Rethinking Consequential Validity: Linguistic Deconstruction of "Gold Standard" Math Story Problems to Uncover Implicit Gender Bias

• In Event: 34.076 - Key Issues in Validity Research and Theory

Abstract

This work represents a novel exploration of the most micro unit of analysis in measurement—the test item—as a means to expose macro-level social discord. To explore the ways implicit bias and microaggressions continue to permeate the existing body of standardized test questions, a rigorous procedure for the deconstruction of math story problems was developed. Preliminary coding of the data revealed interesting trends in how implicit bias is manifested in the item set. These findings radically challenge fundamental assumptions made about how the measurement community evaluates test questions for bias and sensitivity. They extend the conversation about testing and measurement, especially when it comes to the use of scores in student promotion/placement, teacher retention, and allocation of funding.

Author

• Gabriella Gorsky, University of Washington - Seattle