Structural modeling of mixed longitudinal and cross-sectional data.

In this paper we describe some mathematical and statistical models for dealing with changes over age. We concentrate specifically on the use of a structural equation modeling (SEM) approach (using computer programs like LISREL) to deal with issues of: (1) group differences in regression parameters, (2) differences in longitudinal and cross-sectional results, (3) differences due to longitudinal attrition, and (4) mixtures of these problems. To illustrate these ideas we use data from a previous study of hypertension and intellectual abilities (from Schultz, Elias, Robbins, Streeten, and Blakeman, 1986).