posted on 2021-06-07, 22:15authored byHarry A. Layman
This research has two primary goals. The first is to develop a useful framework for designing rubrics to improve the utility of feedback and the reliability of scoring for critical thinking assessments that use constructed response items. The second is to demonstrate and explore the practicality, effectiveness, strengths, and weaknesses of this approach as applied to some specific data sets. The use of constructed response (CR) for educational assessment has been advocated for decades (Gulikers, Bastiaens and Kirschner, 2004; Palm, 2008; Wiggins, 1990). The primary benefits claimed are more authentic measurement and better feedback to students and teachers. More authentic measurement includes the notion that the construction of a response is more cognitively challenging and more direct evidence than the indirect evidence from selecting among predefined choices. Better feedback is generally limited in practice, however, when use of CR items relies on holistic scoring, generic rubrics, and a regimen of scorer training and calibration to attain consistent and generally valid measurement. The resulting broad, multifactor classification levels are unable to convey response-specific feedback (Bejar, 2017). This research postulates the use of a rubric design framework for the creation of itemspecific, content-centric rubrics for ssessment items that have right and wrong, better and worse possible responses. The framework establishes a uniform mechanism for identifying essential elements of item responses, with explicit weights for varying degrees of correctness and completeness and standardised approaches to calculating overall scores, subscores, and scaled scores. Resultant score reports can provide explicit feedback to response elements present, absent, or less than complete that explicitly justify and explain score differences between responses. Successful use of this rubric design framework promises CR assessment for critical thinking or argumentative writing items that will have score reports able to provide detailed feedback to students and defensible scoring outcomes, with the potential for improved interrater reliability.