This article defines test bias and reviews research methods most aptly applied postpublication to educational and psychological tests. Best methods for estimating bias in criterion-related or predictive validity and determining bias in construct validity across groups are explained. (Author/JDD)

This paper argues that using intelligence as an aptitude benchmark in defining reading disability conceals unsupported assumptions about educational potential and makes it difficult to differentiate the cognitive characteristics of dyslexic children from those of other poor readers. The use of a more educationally relevant aptitude measure such as…