Acknowledgements

For helpful comments on previous drafts, we would like to thank participants at two workshops held at the University of Washington (24-25 March 2000 and 4-5 May 2001), the project members, Steve Hanson, Simon Hug, Dan Kelemen, Jeff Kopstein, Paul Kowert, Jeff Lewis, Andy Moravcsik, Johan P. Olsen, Thomas Risse, Vivien Schmidt, and Gerald Schneider.� Earlier versions were presented at the 2001 meeting of the European Community Studies Association and the 2002 annual convention of the American Political Science Association.

 

Abstract

Three central goals motivate this introductory essay and the articles that follow.� First, we seek better understanding of the EU's own institutions, especially to the extent that they play a role in, or represent, the process of integration.� Second, we seek better integration of the multiple general understandings of institutions, and primarily of rationalist and constructivist conceptions.� Third and most important, we seek to promote the integration of institutional research.� Our overarching argument is that metatheoretical debate about institutions has run its course and must now give way to theoretical, methodological, and carefully-structured empirical dialogue.� To this end, we offer specific strategies for promoting greater synthesis among competing institutional schools.

Keywords: Institutional theory; rationalism; constructivism; European Union; research design; methodology

I. Introduction

The study of the European Union (EU) --how it works, what it is, and the processes by which it has evolved-- poses considerable challenges, but also promises considerable rewards to those who would undertake it.� The EU's architecture and operation are complex, its identity as a political system problematic, and its location within traditional academic subfields ambiguous.� We view these features more as opportunities than as obstacles.� On this view, the EU represents a living laboratory within which to study a variety of economic, social, political, and institutional developments, none of which is unique to it.� EU studies, rather than standing separate from the subfields and from broader developments in the discipline, may instead stand astride them, at once drawing from and influencing them.� This prospect should attract political scientists of all stripes, whatever their theoretical and methodological tendencies, area or subject specializations, or subfield identities.

 

The research reported here focuses on "integrating institutions," which, to us, means three things.� First, we seek better understanding of the EU's own institutions, especially to the extent that they play a role in, or represent, the process of integration.� The rapidly expanding body of research into EU institutions, while addressing many important questions, lacks coherence, often of the most basic sort, and thus risks fragmenting into a whole worth less than the sum of its parts.� Second, we seek better integration of the multiple general conceptions of institutions, and primarily of rationalist and constructivist (or sociological) conceptions.� While many recognize potential complementarities between these various institutionalisms, metatheoretical debate has often hindered theoretical dialogue and fruitful empirical work.� This is as true of the generic institutional literature as it is to applications of it to the EU.� Third, we seek to promote the integration of institutional research across the subfields of political science (primarily, comparative politics and international relations), which increasingly employ similar institutionalist language but, for now, speak far too little to each other.

 

Our overarching argument is that the metatheoretical debate about institutions has run its course and must now give way to theoretical, methodological, and carefully-structured empirical dialogue.� To this end, we offer specific strategies for channeling the institutionalist research program in fruitful directions.� The papers that follow implement some of these suggestions, collectively indicating the current measure of integration and opening opportunities for additional efforts.� We hasten to emphasize that many of our suggestions are ambitious and are not taken up in the papers that follow.� Rather than tailoring our advice to the collected contributions, however, we have chosen to set forth an ambitious agenda in the hope that it abets ongoing efforts by a broad scholarly community.

 

The paper proceeds in five parts.� After this brief introduction, Part II locates the study of the EU within broader disciplinary trends, most notably the embrace of institutional analysis across the substantive subfields of political science.� Part III summarizes the rational choice and sociological (or constructivist) institutionalist variants, highlighting possible points of commonality, complementarity, and divergence.� Part IV comprises the bulk of the paper.� It draws out various models of dialogue whereby the rationalist-constructivist discussion might be made more theoretically and empirically fruitful, and proposes research design strategies that might be used to implement them.� Part V summarizes our arguments and introduces the papers that follow, identifying both individual and collective contributions to the project of integrating institutions.

II. European Union Studies and the Discipline

The relationship of EU studies to traditional political science subfields is a long and complicated one.� The neofunctionalists' early and sustained attempt at developing a theory of regional integration, while clearly drawn from the functionalist approach to international organization, was highly original and stood partly separate from the core concerns of and trends in comparative politics and international relations.� While neofunctionalism's self-proclaimed demise in the mid-1970s (Haas, 1976) gave rise to a less self-consciously theoretical, more narrowly empirical body of EU (then EC) research (Smith and Ray, 1993), EU scholarship did not stand unrelated to the discipline.� Work on the EC during the 1970s devoted itself to a variety of concerns often associated with comparative politics and public administration, including confederalism (Taylor, 1975), policy implementation (Puchala, 1975), and a spate of studies in particular issue-areas such as agricultural policy (Feld, 1979).� In IR, neofunctionalism provided the "intellectual opening" (Katzenstein, Keohane and Krasner, 1998, p. 654) for the development of the theory of complex interdependence (Keohane and Nye, 1975), and ultimately of a neoliberal rival to the dominant realist paradigm, setting the stage for the "great debate" of the 1980s between those two approaches.� That said, EU studies experienced perhaps its greatest intellectual isolation and lowest level of theoretical ambition during this period.

 

A renewed interest in general theory accompanied the relaunch of the European Community in the mid-to-late 1980s and beyond.� This found expression most notably in the rediscovery of neofunctionalism (Sandholtz and Zysman, 1989) and the reformulation of intergovernmentalism (Moravcsik, 1991, 1993), but also occurred (more subtly) in the continued application of theories and methods more commonly associated with comparative politics (see generally Sbragia, 1991).� These included work on public opinion (Eichenberg and Dalton, 1993), political parties (Attina, 1990), consociationalism (Taylor, 1991), interest representation (Streeck and Schmitter, 1991; Andersen and Eliasssen, 1991; Kohler-Koch, 1994), judicial politics (Volcansek, 1986, 1992), and comparative federalism (Scharpf, 1988; Sbragia, 1993), among many others.� Indeed, by mid-1990s Hix (1994) stood prepared to declare IR and CP separate but equal in EU studies, the former most relevant in explaining the over-time integration of the EU, and the latter its everyday politics (see also Hurrell and Menon, 1996; Hix, 1996).

 

At the same time, converging empirical and intellectual trends, especially in the area of political economy, increasingly undermined the distinction between comparative and international.� Empirically, trends now falling under the rubric of "globalization" (growing trade liberalization, the diffusion of values and norms, increasing capital mobility, and heightening regulatory competition) deprived state boundaries of at least part of their meaning, to some extent effacing the distinction between politics within states (CP) and politics among them (IR).� Several intellectual trends also converged, most notably increasing attention to domestic-international linkages (for a review see Caporaso, 1997), recognition of the importance of social processes within and among states, and renewed attention to the creation, maintenance, and effects of institutions in the political economy.� Together, these developments rendered subfield distinctions increasingly anachronistic and potentially counterproductive.� Institutionalism, especially, seemed to provide an intellectual bridge, promising (according to its advocates) a general theory applicable to comparative, international, and American politics (Milner, 1998). [1]

 

These broader trends did not fail to impact the study of the EU.� While both the politics of economic interdependence and the reality of "multi-level games" had long formed central pillars of EU studies, the "institutional turn" proved more novel, and EU scholars embraced it with relish.� We will not narrate the institutional turn in EU studies here, as this terrain has been amply covered elsewhere (Jupille and Caporaso, 1999; Aspinwall and Schneider, 2000; Dowding, 2000).� Instead, we will content ourselves with two related points.� First, "institutionalism" itself proved a compound and, some say, contradictory category.� Indeed, astute analysts observed relatively quickly that there were, in fact, multiple institutionalisms (Hall and Taylor, 1996).� Among these stood two seemingly opposed variants (rational and sociological) and one hybrid that could be shaded toward one or the other pole (historical).� Second, EU institutionalists took on both the virtues and the vices of institutionalism.� While on the one hand the institutional turn promised more general relevance and intelligibility for EU studies (in addition to the supposed inherent benefits of institutional analysis), on the other, EU institutionalists tended to appropriate somewhat uncritically the "institutionalist" label without adequate consideration of what was being taken on.� Given the increasingly heated debate occurring more generally between rationalist and constructivist institutionalists, it is not surprising that EU scholars began reproducing the debate amongst themselves.

