Cross-domain models represent a genre of social science models used to analyze individual change in multiple dimensions simultaneously. Cross-domain models are multilevel structural modeling methods. First, the technique is explained. Second, it is used to evaluate student success in pre-engineering studies.;Academic performance measures were obtained from a sample of 868 pre-engineering students who enrolled in 1991. The measures were collected at three separate time points in two domains: mathematics and history. Growth in each domain was measured as a function of change in course grades over three consecutive terms. Several alternative structural models were constructed to reflect different student performance in each domain. These level 1 models were evaluated. The best fitting were used to construct the level 2, cross-domain growth models. The cross-domain growth models compared student growth in the scientific realm with that in the liberal arts arena. Using multiple goodness-of-fit measures, a suitable model was identified. This model was used to evaluate student success in obtaining admission to the College of Engineering. Thus, the cross-domain model was expanded to include distal outcomes. The final, two-level, cross-domain model demonstrated that student growth in both science and liberal arts contributes significantly to student success.;Cross-domain modeling is a very versatile methodology. It permits investigators to evaluate change in multiple dimensions. It is not restricted to linear growth but can be applied to curvilinear trajectories. Covariance structures of occasion-by-occasion level 1 measurement errors can be modeled explicitly. Furthermore, different error covariance structures can be hypothesized for separate domains. This dissertation provides one example of this analytic technique. |