The Problem of Range Restriction in Test Validation
Abstract
This paper examines some of the legal arguments advanced over the years with respect to the interpretation of the validity coefficient. The issue as to whether or not the validity coefficient is of a magnitude large enough to have "practical significant" is given considerable attention. The Equal Employment Opportunity Commission has argued that the validity coefficient is usually too low to have practical significance. Educational testing Service (ERS), however, has contended that the validity coefficient does have practical significance but it should be corrected for restriction in range. The Department of Justice claimed that range restriction correction formulas cannot be used because the assumptions underlying them cannot be met. The Department of Justice in its defense cited the Division 14 Principles for Validation and Use of Personnel Selection Procedures, which asserts the desirability of having validation samples be as similar as possible to the applicant pool. Thorndike's range restriction correction formulas and their underlying assumptions are carefully reviewed in this paper. These formulas are applied to data on actual selection instruments to obtain the estimated true validity coefficient that would be obtained if the validity coefficient was based on the total applicant population. Also, criterion referenced tests are discussed and suggested a viable alternatives to norm referenced tests, along with factors contributing to criterion biases.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 29, 1981
- Accession Number
- ADP001352
Entities
People
- Robert C. Mckenzie