Teacher Evaluation: What Do We Want to Measure?
Note: Andrew McEachin, an assistant professor of educational policy analysis and program evaluation at NC State University, is guest posting this week.
I want to thank Rick for giving me the chance to (at least partially) fill his large shoes. The current education policy landscape has no shortage of interesting topics. I thought about starting the week by blaming the Common Core for my tomato plants' lack of output (since recent commentary on the Common Core has gotten a tad out of hand). But instead, I'll spend the first three days talking about teacher and school accountability and the last day on the role summers play on accountability and education policy.
A good teacher can forever change a student's life. It is in this vein that teacher evaluation and accountability policy has continued to be analyzed and discussed in the public as well as the research community. Two specific recent examples come to mind. The first is a study by Drs. Morgan Polikoff and Andy Porter funded by the Gates Foundation which garnered a lot of discussion in the popular press and social media (example reactions here and here). In this study, the authors found a limited, to no, association between teachers' student achievement gains and classroom observations, student surveys, or measures of instructional alignment. The second is a report released by the Brookings Institute that evaluated the validity and reliability of teacher evaluation systems in four districts. While the authors found that all four evaluation systems passed a validity and reliability assessment, they also found that observational measures of teacher practice have similar measurement issues to measures of teachers' impact on students' achievement--e.g., teachers with higher achieving students also get higher scores on the observational protocols. These two papers highlight a few important considerations for future policy work on teacher evaluation.
First, it is unclear, to me at least, what measure of teacher quality should be used to benchmark other measures, or if such a benchmark measure currently exists. Those who like measure A will use it to gauge the quality of measures B, C, D, and so on. Take the Polikoff paper as an example; it is unclear whether the measures of instructional alignment, classroom observations, student surveys, or student achievement growth should serve as measure A. In other words, there are a number of normative judgments that must be made about which measure, or measures, capture the key underlying traits of a good teacher.
Second, a number of subjective decisions must be made about which short- and long-term student outcomes to prioritize in a teacher evaluation system. We know that teachers have a lasting impact on students' achievement and labor-market outcomes, but we know less about how teachers influence students' cognitive and non-cognitive abilities. Granted, the latter two measures are quite difficult to measure in a typical school setting. However, the narrow focus of teachers' instructional quality and impact on students' achievement may miss the more important role teachers play in the development of students' social, emotional, and intellectual development.
Third, while nearly everyone agrees that context matters in teacher evaluations, designing a system to address context often leads to difficult discussions. The Brookings paper, as well as work by Cory Koedel et al (among others), discuss the importance of removing factors out of a teacher's control from the evaluation of her practice and impact of student achievement. Adjusting for differences among teachers' classrooms and schools, however, is often seen as setting different standards for different subgroups of students (more on this later in the week). Yet by not adjusting the evaluations for non-school related contextual differences, policymakers are making it even harder to staff the schools that need the best teachers.
And finally, the discussions about the design of teacher evaluation systems often do not spend enough attention on curating the current human capital that exists within the public education system. The best schools in the country, whether they are public or charter, voucher or magnet, spend a considerable amount of time on fostering a learning and professional development environment within their school building. Simply holding teachers accountable for their students' outcomes is unlikely to have a meaningful impact on the achievement gap without first ensuring every teacher has the knowledge and skills needed to maximize their students' outcomes.
It is clear that the policy landscape (e.g., the ESEA waivers) has quickly moved past asking "Should we hold teachers accountable?" to "How should we hold teachers and school accountable?" Yet, it appears that all too often we do not start with the two most important questions: What qualities and practices characterize the ideal teacher? What are the skills and knowledge we want students to obtain from our schools? In trying to answer these questions, it is clear that not only are there many dimensions of instruction and student outcomes to be measured, but also that parents' and other stakeholders' preferences of which dimensions to measure will vary quite dramatically. Tomorrow I focus on a few interesting ideas that take these issues head-on when designing a teacher accountability model.