Interpretational Confounding of Unobserved Variables in Structural Equation Models

Abstract

The problem of assigning empirical meaning to unobserved variables in structural equation models is discussed. Interpretational confounding is discussed as the assignment of the other than a priori assigned empirical meaning of an unobserved variable. Hypotheses conceming the possibility of interpretational confounding as a concomitant of a lack of point variability in unobserved variables are specified, and corresponding chi-square statistics are given. Numerical illustration is provided