*The author is grateful for the advice of David Dessler, Paul Diesing, Gary Goertz, Jill Homer-Dixon, Ned Lebow, and Barbara Torrey.

1. Jack Goldstone, Revolution and Rebellion in the Early Modern World (Berkeley, CA: University of California Press, 1991); Thomas Homer-Dixon, "On the Threshold: Environmental Changes As Causes of Acute Conflict," International Security, Vol. 16, No. 2 (Fall 1991), pp. 76-116; Thomas Homer-Dixon, "Environmental Scarcities and Violent Conflict: Evidence from Cases," International Security, Vol. 19, No. 1 (Summer 1994), pp. 5-40; Arthur Westing, ed., Global Resources and International Conflict: Environmental Factors in Strategic Policy and Action (New York: Oxford University Press, 1986).

2. For many years, political scientists generally thought that testing hinged on falsification: if data clearly contradicted a hypothesis, the theory from which the hypothesis had been deduced was "falsified" and therefore rejected. Knowledge cumulation progressed not by proof but by disproof. Although based on Karl Popper's interpretation of natural science, most methodological experts now acknowledge that falsificationism seriously misinterprets how natural science actually works. Years ago, for example, Quine showed decisively that theories are tested as a whole and that the discovery of evidence that contradicts a particular hypothesis deduced from a theory hardly ever results in the wholesale rejection of the theory. See Willard van Orman Quine, "Two Dogmas of Empiricism," in From a Logical Point of View: Nine Logico-Philosophical Essays, 2nd edition (Cambridge, Mass.: Harvard University Press, 1953). For a general critique of falsificationism and a defense of an alternative understanding of hypothesis testing, see Paul Diesing, How Does Social Science Work? Reflections on Practice (Pittsburgh, PA: University of Pittsburgh Press, 1991), especially pp. 248-254.

3. Barbara Geddes, "How the Cases You Choose Affect the Answers You Get: Selection Bias in Comparative Politics," in James Stimson, ed., Political Analysis: An Annual Publication of the Methodology Section of the American Political Science Association , Vol. 2 (Ann Arbor, MI: University of Michigan Press, 1990), pp. 131-150; Gary King, Robert Keohane, and Sidney Verba, Designing Social Inquiry: Scientific Inference in Qualitative Research (Princeton, NJ: Princeton University Press, 1994), pp. 128-135.

4. Douglas Dion, "Evidence and Inference in the Comparative Case Study," unpublished draft paper, Department of Political Science, University of Michigan, 1994; see also Benjamin Most and Harvey Starr, Inquiry and International Politics (Columbia, SC: University of South Carolina Press, 1989), pp. 52-54.

5. Dessler similarly distinguishes between a focus on outcomes and a focus on causal factors: "The analyst interested in some phenomenon might treat it as an outcome or feature of some process or structure and search for conditions associated with its appearance. Alternatively, the researcher might choose a factor known or thought to play a role in causing the phenomenon and analyze the tendencies of this factor in isolation. Both categories of analysis link factors to outcomes, but convey different information about this link. While the first category (focus on outcome) tells us what configuration of conditions lead to some specified observed outcome in the world, the second one (focus on factor ) tells us what outcomes tend to be brought about by the workings of a specified factor, whether or not these outcomes are actually produced." See David Dessler, "The Architecture of Causal Analysis," unpublished manuscript prepared for the Seminar for Philosophy and Methodology of the Social Sciences, Center for International Affairs, Harvard University, April 1992, p. 8. Dessler derives his distinction from John Stuart Mill's System of Logic: Ratiocinative and Inductive (New York: Harper, 1859).

6. Although widely used by social scientists, the concept of causation is far more imprecise than is usually acknowledged. For insightful analysis, see Paul Humphreys, The Chances of Explanation: Causal Explanation in the Social, Medical, and Physical Sciences (Princeton, NJ: Princeton University Press, 1989); Wesley Salmon, Scientific Explanation and the Causal Structure of the World (Princeton, NJ: Princeton University Press, 1984); Peter Van Inwagen, Time and Cause: Essays Presented to Richard Taylor (Dordrecht, The Netherlands: Reidel Publishing, 1980); and Tom Beauchamp and Alexander Rosenberg, Hume and the Problem of Causation (New York: Oxford University Press, 1981). In addition, there are serious philosophical debates about whether and how causal claims in the social sciences differ from those in the natural sciences. These debates are particularly pertinent to environment-conflict research, because many causal claims in this field mix natural and social variables. See Jerry Fodor, "Introduction: Two Kinds of Reductionism," in The Language of Thought (Cambridge, MA: Harvard University Press, 1975), pp. 1-26; Stephen Schiffer, "Ceteris Paribus Laws," Mind, Vol. 100, No. 397 (January 1991), pp. 1-17; and Alexander Rosenberg, "Obstacles to the Nomological Connection of Reasons and Actions," Philosophy of Social Science , Vol. 10 (1980), pp. 79-91.