III. Rationalism, Constructivism, and the EU

Study of the EU, then, both reflects and informs the various strands of institutional analysis of politics.� We seek to aid in the development of a more coherent, integrated, institutionalism, which we believe will simultaneously improve our understanding of the EU's integrating institutions and promote dialogue across (integration of) a number of other cleavages that characterize the current literature.� Integration of rationalist and sociological (constructivist) institutionalisms especially interests us.� While in the next part of this paper we offer some specific bridging strategies, here we take a step back to consider the nature of the rationalist-constructivist "divide."� We discuss each approach in general as well as the debate between rationalists and constructivists as it has played itself out generally and among students of the EU.� The table thus set, we will then proceed to identify several specific models of dialogue that might advance integration.

A. What Is Rationalism?

Rational choice is not a substantive theory, if by that one means a system of conditional statements about the relationships among specific variables.� Instead, rational choice is best seen as an approach to social inquiry that encompasses specific theories adhering minimally to one simple postulate: "When faced with several courses of action, people usually do what they believe is likely to have the best overall outcome" (Elster, 1989, p. 22).� Rather than developing the approach at length, we will simply work through some of its key assumptions, highlighting variation within rational choice scholarship and points of convergence and divergence with constructivism.� In so doing, we take two of the four allegedly canonical assumptions identified by Abell (1992), individualism and optimality, as our points of departure. [2]

 

Rational choice theory's individualism assumption (methodological individualism) treats individuals as the basic (elemental) units of social analysis.� Both individual and collective actions and outcomes are explicable in terms of unit-level (individual) properties.� The most important of these properties are, on the one hand, transitive, fixed and given (exogenous) preferences over possible states of the world and, on the other, information and beliefs about the causal connections between actions and outcomes (states of the world).� Rational choice typically offers no theory of preferences, but deploys exogenous preferences in explaining individual and social choice.� The optimality assumption establishes a consequentialist logic of action and, more strongly, suggests that actors undertake means-ends calculations in choosing their best course of action.� Whatever actors want (and this is canonically to maximize utility), they choose what they believe to be the best means available to attain it.� As a corollary, in situations of interdependent choice (joint determination of outcomes), actors will of necessity behave strategically (that is, taking into account the expected actions of others).� Sketched in this way, rational choice is straightforward.� Individuals want things, and they act in such a way as best to obtain what they want (to the best that they can discern this, and subject to the constraints they face).

Were these assumptions widely and/or strongly respected by rational choice theorists, the room for non-competitive dialogue with constructivists might be slim. �However, the practice of rational choice scholarship adheres quite variably to these canonical assumptions, and in our view space thus exists for dialogue with constructivism.

 

The alleged methodological individualism of rational choice theory, for example, is not as opposed to constructivist postulates � which place greater weight on social wholes -- as the above might suggest.� First, much rational choice work deals in practice with collective rather than individual agents.� Collectives are very often anthropomorphized -- among other things, they are assigned preferences (utility functions) and allowed to act -- and treated as if they were individuals, despite some well-known problems of translating multiple individual preferences and choices into collective ones.� This move brings mixed consequences for the possibility of rationalist-constructivist dialogue.� On the one hand, that this is done by fiat forecloses some constructivist research opportunities, not least consideration of social processes in the formation of collective preferences and identities and the realization of collective choices.� In that sense, the practice of treating collectives as if they were individuals could be seen to do violence to the constructivist research program.� On the other hand, this practical recognition that we cannot always reduce collective processes and outcomes to individual ones may offer dialogic or even bridge-building opportunities, a point of commonality that can be exploited.

 

Second, even where individual agents do, in practice, hold explanatory pride of place in rational choice theory, they act within exogenous constraints (call them structures) that also exert causal effects, potentially very important ones.� These constraints can be material (e.g., resource limitations), informational, institutional, or even social.� In an important sense, then, agents and structures combine in rational choice theory to produce individual behavior and collective outcomes (Van Hees, 1997).� Here, and especially to the extent that social structures influence individual action, the possibility of common ground among rationalism and constructivism seems high.

 

The optimality assumption, for its part, arguably gives rational choice its power and most clearly differentiates it from non-rational choice (and especially sociologically-inclined) approaches (Coleman and Fararo, 1992).� The implied logic of consequentiality, in March and Olsen's (1989) terms, is both an irreplaceable part of the rationalist enterprise and at least partly contradicted by its constructivist/ sociological corollary, the logic of appropriateness.� Similarly, the centrality of strategic choice in rational choice theory often strongly differentiates it from constructivist/sociological alternatives.

 

That said, rationalism and constructivism, in the abstract, do not necessarily come into conflict when it comes to what actors actually want (that is, in rational choice, the preferences with respect to which they optimally choose).� Contrary to much common usage, utility is a highly open-ended concept, and an actor's utility need be neither self-regarding nor materialist.� An actor's utility function may even contain altruistic or idealistic arguments; to the extent that it does, the lines between rationalism and constructivism are blurred (see also Finnemore and Sikkink, 1998).� This may be bad for theoretical progress (in the sense that it hinders decisive empirical tests of distinct propositions) or it may be good (insofar as it opens space for dialogue).� We address these possibilities below.

 

B. What Is Constructivism

Like rational choice, constructivism is not a substantive theory, but an approach to social inquiry (Finnemore and Sikkink, 2001, p. 393).� It is based on two assumptions: (1) the environment in which agents take action is social as well as material; and (2) this setting can provide agents with understandings of their interests ("constitutes" them).� Put differently, constructivism questions the materialism and methodological individualism upon which much contemporary scholarship, including much rational choice work, has been built (Checkel, 1998, pp. 325-27; see also Adler, 2002: passim).

 

The first assumption reflects a view that material structures, beyond certain biological necessities, are given meaning only by the social context through which they are interpreted.� The second addresses the basic nature of agents, and in particular their relation to broader institutional environments.� Constructivists emphasize a process of interaction between agents and structures; the ontology is one of mutual constitution, where neither unit of analysis -- agents or structures -- is reduced to the other and made "ontologically primitive."� This opens up what for most theorists is the black box of interest and identity formation, where agent interests emerge from and are endogenous to interaction with institutional structures.

 

Constructivists thus question reductionist versions of methodological individualism, which play such a prominent role in rational choice approaches to comparative politics, IR, and the EU. [3] � This agent-centered view asserts that all social phenomena are explicable in ways that only involve individual agents and their goals and actions; the starting point of the analysis is actors with given properties.� In addition, the theories of action offered by constructivists differ from those stressed by rational-choice analysts.� While the latter stress strategic choice and consequentialist behavior driven by means-ends calculations, the former often emphasize arguing/deliberation and appropriate behavior driven by (complex) learning and dynamics of socialization (Risse, 2000). [4] � From this perspective, institutions � conceptualized, say, as social norms � have deeper effects on core properties of agents (interests and identities).� To use that favorite constructivist buzzword, institutions constitute.

 

Nearly all constructivists would accept such a description of their enterprise.� However, beyond this point, differences begin to emerge � over how far to take the "linguistic turn" (language as persuasive/argumentative appeals or as a deep structure of social discourse enabling action), over theory (substantive/operational versus normative/critical approaches) and research methods (those inspired by positivism versus those drawing upon more interpretative epistemologies). [5]

 

Here and in subsequent essays, we highlight what has been called "modernist," "thin" or "conventional" constructivism (Ruggie, 1998).� These terms refer to a group of scholars who take a middle-ground position on a number of the issues noted above.� That is, they do not reject science or causal explanation; their quarrel with mainstream theories has more to do with theoretical specification than ontology or epistemology. This moderate constructivism is well placed to bridge and integrate institutionalisms � the central theme of this special issue.� It is this substantive concern and not epistemological dogma that thus leads us to background the contributions of more interpretative and discourse-theoretic versions of constructivism. [6]

 

In practical terms, moderate constructivists often explore the role of social facts � norms or culture � in constructing the interests and identities of states/agents.� True to their ontological underpinnings, such "construction" comes about not only or primarily through strategic choice, but also through dynamic processes of persuasion or social learning.� In making and documenting such claims, these scholars will typically consider alternative explanations and/or employ counterfactual analysis (Risse, Ropp and Sikkink, 1999; Johnston, 2001, for example).

