Empirical Estimates of the Validity of Four Measures of Change

Abstract

The measurement of change has received considerable attention in the literature of psychology and education, with special consideration being given to the psychometric properties of simple gain scores and related measures. Three alternatives to the simple gain score are the residual difference, the base-free measure of change, and the Lord/McNemar estimate of true gain. Estimates of the criterion-oriented validity of these four measures were obtained from two empirical studies, and changes attributable to a course of instruction in mathematics were moderately correlated with external criteria, with values ranging from .30 to .54. Psychometric theory was used to derive a guideline useful in predicting the rank order of the validity coefficients. The coefficients obtained from the studies were highly consistent with the theoretical predictions.