Public/Catholic Differences in the High School and Beyond Data: A Multigroup Structural Equation Modeling Approach to Testing Mean Differences

Abstract

Previous research based on the large, nationally representative High School and Beyond (HSB) study has compared senior year achievement test scores for public and Catholic high school students after controlling for background variables and sophomore year test scores. These analyses, however, were based on traditional applications of multiple regression with its implausible assumptions that variables are measured without error and that residuals are uncorrelated. The present study demonstrates tests for mean differences on latent constructs using the LISREL approach to multigroup structural equation modeling (SEM) for this substantively important issue. Public/Catholic differences, even after controlling for background and sophomore outcomes, favored Catholic high school students on senior year outcomes (achievement, educational aspirations, and academic course selection) and subsequent college attendance. These public/Catholic differences were similar for students differing in race, SES (social economic status), and initial ability. Public/Catholic differences in achievement, educational aspirations, and college attendance were, however, apparently mediated by the academic orientation of course selection. The flexibility and advantages—but also the limitations—of this multigroup SEM approach are discussed.