C. The Divide � Real But Surmountable [7]

The divide separating rationalist and constructivist institutionalisms is real.� Especially if one stays at the (rarefied) levels of epistemology and ontology, the gap is enormous. �After all, the starting point for (many) rational institutionalists is self-interested agents engaged in strategic choice; for (many) constructivists, it is other-regarding agents engaged in deliberative dynamics.� At this level, the two institutionalisms might just as well occupy separate universes, given the near-zero possibility of serious dialogue.

However, this is decisively not the level on which this special issue operates.� Rather, our goal is to help develop middle-range social scientific approaches, and not to engage in arguments at the level of abstract meta-theory.� We well appreciate that there are big and (possibly) unresolvable issues dividing some proponents of the two institutionalisms.� Yet we are also well aware that debate over such questions and over meta-theory has all too often hindered theoretical dialogue between constructivist and rationalist institutionalists (Diez, 1999a; Moravcsik, 1999b; Christiansen, Joergensen and Wiener, 2001, for example).

 

From a problem-driven, empirically-oriented perspective, such divides rapidly begin to melt away.� The starting point of the analysis moves from "either/or" to "both/and," with the latter pushing questions of scope and domains-of-application to the fore.� The fact that we observe both strategic/calculative and persuasive/other-regarding behavior in the process of European integration encourages us to capture these facts in our theories.

 

A recent exchange between Moravcsik and one of us (Checkel) reinforces the points made above and demonstrates the utility of "going empirical."� In that debate, the issue at hand is the role played by persuasion and learning in the integration process.� Starting from very different core rationalist (Moravcsik) and constructivist (Checkel) assumptions, the two go on to specify competing middle-range hypotheses for modeling such dynamics.� That is, instead of retreating to meta-theoretical name calling, they offer more fine-grained hypotheses on persuasion and learning from both a constructivist and rationalist perspective.� Less time is spent in asserting the superiority of one's preferred social theory and more is spent at a very practical and operational level, asking, for example, how or whether the competing propositions leave distinct empirical "trails" (Checkel and Moravcsik, 2001).

 

We agree with this general approach.� That is, the most productive terms for bridging this putative divide center on the specification of the conditions under which institutions matter, and precisely how the causal mechanisms operate.� True enough, rationalist and constructivist approaches differ meaningfully on many important questions, most notably the degree to which institutions structure and constitute identities and preferences.� On our reading, however, these differences often more accurately represent differences of theoretical specification and causal priority than of which variables and actors to include in one's analysis.

 

The above discussion offers hope that continued cooperation between rationalists and constructivists might bear fruit.� After all, theories can be modeled in different ways, causal sequences can be varied, assumptions about exogeneity and endogeneity can be tested (sort of), and variables can be self-consciously left out or included.� In short, having reached decreasing returns with regard to continued discussion of meta-theoretical issues, it seems high time to take seriously the twin tasks of theoretical integration and empirical testing.

 

Summarizing, we suggest, first, that theoretical dialogue is both desirable and attainable.� It is possible to build bridges.� Second, we argue that the dialogue should proceed via empirical testing of middle-range propositions.� Third, we suggest that the value added of our proposed integration must be assessed through its empirical performance � hence our effort to collect the papers that form the core of this volume.� All these essays represent structured attempts to apply one of the several models of theoretical integration offered below.

 

IV. Theory and Method: the Possibility and Practice of Dialogue

This section considers the possibility of theoretical dialogue, and offers some guidelines as to its practice.� We first address ourselves to an important philosophical objection, having to do with the potential incommensurability of different paradigms.� Having determined that this problem is surmountable, we proceed in the second subsection to advance four models of theoretical dialogue, concluding with a series of specific research designs that advance them.

A. The Commensurability Problem

One philosophical objection to the critical comparison and evaluation of theories must be met from the start.� We begin by setting out the standard view of theory testing, then introduce the philosophical objections, and then briefly suggest how we think these objections can be overcome.

 

The standard view of theory testing is that we evaluate a theory not just against the statistical (null) model but also against the best (or most commonly accepted) alternative theories.� First, we want to know if the connections supplied by our theory are observably different from a random distribution.� Second, if the relationships are significant (statistically), we want to know why.� In this sense, demonstration of statistical significance is only the beginning of the story, a prerequisite to a deeper causal probing.� Third, we may want (and there are good reasons for wanting) to compare various theories with one another. �Only in the rarest of cases is there but one plausible account to explain an outcome, nor is it likely that our starting point is zero knowledge. �It is untenable and methodologically inconsistent to argue simultaneously that we have prior knowledge (and therefore should do literature reviews as a way of identifying the knowledge frontier), and then proceed to test our theories against the statistical null.� Yet, recognition that multiple explanations are usually available raises directly the thorny issue of theoretical comparison and the more difficult task of giving operational content to such comparisons.

 

Since Kuhn's The Structure of Scientific Revolutions (1962), the commensurability issue has posed an obstacle for those who would compare alternative theories. This problem, simply and strongly stated, is that theories are different language systems with limited mutual translatability.� As with natural languages, different theories can be learned but the learning process is likely to be most successful if it occurs through direct experience rather than through translation.� Indeed, full translation may be impossible in principle. �This is so because words (or scientific terms) have different referents (observables) in different theories.� As a result, comprehension is not simply a matter of fitting two different words to the same underlying phenomenon.� Languages do not simply deploy different sounds and spellings.� They also slice and package the empirical world in different ways.� Each theory does its own work at the data level --determining what are relevant data-- and if the observations we aim to use to adjudicate among theories are themselves infected by the theory (if all observations are "theory-laden" to use Kuhn's term), this exercise is doomed to fail.

 

This view of incommensurability can be effectively countered, or at least moderated.� We offer two reasons for holding to the position that theories can be compared, except perhaps under extreme conditions.� First, while theories may contain terms with different referents, it is enough that some of these terms be the same or mutually translatable. �We need a conceptual anchor, a set of mutually intelligible concepts situated in both theories, in order to move ahead. �If we are to compare, it is especially important that the dependent variable be conceptualized in ways that allow comparison. �But, as we have argued, comparability does not equal identity.

 

Second, the notion of a complete translation manual (the absence of which is said to doom theory comparison) is not the only way to get at the problem.� Languages, including theoretical languages, are more than one-for-one (or one to many) rules of correspondence. �One cannot learn a language simply by mastering the relations among words and things because languages are also networks of significance as well as rules about the relations among terms and referents. �Languages have a horizontal (significance) aspect as well as a vertical (referential) aspect. �The same is true of theories.� Through careful specification of the theoretical context, we are more confident that we can understand one another.� This implies that the networks of meaning in which concepts are embedded are grasped.

 

The constructivist/rationalist divide has a parallel with our attempt to compare theories.� On the one hand, the search for a complete translation manual�a full listing of the meanings and mappings of the two theoretical languages�resonates with rationality.� On the other, we accept that learning a language takes place partly by acquaintance, immersion, and direct experience, rather than through a formulaic translation manual.� As Kuhn argues, learning a language is interpretive and hermeneutic (2000, p. 56) because sense as well as referents are in play.� Learning a language�natural or theoretical�is partly a process of acquiring an understanding of the networks of significance, or meaning, underlying the language.� This is why language learning, as well as theoretical comprehension, occurs in discrete chunks and infrequent unanticipated Gestalts rather than incrementally.

B. Models of Theoretical Dialogue

Not surprisingly then, our starting point is a belief that rationalism and constructivism possess a degree of commensurability sufficient for fruitful exchange to take place.� In what follows, we seek to promote such a dialogue � one that builds upon other recent efforts (March and Olsen, 1998; Schneider and Aspinwall, 2001; Fearon and Wendt, 2002) -- and do so by identifying four distinct modes of theoretical conversation: competitive testing; additive theory based on complementary domains of application; sequencing of theories; and incorporation (subsumption).� Not all of these approaches enter the arena in the same spirit.� Competitive testing attempts to confirm some theories and refute others.� Additive and sequential approaches are "nicer" since they attempt not to eliminate or subsume but to build a more comprehensive composite in which the whole provides some gains over partial representations, all the while preserving the integrity of the contributions of the parts.� Subsumption, by contrast, is clearly hegemonic in result if not in aspiration.� More powerful theories absorb less powerful ones, perhaps even by criteria established by the more powerful theories, and then, gallingly, (re)produce these "weaker" theories as derived special cases.

