ABSTRACT
Not surprisingly, most of the literature on multivariate methods in psychology concerns techniques for analyzing empirical data. Yet the simultaneous study of a large number of variables presents as many thorny problems for the theorist as for the empiricist. Theoretical models that could pass scrutiny when all but a few variables were ignored frequently fail in attempting to explain the interactions among a large set of variables. Too often the solution has been to present a set of simple models describing the observed relations between pairs of variables. Although such descriptive models are sometimes accepted as psychological theories, they usually provide little information about underlying psychological processes.
