The increasing number of studies, particularly in the area of comparative and historical research, are using the method of agreement and method of difference proposed by Mill (1872) to infer causility based on a small number of cases. This article examines the logic of the assumptions implicit in such studies. For example, the research must assume: (1) a deterministic approach rather than a probabilistic one, (2) no errors in measurement, (3) the existence of only one cause, and (4) the absence of interaction effects. These assumptions are normally inappropriate, since they contradict a realistic appraisal of most social processes, but are mandatory if we follow Mill's causal analyses based on small N's. Research should not attempt employment of such methods in small N cases without a more rigorous justification of heroic assumptions and a guard against possible distortions. (A)
Abstract