Sage Journals: Discover world-class research

Abstract

In research on individual and group differences, samples may be selected for study on the basis of whole scores that differ from the population mean. Also in clinical practice of assessment of exceptional children, a child is selected (i.e., diagnosed) on the basis of a total score followed by more specific evaluation based on the several part scores making up the whole score. These procedures produce regression to the population mean which can affect the accuracy and adequacy of part score interpretations, especially when each part score has a different correlation, within the population, with the whole score. The expected regression on a part score for a sample selected on a whole score that may differ from the population mean is derived. The effect of unreliability of the whole and part score scores upon the regression is also derived and an example of the results given.

Get full access to this article

View all access options for this article.

References

Jensen, A. R. and Reynolds, C. R. (in press). Race, social class, and ability patterns on the WISC-R. Personality and Individual Differences. (a)

Jensen, A. R. and Reynolds, C. R. (in press). Sex differences on the WISC-R. Personality and Individual Differences. (b)

Kaufman, A. S. (1982). Implications of WISC-R research for school psychologists. In C. R. Reynolds and T. B. Gutkin (Eds.), The Handbook of School Psychology, New York: Wiley.

Lord, F. M. and Novick, M. R. (1968). Statistical theories of mental tests. Reading, Massachusetts: Addison-Wesley.

Reynolds, C. R. and Jensen, A. R. (in press). Intellectual differences among blacks and whites matched on full scale IQ. Journal of Educational Psychology.

Regression Effects on Part Scores Based on Whole-Score Selected Samples

Abstract

Get full access to this article

References