Multivariate approaches to the evaluation of programs for mildly mentally retarded persons

Multivariate analysis usually connotes a set of complex statistical models. It can also represent a way of thinking about research in which behavioral outcomes are seen as having multiple facets and resulting from multiple antecedents. Three areas of research on mentally retarded children that need conceptual multivariate analysis were identified: (a) the characterization of learners, (b) the description of process features of an instructional setting, and (c) the depiction of change. Examples were provided, drawing on a recent National Academy of Sciences report. Finally, statistical developments were described that are appropriate to the data that these conceptualizations yield.