single institutional form is best at maintaining resources across a wide range of environmental and social conditions. Researchers have begun to propose hypotheses about conditions under which particular institutional forms are likely to be successful. Similarly, research has shown that simple bivariate relationships of the sustainability of resource management with the size, heterogeneity, and poverty of the user group may be positive, negative, or curvilinear, depending on contextual factors (Agrawal, Chapter 2). Researchers have responded by developing and testing hypotheses that take these contingencies into account.
Studies with large numbers of cases (large-n studies) are particularly useful for generating such hypotheses because they allow regularities to be observed in subsamples that differ in factors that change the effect of other variables—the factors that make conclusions contingent. For example, Tang (1992, cited by Bardhan and Dayton-Johnson, Chapter 3) reports that heterogeneity among resource appropriators is associated with poorer performance in irrigation systems that are managed by government agencies, but not in community-managed systems (see Figure 13-1). Apparently, some community-managed systems are able to develop rules of allocation and cost sharing that meet the challenges of heterogeneity, while agency-managed systems are not. Similarly, Varughese and Ostrom (2001) show that various forms of heterogeneity within forest user groups depend for their effects on collective action on the specific form of organization established by the group. Identifying this difference in the effects of heterogeneity requires cases that differ in their degree of heterogeneity among both community-managed and government-managed systems.2
A third element of development is a shift from correlational to causal analysis. Researchers hypothesize and search for causal paths or mechanisms that can