A model-based approach to validating education indicators using multilevel structural equation modeling

This article considers an approach to validating the selection of education indicators by incorporating them into a multilevel structural model and using the estimates from that model to engage in policy-relevant simulations. Multilevel structural equation modeling was applied to a subsample of the first follow-up of the National Education Longitudinal Study of 1988 (National Center for Education Statistics, 1988) to demonstrate the potential of this approach. Focus of attention was on science education indicators. A within-school model of science achievement was linked to a between-school model of the academic press of the school. Separate estimation of these models revealed adequate fit to the data after minor modifications. The multilevel model also showed adequate fit to the data. Utilizing the reduced form of the full multilevel model, predictive validity of the model was studied by gauging movements in various outcome indicators as a function of changes in policy-relevant input indicators. The article closes with a discussion of the limitations of the proposed modeling approach, the potential for future model development, and the implications of this approach for quantitative modeling within the domain of education policy.