Teacher Evaluation, Instructional Practice and Student Achievement: Evidence from the District of Columbia Public Schools and the Measures of Effective Teaching Project

Abstract

I dedicate this dissertation to my many families who have offered me so much support through this process-my biological family, my academic family, and my friends. I especially thank Charles for his love and patience this past year. I could not have done this without you. vi ACKNOWLEDGEMENTS I am immensely grateful to my advisor, Jim Wyckoff, for the excellent advising and endless patience he provided while I wrote this dissertation and over the past five years. For co-leading our DCPS research project and his sage advice, I am also very grateful to Tom Dee. I also thank my dissertation committee, Ben Castleman, Julie Cohen, and Bridget Hamre for their feedback and support during the dissertation process, and Allison Atteberry, Matt Kraft, and Roddy Theobald for helpful comments and suggestions on previous drafts of this work. I would also like to thank the District of Columbia Public Schools for providing data used in this dissertation, and Scott Thompson, Kim Levengood, Luke Hostetter, Alden Wells and Austin Zentz of DCPS for answering questions about the data and about IMPACT. I also use data from the Measures of Effective Teaching Project, and I appreciate the work done by the original MET researchers and staff at ICPSR as well.

Enacting the Rubric: Teacher Improvements in Windows of High-Stakes Observation

“…This highlights the statistically significant effects seen for Teach 3, 6, and 8 are driven by the fact that these coefficients are more precisely estimated and not necessarily larger in magnitude. Our results are consistent with those of Adnot (2016), who found that the majority of variation in evaluation scores in the IMPACT program can be explained with a single factor that encompasses all nine of the evaluation components.…”

“…As found in other studies (Adnot 2016), these rubric components are highly correlated, making our hypotheses on observation components highly dependent. We have adjusted the significance levels using a Bonferroni correction because it is very likely that an improvement in one rubric component will also lead to an improvement in another.…”

“…Two studies found that the specificity of the language in observation rubrics may influence teachers to strategically take up practices that are easiest to target. Adnot (2016) found that low-performing teachers facing a dismissal threat in DCPS improved the most on highly specific rubric practices, ranging from a statistically significant effect size of 0.22 to 0.62 standard deviations. Another study found that more explicit, written descriptions of instructional domains correlated with teacher improvement on that rubric component (Kane et al 2010).…”

New Assessments and Teacher Accountability: Lessons for Teachers’ Practice

“…Meanwhile, teachers who perform exceptionally well are eligible for substantial financial rewards. Multiple studies have demonstrated that this incentive structure leads to improved overall performance for teachers on both ends of the performance distribution, with measurable gains on the TLF (Adnot, 2016; Dee et al, 2021; Dee & Wyckoff, 2015). Phipps and Wiseman (2021) looked specifically at the observation component of the evaluation system, and found that the expectation of a classroom observation—proxied by the number of days remaining in the observation window at the point of observation—was associated with better teaching performance.…”

“…Phipps and Wiseman (2021) looked specifically at the observation component of the evaluation system, and found that the expectation of a classroom observation—proxied by the number of days remaining in the observation window at the point of observation—was associated with better teaching performance. Adnot (2016) took a more nuanced look at incentive effects by examining teachers’ improvements across each of the rubric standards; teachers who were at risk of involuntary separation made meaningful improvements, but gains were concentrated on the most-prescriptive and the least-difficult teaching domains.…”