Abstract
The use of standardized regression coefficients makes the comparison of causal models difficult to perform. Using Duncan's (1969) path model for white and black men, this article shows how one can use LISREL to analyze such models with unstandardized coefficients and compare them in terms of overall goodness offit, tests of differential specifications across groups, and assumptions of equality or inequality of corresponding regression coefficients. This article therefore demonstrates that researchers can use LISREL constructively, even if they do not wish to use all of LISREL's many statistical features.

This publication has 23 references indexed in Scilit: