Generalizing Generalizability in Information Systems Research

The concept of generalizability is not homogeneous and monolithic, but can be analyzed into four types: the generalizability of a theory to different settings, the generalizability of a theory within a setting, the generalizability of a measurement or observation, and the generalizability of a variable, construct, or other concept. In this study, we affirm the legitimacy of the statistical, sampling-based conception of generalizability (which falls under the generalizability of a measurement or observation), where we identify it as but a subset of the overall concept of generalizability. We explain that to regard it as the general form of generalizability would, in itself, be an overgeneralization. We offer constructive guidelines, based on established arguments in the philosophy and methodology of science, for how to achieve the generalizability of a theory to different settings. We provide a concrete example of the guidelines by applying them to a classic study (Markus, 1983) to show how a theory may be properly generalized to different settings.