Can Performance-Based Assessments Be Reliable and Valid? Findings From a State Pilot

• In Event: Roundtable Session 117
  In Roundtable Session: Perspectives on Validity

Abstract

In 2010-11 and 2011-12, [State] completed pilots of curriculum-embedded performance assessments in English language arts, mathematics, and science with approximately 140 teachers across 31 schools, and completed research to explore the validity and reliability of these performance tasks as components in a multiple-measures assessment system. The purpose of this paper is to present findings regarding the reliability and validity of the performance based assessments used in the state pilot, including the results of three sets of studies: 1) Generalizability studies; 2) Regression analyses to ascertain the relationships between performance assessment scores and other measures of student achievement (high school GPAs, exit exam scores); and 3) Content alignment and content validity reviews.

Authors

• Ruth Chung Wei, Stanford University

• Ken Mathew Cor, University of Alberta

• Nicole Leigh Arshan, Stanford University

• Raymond Lee Pecheone, Stanford University