ABSTRACT
Social science research is faced with a widening discrepancy between the formal assumptions of statistical techniques and mathematical models and the substantive goals of research. The practicing researcher appears to be confronted with the choice between preselecting his or her topics and data to fit the scientifically legitimated moulds of formal tools on the one hand and of sticking to his or her problems using data and models in a much more exploratory and tentative manner on the other hand. Stanley Lieberson's work has played a crucial role in removing the veils of ignorance and innocence which conventional ideals and practices of empirical research enjoyed for a long time. Lieberson's arguments support a methodology of social research in which individual decisions and attributes should play a much smaller role as dependent variables, and inter-individual differences a much smaller role as independent variables than is currently the case.
