Examining Teacher Effectiveness Using Classroom Observation Scores: Evidence From the Randomization of Teachers to Students

Education policymakers have long recognized the importance of having highly effective teachers in all of our nation’s classrooms. Among recent policy efforts, the federal Race to the Top (RTTT) competition and the No Child Left Behind (NCLB) waivers have highlighted the importance of ensuring highly effective instruction as a central policy consideration, primarily through increased attention to teacher evaluation systems. This policy focus follows from research findings that have emphasized the importance of teacher quality for student achievement above other school-level characteristics (Aaronson, Barrow, & Sander, 2007; Goldhaber, 2002; Rivkin, Hanushek, & Kain, 2005; Rockoff, 2004). Even as federal policy efforts have promoted more rigorous approaches to evaluating teacher performance, little consensus exists about the most salient measures of teacher effectiveness for the purposes of teacher accountability (such as high-stakes tenure decisions) and compensation (such as merit-based pay programs). Traditional measures of teacher performance, such as years of experience or educational attainment, play little role in student achievement (see Steinberg & Sartain, in press, for a summary). Newer efforts incorporate student test score data through valueadded measures of teacher performance, yet such measures can be statistically unreliable over time (Goldhaber & Hansen, 2010; McCaffrey, Sass, Lockwood, & Mihaly, 2009), and prone to severe bias due to the nonrandom assignment of students 537551 EPAXXX10.3102/0162373714537551Garrett and SteinbergExamining Teacher Effectiveness Using Classroom Observation Scores research-article2014