Statistics Seminar by Marie Wiberg from Umeå University

Abstract

Standardized high-stakes tests typically use different test forms to ensure test security. Test equating methods are statistical procedures that adjust for test form differences to make it indifferent for test takers which form they are given. Equating test scores across diverse groups is a fundamental challenge in educational and psychological assessment. In times of societal change or crisis, shifts in test-taker groups regularly occur, which may influence the equating process. The aim of this presentation is to address the problem of equating test scores under nonequivalent group conditions. Methodological challenges arising from ability differences between groups and differences in test forms are explored, and recent developments designed to address these issues are discussed. Finally, practical advice is given on how to achieve fair and comparable test scores across nonequivalent groups.