Category: Spectrum Bias

Spectrum bias is present in diagnostic studies that do not randomly assign participants to groups. Instead, groups on opposite ends of the spectrum are handpicked (e.g., a group of children with moderate to severe language disorders and a group of typically developing children) and a diagnostic test is administered. Since individuals whose diagnostic status is not obvious (e.g., mild language disorder) are not included in the study, the diagnostic accuracy cannot be generalized to a more representative population that includes such individuals. This results in decreased diagnostic accuracy. Because of spectrum bias, the validity of the diagnostic test is compromised.

Relevant Research: Eligibility Criteria for Language Impairment: Is the Low End of Normal Always Appropriate?

Leaders Project

Source URL: View this document on the ASHA website

Many clinicians label a child as language impaired (LI) or typically developing (TD) based on arbitrary cutoff scores (1.5 SD below the mean or 1 SD below the mean) on standardized tests. However, as demonstrated in this article, many standardized tests do not even provide information about validity and reliability.

Read More

Research Text: Why Are So Many Minority Students in Special Education? Understanding Race and Disability in Schools

Leaders Project

Source URL: View this book on the Teacher’s College Press website

This book addresses the constellation of factors that have contributed to the misidentification of minority/culturally and linguistically diverse children as needing special education services. It also provides suggestions for improving the special education referral process.

Read More