1. Competitive Testing

In Conjectures and Refutations, Karl Popper proposes that our theories are merely conjectures the truth of which we cannot know in advance.� At best, all we can do is to subject our hypotheses to critical scrutiny, and invite others to do the same.� The curiosity of the scholar should be directed not just at advancing knowledge claims but also toward devising ways to prove one's own theories false, if indeed they are false.� Strong tests are those that set the bar of confirmation high �that is, they make confirmation more difficult--, while easy tests set the same bar low.� Strong tests are those that explain outcomes when other theories do not.� They discriminate.� Weak tests share explanatory success with many other theories.

 

Competitive testing means that we do not evaluate our claims only against "the evidence," but against other theories as well (and, of course, other theories imply other evidence).� If we have absolutely no baseline knowledge, that is, if our best presumption outside of our theory is a random relationship between X and Y, then our competition is a random distribution (Stinchcombe 1967, pp. 23-24). �But if there are other applicable theories, as there surely almost always are, then our empirical probes should be set up in such a way as to compete with one another.� This is the fundamental starting point of the Campbell and Stanley (1963) approach to competitive hypothesis testing.

 

Exactly how we set up these competitive probes is difficult.� Unless the theories in question are completely orthogonal to one another, this will not be straightforward and any attempt to set up the evaluation one way instead of another will affect the relative strength of the results. �To say this is to recognize that theoretical specification is not innocent.� Different causal diagrams imply differences in correlations among any pair of related variables. [8] � We have more to say on this important matter in the next sub-section.� Suffice it to mention here that competitive testing implies control of variables outside one's theory and that how these controls are executed can affect the comparisons being made.

 

The main problems, however, are not technical but conceptual and methodological in the broadest sense.� How do we evaluate theories against one another?� What standards do we use?� One can imagine various relevant criteria, among them empirical fit (between theory and evidence), scope (how broad is the theory?), logical coherence (to what extent can parts of the theory be connected, even deduced, from other more basic parts, the axioms), falsifiability (the number of ways a theory can be proved wrong, its refutable implications), parsimony (fewness of parameters) and robustness (the resilience of a theory in the face of changing parameters).� We only consider empirical fit, and even here the challenges are enormous.� And whatever criterion we examine in comparing theories, the value-added criterion is always important.� Yet how to establish the relative contributions of two theories even under the simplest conditions can be tricky.� Just think of the "first-mover advantages" of having your favorite theory go first, subsequently leaving to others the task of explaining the residual variance (see below).

2. Domain of Application

The domain of application approach strives for a minimal synthesis in the sense that, while rationalism and constructivism might appeal to completely independent explanatory factors, when combined they could increase our ability to explain the empirical world.� All theories have scope conditions.� The domain approach works by identifying the respective turfs and "home domains" of each theory, by specifying how each explanation works, and finally by bringing together each home turf in some larger picture.� Each theory is specified independently and the result, if successful, is an additive theory that is more comprehensive than the separate theories.� The view of synthesis offered here is modest (certainly no idea of a Hegelian synthesis or transcendence).

 

Scholars have already advanced a number of domain-of-application propositions.� For example, we might imagine that high substantive stakes invite rational calculation, while relatively low stakes allow for non-calculative decision making (Sj�blom, 1993, p. 401; March and Olsen, 1998, pp.952-53; Aspinwall and Schneider, 2000, p. 27).� Or we might postulate that the more routine the behavior, the more easily it is institutionalized (backgrounded).� In organizational theory and general systems theory, for example, those parts of the environment that can be "mapped" in some stable sense are hardwired into the organization and become part of its "lower" (administrative) functions.� Less stable, less easily mapped aspects of the environment remain on the strategic agenda.

 

With regard to the EU literature, it is easy to imagine a theoretical division of labor (hence also a complementarity) based on the explanation of "grand bargains" and day-to-day decision making.� Indeed, Moravcsik (1998) has long held that liberal intergovernmentalism only explains a piece (albeit a preponderantly important one) of the EU puzzle.� But there are other possibilities.� Might it be the case, for example, that intra-institutional, private, and/or highly technical deliberation involves non-strategic, preference- and identity-shaping discourses, while inter-institutional, public, and/or highly political deliberations conform more closely to strategic bargaining among agents with fixed and stable identities and preferences (Checkel, 2001b; see also Risse, Ropp and Sikkink, 1999)?� If we were to throw one or the other approach out, the question itself would be foreclosed.

 

The key to this model of dialogue is properly to specify the scope conditions of each theory, what its domain is, and how it relates to other theories. �If one theory provides some value-added to the other, we can improve our efforts by this approach.� Admittedly, this works best when multiple theories focus on similar explananda, when explanatory variables have little overlap, and when these variables do not interact in their influence on outcomes.

3. Sequencing

Dialogue based on the sequencing of constructivist and rationalist accounts relates closely to the second approach but goes one step further, suggesting that each depends on the other temporally to explain a given outcome.� Where domain-of-application approaches posit different empirical domains within one frame of time for rationalism and constructivism, sequencing approaches suggest that variables from both approaches work together over time in order fully to explain a given domain.� Legro's (1996) "cooperation two-step," in which a culturalist account of preference formation precedes a rationalist account of conflict and cooperation in international relations, provides a clear example. [9]

 

Some of the contributions to a recent volume on the domestic structural effects of Europeanization (Cowles, Caporaso and Risse, 2001) could be characterized as pursuing constructivist-rationalist dialogue through sequencing, although the sequence remains implicit.� In particular, one might grant that the EU treaties respond to rational, lowest-common denominator interstate bargaining (Moravcsik, 1998) but follow Risse (2001) or Checkel (2001a) on Europeanization and identity formation, thereby suggesting that rational institutional choice both shapes and responds to (endogenous) national preferences.� Whatever one's focus, a sequential approach seems possible, if only by problematizing preferences rather than taking them as fixed and given.� Crombez's (1997) discussion of Commission nomination procedures represents a creative attempt to theorize, rather than assume, the preferences of key actors (see also Hug, this volume).� Occam's razor may cut against full theoretical specification of preferences.� Yet, we may find that asking about them and their political/institutional determinants opens up new questions and provides new answers to old ones (see also Johnston, 2001, pp. 491-92).

4. Incorporation (Subsumption)

A fourth way to integrate theories is to show that one can be logically derived from the other.� Theory X is a special case of Y, or vice-versa.� This is a form of reductionism without the "levels of analysis" connotation of that term. �A theory of culture that interprets preferences and networks of significance within a framework of changing relative scarcities is attempting to produce cultural theory from basic economic theory (Becker, 1996; Bowles, 1998; Stigler and Becker, 1977; Sunstein, 1998; von Weisz�cker, 1971, 1984). The latter is thought of as more fundamental because the relation is asymmetric.� Changes in relative scarcities produce cultural changes, not vice-versa.

 

Along similar lines, if constructivism could explain cultural variations, institutions, and preferences then it would have succeeded in stealing a lot of the rationalist thunder.� Weberian sociological theory treats market society and homo oeconomicus as special cases of a wider range of possible (and historically realized) social formations, rather than as natural, inevitable products of modernization or as uniquely efficient equilibria independently discovered by all societies (Weber, 1964, pp. 181-186).� The market itself is a social construction, and the strategic maximizing behavior that it supposedly demands may itself be reducible to social or cultural factors (Elster, 1986, p. 23; Granovetter, 1985, 1992).� A (constructivist) theory of the latter, in this example, would obviate the need for (or, better, provide) a theory of the former.� While as a general matter this approach holds great appeal, advocating it amounts to believing that one approach is derivative of the other, hence does not stand on its own (distinctive) theoretical foundation.� While not always recognized, the imperial project can run in both directions, and not just from rationalism to other areas.

 

We can imagine a variety of EU topics that might be susceptible to such an effort.� For example, the puzzle of sovereignty transfer or delegation to supranational agents, as well as the research program to which it has given rise, might be explainable in terms of a modernist global culture emphasizing logics of specialization and efficiency (see Finnemore 1996, interpreting the work of, among others, John Meyer).� Why do member states weigh costs and benefits of integration and delegation and maximize net gains through efficient bargaining and institutional design? �They do so, by this logic, because they are instructed to do so by a modernist culture or, at a minimum, by supranational entrepreneurs who tell them who they are and where their interests lie (but see Moravcsik, 1999c).� Similarly, we could entertain a variety of propositions that would ground constructivist claims on a foundation of rational maximizing among actors with given identities and preferences.� It may be instructive to think in terms of a repertoire of strategic frames which one "uses" opportunistically in different situations, perhaps � la Goffman's "frame analysis" or "presentation of self in everyday life" (see the contribution of Schimmelfennig in this volume), or in terms of "strategic social construction" of others' preferences and identities (Finnemore and Sikkink, 1998, 2001).

