Toward a Non-experimental Method for Causal Analyses of Social Phenomena

A method is described for detecting causal relationships among social phenomena in the natural setting. Based primarily upon obtaining an association between two factors, one measured earlier in time than the other, and controlling the influences of outside factors, it considers causality to be a matter of degree and allows for symmetrical causal relationships. A study of organizational behavior is described to demonstrate the applicability of the method. After contrasting the method to path analysis and panel-study analysis, it is suggested that behavioral scientists think in terms of three levels of research on complex cycles of social behavior: one point-in-time correlations to determine general associations among factors, methods like the one proposed here to determine general causal relationships, and, when possible, experiments to determine more precisely the causal relationships among factors of particular interest .