Anchor Tests, Score Equating and Sex Bias

Abstract

This paper discusses the use of anchor tests (scaling tests) to bring two or more sets of scores to a common scale. Particular attention is given to the rescaling of school-based assessments against an external test or examination and to potential sources of bias in this procedure. The need for routine validity checks is emphasized, and a latent trait approach to constructing a statistical framework for tests and examination score equating is described and illustrated. Bias caused by rescaling school assessments against an inappropriate anchor test is illustrated using a 1984 attempt to rescale students' assessments in English against the Australian Scholastic Aptitude Test.