Composite Versus Component Scores: Consistency of School Effectiveness Classification

Abstract

This study examined consistency of school effectiveness classification when using a composite score as the school effectiveness index (SEI) versus using individual subject area scores. Employing the five components of the Louisiana Graduation Exit Examination (GEE) as dependent variables, schools were classified as effective, average, or ineffective. Two comparison studies were conducted: First, school classification results derived from subject area scores were compared to those of the composite and each other (n = 324); second, classification results of within-school split samples were compared for the composite and components (n = 255). A subcomposite of two subjects was also examined. Analysis was done by computing agreement ratios and kappa coefficients as well as correlations of raw scores and residuals. In general, the overall composite-score model yielded the highest relationship and consistency with all component-score models, and the highest split- sample consistency. Results indicate that a single subject area or grade level provides a less consistent and more narrow perspective of school effectiveness (academic achievement) than does a subcomposite made up of only two subject areas. A composite score composed of as many subjects as possible provides a broader, more consistent basis from which to construct a model to judge school effectiveness.