C. Research Design and Methodology

Whichever model of dialogue proves desirable or appropriate, its success will depend upon careful testing based upon sound research design. �The current impasse in the institutional turn requires that we take the next logical step by evaluating our theories in appropriate empirical contexts.� This injunction applies as much to rationalist as to constructivist work (see e.g., Green and Shapiro, 1994; Moravcsik, 1999a), and certainly applies no less to research seeking to integrate the two.� Indeed, we in no way want the quest for dialogue to replace the examination of mid-range propositions in light of evidence.

Beyond �or rather, in service of-- theoretical dialogue, then, we advocate increasing methodological transparency in moving from theoretical to operational concerns.� Before getting into the specifics, a general point suggests itself.� We embrace a view of science committed to adjudicating rival knowledge-claims with evidence. [10] � That is, given that our interest lies primarily in theory and evidence rather than in epistemology, we seek to isolate the former in part by controlling for the latter.� Katzenstein's Culture of National Security (see especially Jepperson, Wendt and Katzenstein, 1996) provides the model and clearly demonstrates the fruitfulness of this approach.� By controlling for epistemological differences, the contributors to that volume sharpened the theoretical contrast between their own constructivism and the rationalism to which they opposed themselves.� While we seek to move beyond the opposition central to that volume, we take up its approach to epistemology.� The choice bears an inevitable cost in the practical exclusion of a body of scholarship of a different epistemological bent. [11] � We thus knowingly proceed partially and incrementally, aware of the terrain left uncovered.� But if Aspinwall and Schneider are right in suggesting that transcending epistemological differences represents a bridge too far, then our choice is one that prevents the best (epistemological agreement) from being the enemy of the good (intra-epistemological, inter-theoretical progress). [12]

 

With this general point in mind, we asked contributors to the volume to confront middle range theoretical propositions with evidence in such a way as to offer discriminating tests of fair, clear, rival hypotheses.� By "middle range" we simply mean operational.� By "theoretical propositions" we have in mind conditional statements that posit explanations for relationships between substantive variables and are, in principle, falsifiable.� By "discriminating tests" we mean that the evidence should permit adjudication as between rival propositions; by implication, propositions should not be observationally equivalent with respect to the issue/evidence in question (see also Checkel and Moravcsik, 2001).� By "fair," we simply intend that straw person arguments should be avoided.

 

More specifically, we foresee four types of tests, or research designs, which we present here in order from least to most demanding.

 

(1)   Single Theory Null Model Test. The first approach tests either rationalist or constructivist propositions in an uncontrolled way against a null model of no effect. beyond random variations. [13] �Such an approach resembles most existing research and least serves the distinctive interests of this volume.� We remain open to it, however, especially to the extent that it represents "state of the art" for each approach.� Negative results, implying acceptance of the null hypothesis, would clarify the domain of application of the tested approach quite definitively.� Positive results would offer less clarity, in line with the Popperian notion of logical asymmetry between confirmation and disproof.� The degree of confirmation of a theory depends not only on the existence of the relationships it asserts but also on the elimination of explanatory causes proposed by other theories (Z, M, and N).� Elimination of the null hypothesis is thus a prerequisite to consideration of other rival hypotheses, not a substitute for it.� Focusing exclusively on one theory may inflate its causal significance to the extent that explanatory factors are correlated.� In any event, contributions using designs of this sort should, at least, contemplate what the "other side" would have to say, and think about ways in which more might be done.

 

(2) Zero-Order Competitive Model. The second approach we term the zero-order but competitive model.� This approach is a serial extension of the first.� Each theory is evaluated against a null model of random effects without controlling for the variables in the other theory.� It is an improvement in that it requires us to elaborate the implications of each theory, and to compare the expectations of each against the empirical evidence. �By doing so, we are likely to learn something about the scope (domain) of each theory's predictions, where the overlap occurs, and thus where tthey are observationally equivalent or distinct.

 

In a world where both theories propose explanatory factors that are uncorrelated, an uncontrolled analysis would be relatively unproblematic, though ironically we have no way of knowing if these factors are uncorrelated without doing the empirical analysis.� Estimates of the separate relationships in both theories would tell us something distinctive and, in the best of cases (where relationships are independent, linear, and additive), we could combine separate empirical results into one comprehensive explanation (Blalock, 1969, pp. 35-36).� However, we are unlikely to find completely uncorrelated causes of a common dependent variable, and when causes are themselves correlated we have no reliable way (within this approach) to disentangle the respective contributions.� Each theory silently profits from its correlation with variables in the other theory, and worse, with variables that may be untheorized but still operate.� Variable X1, a proposed explanatory variable in theory one, will account for more variation in Y (an outcome variable), to the extent that X1 correlates with X2 (in a rival theory) and X3 (an untheorized variable that affects Y and is correlated with X1).

 

(3) Multiple Theory Controlled-Competitive Model. A third approach we label the controlled competitive model.� It attempts to respond to the obvious defects of the second approach by controlling for the explanatory variables in the rival theories. The relevant overall model is X causes Y controlling for Z and Z causes Y controlling for X.� This approach comes closer to the scientific ideal in that control is taken seriously.� The basic idea is to take into account factors that are not included in one's own theory, but that may still affect the outcomes of interest.

 

Control can be accomplished in a number of ways.� The ideal is the fully controlled randomized experiment where individuals are assigned to groups randomly, assuring their equivalence with regard to theorized and untheorized variables. �In Stinchcombe's terms (1968, p. 24), theoretical variables are turned into statistical variables, since all the explanatory variables in rival theories are randomized across groups.� While the classical experiment can rarely be used to study political phenomena, approximations are available.� If one suspects a third variable of affecting outcomes, one can sample observations such that they are homogenous with regard to this variable.� If X is an independent variable and Z a control variable, one can design his/her research so that X varies but Z is constant.� Finally, if we can assess the relation between a third variable and the dependent variable, we can calculate it and take it into account, i.e. we can "take out" the effect of the third variable and see whether there is still covariation between the causal variable and these corrected values.� This is the technique of partial correlation or standardization (Stinchcombe, 1967, p. 37).� A further refinement of this basic logic is represented by path analysis and path coefficients, which are essentially partial, standardized regression coefficients. [14] � In a complex set of relationships among variables with different causal paths, the function of path analysis is to identify the relative strengths of each.

 

(4) Residual Variation Approach. A fourth approach we label the residual variation or value-added model.� It is not so different in underlying logic from the previous approach (it is really a special case) but it provides a different vantage point for comparing theories. We can ask what is the value-added of X?� As a rhetorical device, this is often deployed as a way of defending one's favored theory.� One often hears colleagues ask one another, "what can your theory explain that mine can't already explain?" �If we take any two theories, each with two explanatory components�the first component is shared by both theories, the second is distinctive-- whichever one is entered first will likely be the more powerful, for the simple reason that the first will benefit from its correlation with the second, while the second theory will be discounted to the extent it overlaps with first.� Paradigmatic privileging, then, tilts the playing field in favor of the theoretical first mover.

 

In this category we may place the simple analysis of residuals and stepwise regression.� In the analysis of residuals, one first calculates the regressions of Y on X, then takes the residuals and treats them as new data, using a second set of variables to explain them.� As we move along, things get tougher and tougher, since we are removing the easiest explained variance first and leaving the most difficult for last.� In stepwise regression, we construct a model in which the variable "explaining" the most variation in the dependent variable enters first, then the second, the third and so on.� The model is really constructed out of the empirical algorithm that instructs the computer to select the most important variable first, and so on in accordance with the criterion mentioned above.� Variables entered later are penalized, since variation that might have been explained by the fifth variable has now already been removed by prior variables.� To the extent that overlap exists between later and earlier variables, or between extraneous variables correlated with both early and late entries, the effect of later variables will be underestimated.� As two astute critics put it, stepwise regression cannot simultaneously provide both the model and reliable estimates of underlying relationships (Hanushek and Jackson, 1977, p. 96).

