DIMENSIONAL SAMPLING: AN APPROACH FOR STUDYING A SMALL NUMBER OF CASES.

This article presents a logic for small-numbers research that takes advantage of the number of cases involved, and that has definite advantages for the development of theory not found in either the single-case study or the large-numbers approach. Briefly, the approach is a three-step one: first, explicitly delineate the universe to which you eventually wish to generalize; then, spell out what appear to be the most important dimensions along which the members of this universe vary and develop a typology that includes the various combinations of values on these dimensions; and, use this typology as a sampling frame for selecting a small number of cases from the universe, typically drawing one case from each cell of the typology. Most sociological research involved either the study of single cases or the statistical analysis of large numbers of cases. The middle ground, the study of a small number of cases, is occupied only rarely. Furthermore, most of what I shall call small-number studies involve either a direct extension of the single-case-study logic or an attempt at following the logic of statistical studies with an inadequate number of cases.