Examining the Validity of Claims Based on Constructed-Response Task Performance Consistency on Parallel Exam Forms

• In Event: Topics in Validity Evaluation

Abstract

Many assessment programs that use constructed response (CR) items design test forms to be parallel by utilizing the same type of prompt and scoring rubric across forms; thus the challenge presented to the test taker remains constant. Typically an identical or similar holistic or analytical rubric with points awarded for completing specific elements is used.
This study presents a methodology applied to the AP® World History Exams for evaluating consistency of skills elicited by parallel prompts for CR items with analytical rubrics. The results of studies using this methodology can be used to inform test development and design of task models and also address directly the validation of claims about student performance to support the Interpretation/Use Argument (Kane, 2013).

Authors

• Rosemary A. Reshetar, The College Board

• Barbara S. Plake, University of Nebraska - Lincoln