Realistic Models? Critical Realism and Statistical Models in the Social Sciences.

Abstract

Critical realist scholars' skeptical attitude toward the use of statistical models in social scientific research is examined. After offering a brief review of the philosophy of social science provided by critical realism, the thought of Roy Bhaskar & Tony Lawson is scrutinized, indicating that their critical realist critique of social scientific statistical models as "deductivist" is unjustifiable. On the contrary, it is contended that critical realism provides an approach that answers several philosophical questions concerning the specification & interpretation of statistical models & supplies a robust approach to critiquing mainstream economics. It is contended that Structural Equation Modeling methods possess considerable explanatory power; indeed, the ability of Structural Equation Modeling approaches for determining "goodness of fit," interpreting model coefficients, & identifying "omitted & included variable bias" are stressed. The potential implications of adopting this alternative critical realist understanding of social scientific statistical models for improving investigations of causality are also pondered. 75 References. J. W. Parker