In the first chapter of my dissertation, I examine the robustness of typical teacher quality measures to alternate ranking systems factoring in the dispersion of value-added. The typical measure used by researchers and school administrators to evaluate teachers is based on how the students' achievement increases after being exposed to the teacher, or based on the teacher's ``value-added''. When teacher value-added is heterogeneous across her students, then the typically used measure reflects differences in the average value-added the teacher provides. However, researchers, administrators, and parents may care not just about the average value-added, but also its variance. Encouragingly, ranking systems factoring in the dispersion produce similar rankings as the ranking system based only on the mean. In the second chapter, I examine the effect of measurement error in the dependent variable on quantile regression, because unlike OLS regression, even classical measurement error can generate bias. I examine the pattern and size of the bias using both simulation and an empirical example. The simulations indicate that classical error can cause bias and that non-classical measurement error, particularly heteroskedastic measurement error, has the potential to produce substantial bias. Using restricted access Health and Retirement Study data containing matched IRS W-2 earnings records, I examine whether estimates of the returns to education statistically differ using a precisely measured and mismeasured earnings variable. I find that returns to education are over-stated by roughly 1 percentage point at the median and 75th percentile using earnings reported by survey respondents.In the third chapter, my coauthors and I investigate how the precision and stability of a teacher's value-added estimate relates to student characteristics. We find that the year-to-year stability of teacher value-added estimates can depend on the previous achievement level of a teacher's students. The stability level of the estimates are typically 25\% to more than 50\% larger for teachers serving initially higher performing students. We offer a policy simulation demonstrating that teachers who serve low-achieving students may be differentially likely to be the recipient of sanctions in a high stakes policy based on value-added estimates.
Read