D. Summary

Table 1 summarizes some of the key relationships between the models of dialogue identified above and the research designs set forth to implement them.� We do not intend it exhaustively to list all of the possible connections, simply to identify some of the key inferences that each design might enable in the context of each type of dialogue.� Similarly, while we have employed statistical terminology in adducing each research design, we believe that each applies to carefully-structured qualitative work, including comparative case studies, process-tracing, or narrative analysis of developmental sequences (Becker, 1998).

 

Table 1 .� Research Designs for Theoretical Dialogue

		Research Designs
		Single-Null	Zero-Order Competitive	Multi-Theory Controlled	Residual Variation
Model of Dialogue	Competitive	Cannot advance competitive dialogue, except to establish that one theory works (or doesn't) as a precursor� to further testing, or unless one theory is equivalent to the null hypothesis.	Can advance competitive dialogue.� If theory A outperforms the null and theory B does not, the judgment is easy.� Under certain conditions summary statistics (e.g., R2) can be used to advance competitive dialogue, even where both outperform the null.	Directly advances competitive dialogue, showing effects of which while effects of the other are accounted for.	May advance competitive dialogue, though "unfairly", by allowing claims of theoretical priority for those approaches contributing greater variance explained.
	Domain of Application	Minimally advances DOA dialogue by showing that one does or does not apply in a given domain.� Does not address whether the other does or does not apply there or elsewhere.	Minimally advances DOA dialogue, but only minimally, if serial tests demonstrate that one or the other does or does not apply with respect to a given aspect of the phenomenon to be explained	Can advance DOA dialogue by specifying where each theory applies, controlling for the other.	May advance DOA dialogue, by identifying relative variance explained in a given empirical domain.
	Sequencing	Can minimally advance sequencing dialogue, by showing that one does or does not apply at a given point in a process.� Does not address whether the other does or does not apply at that point or elsewhere.	Can advance sequencing dialogue, but only minimally, if demonstrates that one or the other does or does not apply at a given stage of the process to be explained.	Can advance sequencing dialogue by specifying when each theory applies (to a temporal sequence of interest), controlling for the other.	May advance sequencing dialogue, by identifying relative variance explained at a given point in a temporal process.
	Subsumption	Cannot advance claims of subsumption.� That theory A outperforms the null establishes nothing about the relationship between theory A and theory B.	Cannot advance claims of subsumption.� That theory A and/or theory B outperform (or not) the null or the rival theory says nothing about their relationship to each other.	May be used to disconfirm claims of subsumption, but probably cannot be used to advance them.	May be used to disconfirm claims of subsumption, but probably cannot be used to advance them.

 

 

 

In the context of any given piece of research, the model of dialogue will be a matter not just of taste (e.g., which seems "nicer", or most likely to produce "victory" for one's preferred approach), but also of the phenomenon to be explained.� Dialogue based on sequencing, for example, implies the existence of a sequential phenomenon of interest, and would be inappropriate where phenomena are temporally or causally isolated.� Similarly, the appropriate research design depends in part upon the model of dialogue that one has in mind, and in part upon the types of inferences one wishes to make.� To illustrate with an obvious example, it would make little sense to pursue a competitive theoretical dialogue through a single theory null research design, which implies little about the relative merits of competing theories.� Rationalist-constructivist dialogue is more than a matter of conviction, or of good- (or ill-) will.� It is a matter of careful design and testing.� But each model of dialogue, and each research design, permits certain inferences and forecloses on others.� Thus, these design and methodological choices must be made carefully and transparently.

V. Conclusion

In sum, our approach to integrating institutions involves moving beyond meta-theoretical claims via careful attention to both theory and method.� We sketched four models of theoretical integration (dialogue), ranging from basic opposition (competitive testing), through complementarity (domains of application, sequencing), and all the way to subsumption.� Suggesting that rationalism and constructivism can speak to and understand each other represents only the first step, however.� In addition, we offered four specific research designs aimed at producing empirical tests that can clarify the theoretical landscape.� Each offers distinct advantages and disadvantages, but collectively they point to a range of possibilities that might find application in a variety of circumstances, depending upon the questions being asked, the data being used, and so forth.� Our goal is certainly not to exhaust the possibilities, nor simply to offer a taxonomy of current practice.� We seek instead to identify a menu of alternatives, the refinement or extension of which will depend, in part, upon actual implementation of these ideas.

 

The papers that follow all directly involve themselves with integrating institutions.� Many of them individually, and the collection as a whole, provide a first pass through what is, to a large extent, new terrain.� While we have strongly encouraged each contributor to take up our suggestions in his/her paper, we have not dictated models or methods, nor have we insisted on individual or collective coverage of all of the dialogic and design possibilities.� The collection thus represents, we hope, a beginning rather than a conclusion of the project of integrating institutions.

Contributions

Simon Hug's "Endogenous Preferences and Delegation in the European Union" extends rationalism to an area that it has normally eschewed --preference formation.� The dialog implied by his study is a competitive and possibly a subsumptive one.� He implements it using a single theory null research design, in which he tests rationalist propositions about European Commission preferences and member state delegation to it.� His empirical analysis, while not unequivocal, confirms that certain preferences �those of agents such as the Commission and the European Court of Justice-- can be explained within a rationalist framework.� While Hug's analysis broadens the domain of application of rationalism, we also view it as potentially contributing to a form of sequential synthesis, in which rational intergovernmental bargaining constitutes another actor (i.e., determines its preferences and its very identity).� Far from being problematic, we find just this sort of blurring of lines to be a positive development, an indicator that meta-theoretical dispute is giving way to theoretical synthesis based on transparent and solid empirical testing.

 

Kreppel and Hix's "From 'Grand Coalition' to Left-Right Confrontation" examines the changing nature of party competition in the EU's increasingly powerful, directly-elected European Parliament following the 1999 elections.� They offer a zero-order competitive test of rival rationalist and constructivist propositions that speak directly to the same issue but come to different conclusions.� Refreshingly, they report disconfirming results for both approaches.� That is, neither rational choice nor constructivism can explain the shift from grand coalition to ideological competition that followed the 1999 EP elections.� Given that no singular account performs satisfactorily in this domain, they conclude by sketching a "mixed" (synthetic) model in which a short-run, rational-calculative logic coexists with a longer-run, constructivist account grounded in logics of appropriateness and identity.� We thus view Kreppel and Hix's piece as an important check on simplistic singular accounts and as a starting point for a more robust account of the various determinants of party behavior in the EP and democratic legislatures more generally.

 

Jeffrey Lewis's "Institutional Environments and Everyday EU Decision-Making" takes up a key theme in sociological institutional analysis � the possible constitutive and socialization effects produced by recurring face-to-face interaction within established institutions.� Everyday decisionmaking in the Committee of Permanent Representatives (COREPER) has proven fertile ground for adversarial clashes between rationalists, who are skeptical of socialization effects, and constructivists, who often emphasize their importance.� Lewis undertakes a multiple theory controlled competitive test of these rival claims.� The case study results, while nuanced, provide solid support for a variety of "thin" constructivist claims, including arguments about the importance of institutionally/socially-induced "process interests" that temper hard-headed pursuit of narrowly-calculated national interests.� His study, then, provides considerable insight into the respective domains of application of rationalism and constructivism, using both confirming and disconfirming evidence to clarify the scope conditions of each.� Lewis concludes with an interesting reflection on the Janus-like nature of national officials operating in EU institutions, suggesting that further dialogue, perhaps in the form of sequential synthesis, is both desirable and possible.

 

Mark Pollack's "Control Mechanism or Deliberative Democracy" directly pits the author�s preferred rationalist account, based on principal-agent theories of delegation, against rival constructivist claims in an empirical domain, the EU's complicated system of executive committees known as "comitology," in which the latter has enjoyed previous empirical successes. �The model of dialogue here is competitive, implemented through a multi-theory controlled research design.� Pollack finds that rationalism provides a more compelling account of institutional choice (of comitology procedure) than does constructivism.� Further, he suggests that this result, while preliminary, informs the possibilities for other sorts of dialogue, undermining the applicability of constructivism in related domains, or at other points in a temporal sequence.� Fuller empirical tests of these ideas in the area of comitology would seem to suggest themselves.

 

Frank Schimmelfennig's "Strategic Action in a Community Environment" proposes and tests a sophisticated rational choice approach in the important and timely area of the EU's enlargement to the east.� In particular, he develops a series of testable propositions that help to tease out the domains of application of rationalism, constructivism, and his own synthesis, and proceeds to implement a controlled model of theory testing to assess them.� His empirical analysis considers four stages in the evolution of the enlargement process, showing his synthesis both to account for factors captured in singular (rationalist, constructivist) theories and to explain "new facts."� In that sense, his contribution suggests that both approaches apply in part of the enlargement domain, and that a sequential model of dialogue opens space for both approaches to explain a piece of the puzzle.

