nity pursues this path self-consciously. This implies some changes in styles of research and an increased coordination of the research community around theory development and testing.
Case studies have contributed greatly to knowledge by documenting the limitations of the Tragedy of the Commons model, identifying key variables, and generating hypotheses. Case studies will continue to break new ground, particularly in investigations of “new commons” and interinstitutional linkages, as well as in participatory research (to be discussed). In such frontier research areas, in-depth observation is needed to uncover phenomena or variables that might be missed if researchers looked only at variables known to be important in well-studied areas of the field. However, it is now possible to use case study methods within theoretically driven research programs, for example, using methods of focused and structured case comparison (Bennett and George, in press) and theory-driven evaluation (Birckmayer and Weiss, 2000; Chen, 1990; Chen and Rossi, 1992). It is also possible to mine existing case studies by developing structured coding forms to extract common information about theoretically relevant variables and thus to test propositions (see Ragin, 1987, 2000; Ragin and Becker, 1992; Tang, 1992; Schlager, 1994). The results of such systematic assessments of previous case studies can also point toward critical questions for new case studies. All these strategies should become much more prominent in future uses of case study methods. However, it is important to remember that there are inherent limitations to case approaches, such as those created by the need to compare any case history with a counterfactual scenario based on what might have happened instead (see Tetlock and Belkin, 1996; Roese, 1997).
Theory has developed to a point that it provides contingent generalizations that can be illuminated by multicase, multivariate research methods. This development implies an increased role in the next decade for relatively large-n, multivariate research as well as for the new case study-based methods already described. Efforts to develop large-n multivariate databases (Agrawal and Yadama, 1997; Ostrom, 1998) provide essential infrastructure for the quantitative multivariate style of multicase research. Syntheses of existing research, such as those offered in several chapters of this volume, are also essential because they generate hypotheses that involve variables that are missing from the large databases but that can be investigated by focused case comparison methods.