Value-Added Models, Outcome-Based Teacher Performance, and the Teaching-Learning Process

Current educational reform efforts are focused on enhancing school and teacher accountability as a means of improving student learning. One policy initiative resulting from these efforts is the inclusion of outcome-based teacher performance (OTP) as a component of teacher evaluations. OTP is defined as a teacher's contribution to the learning outcomes of her students. An increasingly popular method for measuring OTP is a set of statistical techniques known as value-added models (VAMs). Despite their popularity, the validity of VAM estimates of OTP has not been fully established. This dissertation contributes to the ongoing validity debate by evaluating the construct validity of VAM estimates. Specifically, construct validity is evaluated by examining the theory of the process by which teachers contribute to student learning that is implicit to VAM estimation of OTP. Embedded within each VAM is a set of theoretical implications about the teaching-learning process. In addition, the identifying conditions necessary for the estimation of OTP from observational data impose restrictions on the teaching-learning process and therefore have theoretical implications. Evaluation of these implications indicates that most are inconsistent with theoretical, empirical, and logical evidence how teachers contribute to student learning outcomes. This evidence weakens the validity of the claim that VAMs can be used to measure OTP. The degree to which this claim is weakened, however, remains a question for future research.