7. David Dessler, "How to Sort Causes in the Study of Environmental Change and Violent Conflict," in Nina Graeger and Dan Smith, eds., Environment, Poverty, Conflict, PRIO Report 2/94 (Oslo: International Peace Research Institute, 1994), pp. 91-112.

8. See Most and Starr, Inquiry and International Politics, pp. 47-67, for a discussion of necessity and sufficiency. The strength of a cause can be measured by the probability of the cause producing a given effect; if the probability is 1.0, then the cause is sufficient.

9. Goertz notes that causal proximity is influenced by theoretical and pragmatic concerns, because it is usually possible to specify the variables and links in the causal process with greater and greater detail and thereby reduce proximity, especially by dropping down to lower levels of analysis. Gary Goertz, Contexts of International Politics (Cambridge, UK: Cambridge University Press, 1994), pp. 181-182.

10. See Homer-Dixon, "Environmental Scarcities and Violent Conflict," pp. 35-36. When environmental degradation becomes irreversible, it takes on the character of a "barrier" cause - that is, a cause that constrains opportunities and precludes options. See Goertz, chapter 6 "Barrier Models of Context," in Contexts of International Relations, pp. 90-113.

11. In contrast, "environmental determinists" tend to assume that few factors operate and that the environmental ones are powerful.

12. Such a claim is like the assertion that the 3 in the product term 2 x 3 = 6 contributes more to the 6 than the 2 does. The error of treating an interactive relationship among variables as an additive one is common in debates about the relative contribution of nature and nurture to such human characteristics as height and intelligence; commentators will often claim that some proportion of measured height or intelligence - say, 60 percent - is a consequence of nature, while the remainder, additively, is a result of nurture. As Sober points out, such a claim is meaningless when applied at the "local" level - that is, at the level of the individual person or event. At the level of the population of persons or events, however, the claim might be a meaningful interpretation of the results of an analysis of variance, if it means that a certain proportion of the variance in the population can be explained by the factor in question. See Elliott Sober, "Apportioning Causal Responsibility," The Journal of Philosophy, Vol. 85, No. 6 (June 1988), pp. 303-318. At the local level, the only time it is appropriate to differentially weight the causal contribution of variables in an interactive relationship is when they have different rates of change over time. Holdren provides a mathematical method for estimating the relative causal importance of variables in such situations. See John Holdren, "Population and the Energy Problem," Population and Environment, Vol. 12, No. 3 (Spring 1991), pp. 231-255, especially pp. 243-244.

13. Such opportunistic exploitation of drought in Africa is discussed by Peter Wallensteen in "Food Crops as a Factor in Strategic Policy and Action," in Arthur Westing, ed., Global Resources and International Conflict: Environmental Factors in Strategic Policy and Action (New York: Oxford University Press, 1986), pp. 154-155.

14. Homer-Dixon, "Environmental Scarcities and Violent Conflict."

15. A statement of the form, "If X, then Y," implies that the antecedent X is a sufficient cause or condition for the consequent Y. When X is hypothesized to be a necessary cause, then the statement should be of the form, "Only if X, then Y." When X is proposed as a necessary and sufficient cause, then the statement should be, "If and only if X, then Y ."

16. Alexander George and Timothy McKeown, "Case Studies and Theories of Organizational Decision Making," in Robert Coulam and Richard Smith, eds., Advances in Information Processing in Organizations, Vol. 2 (London: JAI Press, 1985), pp. 21-58, especially p. 24.

17. Of course, as often noted, correlation does not prove causation; so a correlational analysis by itself will not adequately answer the three key questions posed in the introduction of this article.

18. A similar approach is crucial case analysis, in which hypotheses deduced from a theory are tested against a case that would appear to be better explained, prima facie, by an alternative, competing theory. See Harry Eckstein, "Case Study and Theory in Political Science," in Fred Greenstein and Nelson Polsby, eds., Handbook of Political Science, Vol. 1, Political Science: Scope and Theory (Reading, MA: Addison-Wesley, 1975), pp. 79-133.

19. For a thorough discussion of counterfactual analysis, see James Fearon, " Counterfactuals and Hypothesis Testing in Political Science," World Politics, Vol. 43, No. 2 (January 1991), pp. 169-195.

20. Controlled-case comparison and process tracing are both discussed in George and McKeown, "Case Studies and Theories of Organizational Decision Making," pp. 24-43.

21. To the extent that these causal linkages are specified by the researcher's hypotheses, process tracing increases the number of empirical observations that can be used to test the hypotheses. This is one way of dealing with the problem of an inadequate number of observations for the number of causal variables hypothesized - the "small-N problem" - that many analysts believe bedevils comparative case-study methodology. See King, Keohane, and Verba, Designing Social Inquiry, pp. 226-227.

22. George and McKeown, "Case Studies and Theories of Organizational Decision Making," pp. 31-32. Process tracing provides a particular type of explanation of the independent variable, which Abraham Kaplan calls the "pattern" model of explanation. Kaplan writes: "According to the pattern model . . . something is explained when it is so related to a set of other elements that together they constitute a unified system. We understand something by identifying it as a specific part in an organized whole." Kaplan notes that the pattern model of explanation is distinct from the "deductive" model: "Very roughly, we know the reason for something either when we can fit it into a known pattern, or else when we can deduce it from other known truths." See Abraham Kaplan, The Conduct of Inquiry (San Francisco: Chandler, 1964), pp. 332-335. Kaplan's deductive model corresponds to Hempel's "deductive-nomological" or "covering-law" model of explanation, whereby a phenomenon is said to be explained if its occurrence can be shown to be logically expected, given certain general laws. However, Hempel similarly distinguishes between covering-law explanations and what he calls "genetic" explanations, which, he argues, are generally a better form of explanation for social events. A genetic explanation "presents the phenomenon under study as the final stage of a developmental sequence, and accordingly accounts for the phenomenon by describing the successive stages of that sequence." See Carl Hempel, Aspects of Scientific Explanation (New York: Free Press, 1965), p. 447. Thagard makes an analogous distinction among deductive, schematic, and causal modes of explanation. See Paul Thagard, Conceptual Revolutions (Princeton, NJ: Princeton University Press, 1992), pp. 118-126.

23. Homer-Dixon, "Environmental Scarcities and Violent Conflict."

24. The conjunction in this statement could also be "or." Thus, an exhaustive statement of the conditions would be of the form "X causes Y when conditions A and/or B and/or C . . . and/or N are true." The relationship between X and these conditions is interactive. Ragin discusses the methodological implications of such "multiple conjunctural" causation in which multiple causes interact in different combinations to produce effects of interest to researchers. See Charles Ragin, The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies (Berkeley, CA: University of California Press, 1987), pp. 23-30.

25. Donald Ludwig, Ray Hilborn, and Carl Walters, "Uncertainty, Resource Exploitation, and Conservation: Lessons from History," Science, Vol. 260, No. 5104 (2 April 1993), pp. 17 and 36; Wallace Broecker, "Unpleasant Surprises in the Greenhouse?" Nature, Vol. 328, No. 6126 (9 July 1987), pp. 123-126; Robert Chen and Myron Fiering, Climate Change in the Context of Multiple Environmental Threats, Research Report RR-89-1 World Hunger Program, Brown University (Providence, RI: March 1989); Vaclav Smil, Global Ecology: Environmental Change and Social Flexibility (London: Routledge, 1993), especially chapter 5, "Cascading Complications"; and Bo Wiman, "Implications of Environmental Complexity for Science and Policy," Global Environmental Change, June 1991, pp. 235-247.

26. The problems of control in case-study research are highlighted in George and McKeown, "Case Studies and Theories of Organizational Decision Making," p. 27. See also, Fearon, "Counterfactuals and Hypothesis Testing in Political Science," p. 174, n. 11.

27. Marc Levy advocates this strategy in "Global Environmental Degradation: National Security and U.S. Foreign Policy," Working Paper No. 9, Project on the Changing Security Environment and American National Interests, John M. Olin Institute for Strategic Studies, Harvard University (November 1994), p. 25.

	TOP
	Part 1
	EPS Homepage
	Publications on the Web