Identifying the Effects of Classroom Observations on Teacher Performance
Hunter, Seth Baxter
:
2018-06-18
Abstract
In 2013 the Tennessee Board of Education adopted a policy introducing discontinuities in the number of observations assigned to teachers based on their prior level of effectiveness (LOE), a continuous composite measure of teacher effectiveness. After each policy-assigned observation observers should provide feedback so teachers can improve their performance. Given the importance of teacher effectiveness to short- and long-run student outcomes and increase in policy-assigned observations by many states since the early 2010s, it is important to understand whether more frequent classroom observations improve teacher performance. I address this need by identifying the effects of more observations on student growth.
Treating variation in the number of observations received as exogenous is problematic. Teacher and school administrator (i.e. observer) effectiveness plausibly influence the number of observations received and student growth scores. I overcome this endogeneity problem by exploiting policy-assigned discontinuities in the assignment of observations using a regression discontinuity design. To implement this research design I use a robust administrative dataset from 2012-13 through 2014-15.
I estimate four sets of effects: contemporaneous, extended, cumulative, and heterogeneous effects. While the Tennessee observation system theory of action asserts more observations per year should improve teacher performance (i.e. student growth) in the same academic year when observations are conducted, it is plausible teacher performance improvements take time. Thus, I identify the effects on performance measured one year after treatment (i.e. extended effects), and the effects of all observations received after 2011-12 through year t (i.e. cumulative effects). I also identify effects moderated by: teacher years of experience, teacher perceptions of the observation/ evaluation system, grade level (e.g. elementary), and measures of school administrator (i.e. observer) effectiveness.
There is no evidence more observations per year improve teacher performance as measured by student growth. This is true across short-term, longer-term, and heterogeneous effects. The dissertation ends with a discussion of potential explanatory mechanisms and policy implications.