 

R. Dan Kelemen's contribution, "The Structure and Dynamics of EU Federalism," considers the EU through the lens of comparative federalism.� Kelemen suggests that in one key area, dealing with the expansion of federal powers, constructivism and rationalism offer complementary rather than competing theoretical claims.� In the area of state autonomy, by contrast, they compete directly.� His research design reflects both of these properties, involving a zero-order competitive�that is, each against the null�test of federal expansion, and a multi-theory controlled test of state autonomy.� He finds that both rationalism and constructivism help to explain different aspects of federal expansion, while rationalism provides the superior account of state autonomy.� These results, then, paint a highly differentiated picture of EU federalism, one in which "either-or" theorizing is, in the aggregate, in appropriate, and in which constructivism and rationalism ally to provide a thoroughgoing account of these important issues.

 

Jeff Checkel's "Going Native in Europe" takes us outside of the EU to examine social interaction in Europe's "other" Council, the Council of Europe.� Checkel develops a series of propositions about the conditions under which argumentative persuasion --a process of convincing someone through argument and principled debate-- will be effective.� Though implicitly competitive with a rationalist rival, these propositions are assessed systematically against a null using a narrative, or process-tracing method.� Checkel thus devotes himself to the identification of scope and boundary conditions for both approaches, which he views as largely complementary.� His empirical analysis, based on panel-sample interviews and documentary research, finds considerable evidence of effective argumentative persuasion within the Council's committees working on citizenship and nationality issues, which in some senses (compared to EU institutions) might be construed as "hard cases" for socialization accounts.� That said, argumentative persuasion confronts specifiable obstacles, as does (rationalist) hard-headed bargaining.� Like many of the other contributors, then, Checkel issues a call for more "both/and" theorizing, work that takes seriously both rationalist and constructivist claims, and attempts to grapple empirically with sorting them out.

 

In sum, the papers provide broad and timely empirical coverage.� Kelemen's account of EU federalism gives a bird's eye comparative perspective of the "nature of the beast" (Risse-Kappen, 1996).� The analysis is all the more relevant given the convening in early 2002 of a constitutional convention where debates over the EU�s federal nature are playing no small role.� Kreppel and Hix's study of the current (1999-2004) European Parliament represents one of the first attempts to make empirical sense of the political and institutional determinants -- and implications -- of the EP's new ideological divide.� Schimmelfennig's analysis of enlargement informs what may be the single most important issue --how to incorporate a raft of new members-- facing the EU in the next ten years.� Hug, Lewis, and Pollack cover the rest of the EU's major institutions [15] (and, to a considerable extent, relations between them), and Checkel's essay extends our institutional coverage to Europe's key human rights organization, the Council of Europe.� In short the collection provides both broad and deep empirical coverage of the Europe�s integrating institutions.

 

Perhaps as important as the substantive glue that holds the collection together is coherence at the levels of theory and design.� Here we gather institutionalists of various stripes, all undertaking carefully structured research (using a variety of techniques, from narrative/process tracing to formal to econometric) that advances the rationalist-constructivist dialogue.� If we take Moser, Schneider, and Kirchg�ssner's (2000) Decision Rules to be the state-of-the-art in rational institutional analysis of the EU, and the special issue of Journal of European Public Policy (1999) to fill an analogous role for constructivists, then we have a "state of the art" or "best of approach" volume for each perspective.� What is more, Schneider and Aspinwall's Rules of Integration (2001) brings rationalists and constructivists under one cover, and encourages communication by having scholars from each perspective comment on work done by those working in a different tradition.� We seek to move the agenda forward by encouraging structured dialogue, and by doing so in a way --through subjection of clearly specified claims to empirical tests-- that will lead to scientific progress rather than metatheoretical bloodletting.

 

This exercise, to the extent that it is successful, serves our threefold attention to integrating institutions.� That is, it represents a collection of papers that explicates the nature and functioning of Europe's integrating institutions, presenting these from a variety of angles, and as such painting a richer picture than tends to be available in work that considers only rationalist or constructivist propositions.� It should also contribute, beyond EU studies, to more introspection and dialogue among institutional theorists, suggesting ways of understanding and using institutions that tend not to occur from within a single approach.� Finally, the collection integrates institutional analysis across the subfields of comparative politics and international relations.� That the EU might effect such an outcome befits its status as a novel political form, one that fuses national and transnational contestation and cooperation --well established and centrally important categories-- in unique but informative ways.

References

Abbott, Andrew. (2001).� Time Matters: On Theory and Method.� Chicago, IL: University of Chicago Press.

Abell, Peter.� (1992).� Is Rational Choice Theory a Rational Choice of Theory?� In J.S. Coleman and T.J. Fararo (Eds.), Rational Choice: Advocacy and Critique (pp. 183-206). Thousand Oaks, CA: Sage Publications

Adler, Emanuel. (1997). Seizing the Middle Ground: Constructivism in World Politics. European Journal of International Relations 3, 319-63.

Adler, Emanuel. (2002). Constructivism and International Relations. In Handbook of International Relations, edited by Walter Carlsnaes, Thomas Risse and Beth Simmons. London: Sage Publications.

Andersen, Svein S., and Eliassen, Kjell A. (1991).� European Community lobbying.� European Journal of Political Research, 20, 173-187.

Aspinwall, Mark, and Schneider, Gerald. (2000).� Same menu, separate tables: The institutionalist turn in political science and the study of European integration.� European Journal of Political Research, 38, 1-36.

Attina, Fulvio. (1990). The voting behaviour of European Parliament members and the problem of Europarties.� European Journal of Political Research, 18, 557-579.

Becker, Gary S. (1996). Accounting for Tastes.� Cambridge, MA: Harvard University Press.

Becker, Howard S. (1998).� Tricks of the Trade. Chicago, Ill.: University of Chicago Press.

Blalock, Hubert M. Jr. (1969). Theory Construction: From Verbal to Mathematical Theory. Englewood Cliffs, N.J.: Prentice Hall.

Bowles, Samuel. (1998).� Endogenous Preferences: The Cultural Consequences of Markets and other Economic Institutions.� Journal of Economic Literature, 36, 75-111.

Caporaso, James A. (1997). Across the Great Divide: Integrating Comparative and International Politics.� International Studies Quarterly, 41, 563-592.

Caporaso, James and Joseph Jupille. (2001). The Europeanization of Gender Equality Policy and Domestic Structural Change.� In M.G. Cowles, J. Caporaso, and T. Risse (Eds.), Transforming Europe: Europeanization and Domestic Change (pp. 21-43). Ithaca, NY: Cornell University Press.

Checkel, Jeffrey T. (1998). The Constructivist Turn in International Relations Theory (A Review Essay). World Politics, 50, 324-48.

Checkel, Jeffrey T. (2001a). The Europeanization of Citizenship? In M.G. Cowles, J. Caporaso, and T. Risse (Eds.), Transforming Europe: Europeanization and Domestic Change (pp. 180-197). Ithaca, NY: Cornell University Press.

Checkel, Jeffrey T. (2001b). Why Comply? Social Learning and European Identity Change.� International Organization, 55, 553-88.

Checkel, Jeffrey T. (2002). International Institutions and Socialization in the New Europe � Chapter 1: Introduction. Typescript. Department of Political Science and ARENA, University of Oslo (April).

Checkel, Jeffrey T. and Andrew Moravcsik. (2001). A Constructivist Research Program in EU Studies? (Forum Debate). European Union Politics 2, 219-49.

Christiansen, Thomas, Knud Erik Joergensen and Antje Wiener, eds. (2001). The Social Construction of Europe. London: Sage Publications.

Coleman, James S., and Fararo, Thomas J. (1992).� Introduction.� In J.S. Coleman and T.J. Fararo (Eds.), Rational Choice: Advocacy and Critique (pp. ix-xxii). Thousand Oaks, CA: Sage Publications.

Cowles, Maria Green, James A. Caporaso, and Thomas Risse, eds. (2001).� Transforming Europe: Europeanization and Domestic Change.� Ithaca, NY: Cornell University Press.

