Randomly Parallel Tests: Theory, Use, and Benefits

Session Type: Coordinated Paper Session

Abstract

This session will explore theoretical and practical perspectives of this old/new innovation of creating and using randomly parallel tests (including the large pools of items) or the resulting randomly parallel test forms. Presentations will cover the history and theory of randomly parallel tests, including the closely related topic of stratified randomly parallel tests. The definitions of sources of measurement error, both random and systematic, and for true scores are revisited. Also presented are recommended procedures that quantify various psychometric properties of such testing, such as reliability. Procedures for efficiently building such tests today will be showcased. Finally, simulations and empirical research will provide evidence for the use of randomly parallel tests in operational testing programs.

Sub Unit

• Individual Papers, Innovation Demonstrations, Coordinated Sessions, or Organized Discussions

Session Organizer

• David Foster, CAVEON TEST SECURITY

Presenters

• David Foster, CAVEON TEST SECURITY

• Robert Brennan, UNIVERSITY OF IOWA

• Won-Chan Lee, UNIVERSITY OF IOWA

• Stella Kim, University of North Carolina at Charlotte

• Craig Wells, UMass Amherst

• Jungwon Kyung, Umass Amherst

• Sergio Araneda, Caveon

• Andrew Marder, Caveon

Individual Presentations

• Randomly Parallel Forms: Some History, Theory, and Challenges - Robert Brennan, UNIVERSITY OF IOWA

• Psychometric Framework and Properties of Randomly Parallel Forms - Won-Chan Lee, UNIVERSITY OF IOWA; Stella Kim, University of North Carolina at Charlotte

• What is the Generalizability of Randomly Parallel Forms? - Craig Wells, UMass Amherst; David Foster, CAVEON TEST SECURITY; Jungwon Kyung, Umass Amherst

• Evaluating the Test Security Benefits of Randomly Parallel Tests: Protecting Against Pre-Knowledge - Sergio Araneda, Caveon; David Foster, CAVEON TEST SECURITY; Andrew Marder, Caveon

Discussant

• Wayne J. Camara, LSAC