clusions on a very small number of cases, suffer from the problem of not specifying carefully the causal model they are testing. In the absence of such specification, qualitative studies of the commons are potentially subject to significant problems of method. Two of the most important of these problems are those stemming from “omitted variable bias” and the problem of endogeneity (King et al., 1994:168-182, 185-195). These biases resulting from deficiencies of method have the potential to produce an emphasis on causal factors that may not be relevant, ignoring of other factors that may be relevant, and the generation of spurious correlations.

An incorrect emphasis on some causal variables also may result from the underlying problem of multiple causation, where different causal factors or combinations of causal factors may have similar impacts on outcomes (Ragin, 1987). Thus unpredictable benefit flows and unfair allocation both may have adverse effects on durability of institutions. But in a particular case, it is possible that although benefit flows are unpredictable, they have a much smaller effect on outcomes compared to “unfair allocation of benefits,” and that the researcher has ignored the nature of allocation. In such a situation, the conclusions from the study would be flawed in that they would under- or overemphasize variables inappropriately. This issue is especially acute for commons researchers because conclusions from much case study analysis are couched in terms of directional effects of independent variables: positive or negative. “Unpredictable benefit flow,” it can be argued, undermines the sustainability of commons institutions. But in a case study it may be difficult to discover how particular independent variables are related to each other, or the strength of their relationship to observed outcomes. In an important sense, single-case analyses, especially when they cover a single time period, limit conclusions about cause-effect relationships to bivariate statements when actual relationships are likely to be more contingent, or continuous.

The large number of variables potentially affecting the sustainability of institutions that govern common resources, thus, has important theoretical implications for future research. The most important implication is perhaps for research design. Because the requirement of a random or representative selection of cases is typically very hard to satisfy where common-pool resources are concerned (even when the universe of cases is narrowed geographically), purposive sampling easily becomes the theoretically defensible strategy for selecting cases whether the objective is statistical analysis or structured comparative case analysis. In purposive sampling, the selected cases will be chosen for the variation they represent on theoretically significant variables. This strategy can be defended both because it is easier to implement than an effort to select a representative sample, and because it requires explicit consideration of theoretically relevant variables (Bennett and George, 2001; Stern and Druckman, 2000).49

There is no general theory of purposive sampling apart from the common-sense consideration that selected cases should represent variation on theoretically



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement