Origins of the EU 49 influence behavior have been well elaborated by the overlapping"constructivist" and "sociological institutionalist"schools in international relations and by compar- ative political economists like Peter Hall.4 These scholars have had difficulty moving from ontological propositions to strong empirical demonstrations,however, due to two related problems.The first is the "Janus-faced"nature of ideas. Sometimes actors'beliefs guide their actions;sometimes apparent beliefs only rationalize strategies chosen for other reasons."Distinguishing between the two situations can be difficult.Second,even assuming ideas have causal effects,they do so as interpretations or "filters"of the objective environment.Wherever ideas have causal effects-in Max Weber's famous phrase,as "switchmen"among various material possibilities-so does the objective context they interpret.The challenge for causal ideational argument is to isolate the ideational filter from its context,to separate the subjective components of actors'perceived"interests"from their direct responses to the environment. The postmodern extreme of the ideational literature meets this challenge by denying it.The very notion of an accessible objective context is rejected,making interpretation of ideas the entire exercise,without attempts to assess how ideas relate to objective pressures.'More mainstream ideational arguments,however,use fairly standard social-science methods to suggest the need for similar kinds of interpre- tation.Most common in the constructivist.sociological institutionalist.and com- parative literatures is a reliance on "process tracing."The observer traces the objective pressures impinging on certain decisions and concludes that they did not fully determine a choice.Thus we need to interpret the beliefs that did.Cross-case comparisons are also often used to set up interpretation.In contexts ranging from the early Industrial Revolution to interwar military strategizing,scholars suggest that actors in objectively similar situations adopted different strategies due to different ideas.10 Other studies show the need for interpretation in the proliferation of similar policies across structurally different cases. Especially when combined,these methods produce strong claims that ideas can affect politics to some degree.Their main weakness,however,is an apparent inability to show how much ideas cause certain outcomes.Scholars using these methods offer qualitative assessments of indeterminacy left by objective pressures and interpretations of how certain ideas resolved it.Yet skeptics can always question the former assessment,suggesting(for example)that it underestimates the objective economic pressures toward a choice-meaning that ideas caused it less than has been claimed.A process-tracing focus on a single course of action also offers little 4.See,in particular,Wendt 1989;Onuf 1989;Finnemore 1996;and Hall 1989. 5.See Laitin 1986,11-20:Tetlock 1991,47:Shepsle 1985:and Krasner 1993. 6.Weber1958.280. 7.For a recent example in EU studies,see Diez 1999. 8.George 1979. 9.See Hall 1989 and 1993;Goldstein and Keohane 1993,26:and Adler and Haas 1992. 10.See Biernacki 1995:Kier 1997:Berman 1998;and Hattam 1992. 11.Finnemore 1996.influence behavior have been well elaborated by the overlapping “constructivist” and “sociological institutionalist” schools in international relations and by compar￾ative political economists like Peter Hall.4 These scholars have had difficulty moving from ontological propositions to strong empirical demonstrations, however, due to two related problems. The first is the “Janus-faced” nature of ideas. Sometimes actors’ beliefs guide their actions; sometimes apparent beliefs only rationalize strategies chosen for other reasons.5 Distinguishing between the two situations can be difficult. Second, even assuming ideas have causal effects, they do so as interpretations or “filters” of the objective environment. Wherever ideas have causal effects—in Max Weber’s famous phrase, as “switchmen” among various material possibilities—so does the objective context they interpret.6 The challenge for causal ideational argument is to isolate the ideational filter from its context, to separate the subjective components of actors’ perceived “interests” from their direct responses to the environment. The postmodern extreme of the ideational literature meets this challenge by denying it. The very notion of an accessible objective context is rejected, making interpretation of ideas the entire exercise, without attempts to assess how ideas relate to objective pressures.7 More mainstream ideational arguments, however, use fairly standard social-science methods to suggest the need for similar kinds of interpre￾tation. Most common in the constructivist, sociological institutionalist, and com￾parative literatures is a reliance on “process tracing.”8 The observer traces the objective pressures impinging on certain decisions and concludes that they did not fully determine a choice. Thus we need to interpret the beliefs that did.9 Cross-case comparisons are also often used to set up interpretation. In contexts ranging from the early Industrial Revolution to interwar military strategizing, scholars suggest that actors in objectively similar situations adopted different strategies due to different ideas.10 Other studies show the need for interpretation in the proliferation of similar policies across structurally different cases.11 Especially when combined, these methods produce strong claims that ideas can affect politics to some degree. Their main weakness, however, is an apparent inability to show how much ideas cause certain outcomes. Scholars using these methods offer qualitative assessments of indeterminacy left by objective pressures and interpretations of how certain ideas resolved it. Yet skeptics can always question the former assessment, suggesting (for example) that it underestimates the objective economic pressures toward a choice—meaning that ideas caused it less than has been claimed. A process-tracing focus on a single course of action also offers little 4. See, in particular, Wendt 1989; Onuf 1989; Finnemore 1996; and Hall 1989. 5. See Laitin 1986, 11–20; Tetlock 1991, 47; Shepsle 1985; and Krasner 1993. 6. Weber 1958, 280. 7. For a recent example in EU studies, see Diez 1999. 8. George 1979. 9. See Hall 1989 and 1993; Goldstein and Keohane 1993, 26; and Adler and Haas 1992. 10. See Biernacki 1995; Kier 1997; Berman 1998; and Hattam 1992. 11. Finnemore 1996. Origins of the EU 49