An Argument-Based Validation Study of the Teacher Performance Assessment in Washington State

Abstract

This study examines the validity assumptions of the Teacher Performance Assessment (TPA) using data collected from teacher candidates, mentor teachers, university supervisors and university faculty in two programs at one university during the 2012 field test in Washington State. Applying the work of Michael Kane (2006) on argument-based validation, this study developed interpretations and assumptions of TPA test score use using the following five inferences: Construct Representation, Scoring and Evaluation, Generalization, Extrapolation, and Decision Making. This multi-method study utilizes survey, case study, and test score data. The overarching research question that guided the study was “Is the TPA a valid measure for determining teacher readiness?” The overall findings suggest that the operationalized construct of readiness is stable but scores are not generalizable across populations and guidance was not in place regarding score meaning and use prior to the field test. Low correlation between the TPA and university instruments provided divergent evidence for the use of TPA scores, indicating that decisions made based solely from TPA scores may not be reliable.