationship between different measures of teacher quality (experience, licensure test scores, and value added) and the probability that teachers are hired into, move between, or leave schools that serve different percentages of disadvantaged students (measured by race/ethnicity and poverty level); and b) the relationship between existing teacher quality measures within schools and districts and subsequent changes in student demographics. Second, we develop a dynamic model of the assignment of students to teachers in different school settings that is a function of all four processes described above and builds on a simpler model described in detail in Goldhaber, Lavery, and Theobald (2016). Finally, we use the estimated relationships from step 1 and the dynamic model developed in step 2 to quantify the contribution of each process to overall TQGs through simulations described in Goldhaber et al. (2016).

Findings from this study will inform the process that policymakers should seek to influence to close TQGs. For example, if patterns in teacher hiring explain most of the TQGs, policymakers could develop recruitment policies to attract high-quality teachers to disadvantaged schools. But if patterns in teacher attrition drive the observed inequities, policymakers may wish to focus on retention policies designed to keep high-quality teachers in disadvantaged schools