Multivariate Regressed Scores

Abstract

Recall from Section 5.5.2 that Kelley’s regressed score estimates of true score are easily extended to estimates of universe score in univariate gen-eralizability theory. The basic equation is

$${\hat \mu _p} = (1 - E{\rho ^2})\mu + E{\rho ^2}{\bar X_p}$$

((12.1))

where p stands for an object of measurement. The principal way in which univariate generalizability theory extends Kelley’s work is through the many different D study designs and universes of generalization that may be employed to characterize Eρ ² and arrive at an estimate of it.