Theory testing within small-N research designs is problematic. Developments in the philosophy of social science have opened up new methodological possibilities through, among other things, a novel notion of contingent causality that allows for contextualized hypothesis generation, hypothesis testing and refinement, and generalization. This article contributes to the literature by providing an example of critical realist (one such new development in the philosophy of social science) theory development for a small-N comparative case study that includes hypothesis testing. The article begins with the key ontological assumptions of critical realism and its relation to theory and explanation. Then, the article presents an illustrative example of an e-government comparative case study, focusing on the concept of trust, which follows these ontological assumptions. The focus of the example is on the nature and process of theory and hypothesis development, rather than the actual testing that occurred. Essential to developing testable hypotheses is the generation of tightly linked middle-range and case-specific theories that provide propositions that can be tested and refined. The link provides a pathway to feed back the concrete empirical data to the higher level (more abstract) and generalizable middle-range theories.
TopHow Does Critical Realism Influence Research?
Critical realism is a relatively new philosophy of the natural and social sciences developed in the late 70s and early eighties (Bhaskar, 1978, 1998b). Since then it has provided the basis for a range of social science research (Carter & New, 2004; Danermark, Ekstrom, Jokobsen, & Karlsson, 2003; Mingers, 2000, 2004d; Pawson & Tilley, 1997). In the information systems literature, the potential benefits of critical realism have already been touted (Carlsson, 2004; Dobson, 2002; Houston, 2001; Mingers, 2004a, 2004c, 2004d; Smith, 2006). Recently, more examples of critical realism actively applied in information systems research are emerging (Bygstad, 2008; Dobson, Myles, & Jackson, 2007; Morton, 2006; Reimers & Johnson, 2008; Volkoff, Strong, & Elmes, 2007).