Considering Fairness Even When Minority Sample Sizes Are Small

• In Event: Assessment in the Higher Education Context

Abstract

Examining test fairness in undergraduate STEM classrooms is a difficult problem. This study examines evidence of fairness when minority sample sizes are too small for DIF. A concept inventory was administered in an introductory mechanical engineering course at nine universities (n = 577). Only 1.82% of the sample identified as African American, 1.82% as Multiracial, and 3.64% as Hispanic/Latino. To examine evidence of fairness, we utilized an explanatory sequential mixed method approach to first identify score discrepancies and item performance, then qualitatively explored reasons behind the score differences. Our findings indicate the context of problematic items assumed personal experiences that not all cultures of students would have experienced and test-taking strategies that not all students would have been trained in.

Authors

• Kerrie Ann Douglas, Purdue University

• David Waller, Purdue University

• Marcus Akeem Rose, Purdue University