Crombez, Christophe. (1997). Policy Making and Commission Appointment in the European Union.� Aussenwirtschaft 52, 1/2: 63-82.

Diez, Thomas. (1999a). Riding the AM-Track through Europe: Or, the Pitfalls of a Rationalist Journey through European Integration. Millennium 28, 355-69.

Diez, Thomas. (1999b). Speaking �Europe�: The Politics of Integration Discourse. Journal of European Public Policy 6, 598-613.

Dowding, Keith.� (2000).� Institutionalist Research on the European Union: A Critical Review.� European Union Politics 1, 125-144.

Duncan, O. Dudley. (1966).� Path Analysis: Sociological Examples.� American Journal of Sociology, 72, 1-16.

Duncan, Otis D. (1970).� Partials, Partitions, and Paths. In Edgar F. Borgatta and George W. Bohrnstedt (Eds.), Sociological Methodology (pp. 38-47).� San Francisco, Ca.: Jossey Bass.

Eichenberg, Richard C., and Dalton, Russell J. (1993).� Europeans and the European Community: the dynamics of public support for European integration.� International Organization, 47, 507-534.

Elster, Jon.� (1986).� Introduction.� In Rational Choice, edited by Jon Elster, 1-33.� New York, NY: New York University Press.

Elster, Jon. (1989). Nuts and Bolts for the Social Sciences.� New York: Cambridge University Press.

Eriksen, Erik Oddvar and John Erik Fossum, eds. (2000). Democracy in the European Union: Integration through Deliberation?� London: Routledge.

Fearon, James and Alexander Wendt. (2002). Rationalism v. Constructivism: A Skeptical View. In Handbook of International Relations, edited by Walter Carlsnaes, Thomas Risse and Beth Simmons. London: Sage Publications.

Feld, Werner J. (1979).� Implementation of the European Community's Common Agricultural Policy: expectations, fears, failures.� International Organization, 33, 335-363.

Feyerabend, P.K. (1968).� How to be a Good Empiricist�A Plea for Tolerance in Matters Epistemological.� In The Philosophy of Science, edited by P.H. Nidditch.� Oxford, England: Oxford University Press, pp.12-39.

Finnemore, Martha. (1996). Norms, culture, and world politics: insights from sociology's institutionalism.� International Organization, 50, 325-347.

Finnemore, Martha andKathryn Sikkink. (1998).� International Norm Dynamics and Political Change.� International Organization, 52, 887-917.

Finnemore, Martha and Kathryn Sikkink. (2001). Taking Stock: The Constructivist Research Program in International Relations and Comparative Politics. Annual Review of Political Science, 4, 391-416.

Galtung, Johan (1967). Theory and Methods of Social Research.� Oslo, Norway: Universitetsforlaget.

Granovetter, Mark. (1985).� Economic Action and Social Structure: The Problem of Embeddedness.� American Journal of Sociology, 91(3), 481-510.

Granovetter, Mark. (1992).� Economic Institutions as Social Constructions: A Framework for Analysis.� Acta Sociologica, 35, 3-11.

Green, Donald P., and Shapiro, Ian. (1994).� Pathologies of Rational Choice Theory: A Critique of Applications in Political Science.� New Haven, CT: Yale University Press.

Haas, Ernst B. (1976). Turbulent fields and the theory of regional integration.� International Organization, 30, 173-212.

Hall, Peter A., and Taylor, Rosemary C.R. (1996).� Political Science and the Three New Institutionalisms.� Political Studies, 44, 936-957.

Hanushek, Eric A. and Jackson, John E. (1977). Statistical Methods for Social Scientists.� New York, NY: Academic Press.

Hesse, Mary (1970).� "Is There an Independent Observation Language?", in R.G. Colodny (ed.), The Nature and Function of Scientific Theories.� Pittsburgh, Pa.: University of Pittsburgh Press, pp.35-77.Hix, Simon. (1994).� The Study of the European Community: The Challenge to Comparative Politics.� West European Politics, 17, 1-30.

Hix, Simon. (1996).� CP, IR, and the EU!� A Rejoinder to Hurrell and Menon.� West European Politics, 19, 802-804.

Hopf, Ted. (1998). The Promise of Constructivism in International Relations Theory. International Security, 23, 171-200.

Hurrell, Andrew and Anand Menon. (1996).� Politics Like Any Other? Comparative Politics, International Relations, and the Study of the EU.� West European Politics, 19, 386-402.

Jepperson, Ronald L., Alexander Wendt and Peter Katzenstein. (1996).� Norms, Identity, and Culture in National Security.� In P.J. Katzenstein (Ed.), The Culture of National Security: Norms and Identity in World Politics (pp. 33-75).� New York, NY: Columbia University Press.

Johnston, Alastair Iain. (2001). Treating International Institutions as Social Environments. International Studies Quarterly, 45, 487-515.

Journal of European Public Policy. (1999).� Special issue on the social construction of Europe. �Volume 6, 4 (December).

Jupille, Joseph and James Caporaso (1999).� Institutionalism and the European Union: Beyond International Relations and Comparative Politics. Annual Review of Political Science, 2, 409-425.

Katzenstein, Peter J., Robert O. Keohaneand Stephen D. Krasner (Eds.). (1998).� International Organization and the Study of World Politics.� International Organization, 52, 645-686.

Keohane, Robert O. and Joseph S. Nye, Jr. (1975).� International Interdependence and Integration.� In F. Greenstein and N. Polsby (Eds.), Handbook of Political Science, vol. 8, International Politics (pp. 363-414).� Reading, MA: Addison-Wesley.

Kitcher, Philip (1978).� Theories, Theorists, and Theoretical Change.� The Philosophical Review, vol. 87, no. 4:519-547.

Kohler-Koch, Beate. (1994).� Changing Patterns of Interest Intermediation in the European Union.� Government and Opposition, 29, 166-180.

Legro, Jeffrey W. (1996).� Culture and Preferences in the International Cooperation Two-Step.� American Political Science Review, 90, 118-137.

Kuhn, Thomas. (1962).� The Structure of Scientific Revolutions. Chicago, IL: University of Chicago Press.

Kuhn, Thomas. (2000). The Road Since Structure. Chicago, IL: University of Chicago Press.

March, James G. and Johan P. Olsen. (1998). The Institutional Dynamics of International Political Orders. International Organization 52, 943-69.

Milner, Helen V. (1998).� Rationalizing Politics: The Emerging Synthesis of International, American, and Comparative Politics.� International Organization 52, 759-786.

Moravcsik, Andrew. (1991).� Negotiating the Single European Act: national interests and conventional statecraft in the European Community.� International Organization, 45, 19-56.

Moravscik, Andrew. (1993).� Preferences and Power in the European Community: A Liberal Intergovernmentalist Approach. Journal of Common Market Studies, 31, 473-524.

Moravcsik, Andrew. (1997).� Taking Preferences Seriously: A Liberal Theory of International Politics.� International Organization, 51, 513-554.

Moravcsik, Andrew. (1998).� The Choice for Europe: Social Purpose and State Power From Messina to Maastricht.� Ithaca, NY: Cornell University Press.

Moravcsik, Andrew. (1999a).� Is something rotten in the state of Denmark? Constructivism and European integration.� Journal of European Public Policy, 6, 669-681.

Moravcsik, Andrew. (1999b). The Future of European Integration Studies: Social Science or Social Theory? Millennium, 28, 371-91.

Moravcsik, Andrew. (1999c). A New Statecraft? Supranational Entrepreneurs and International Cooperation. International Organization 53, 267-306.

Moser, Peter, Gerald Schneider and Gebhard Kirchg�ssner (Eds.). (2000).� Decision Rules in the European Union: A Rational Choice Perspective.� New York, NY: St. Martin's Press.

Nachmias, David. (1979).� Public Policy Evaluation. New York, NY: St. Martin's Press.

Price, Richard and Christian Reus-Smit. (1998). Dangerous Liaisons? Critical International Theory and Constructivism. European Journal of International Relations, 3, 259-94.

Puchala, Donald J. (1975). Domestic Politics and Regional Harmonization in the European Communities.� World Politics, 27, 496-520.

Risse, Thomas. (2000). 'Let's Argue!' Communicative Action in World Politics. International Organization, 54, 1-40.

Risse, Thomas, Stephen Ropp and Kathryn Sikkink. (Eds.). (1999). The Power of Human Rights: International Norms and Domestic Change. NY: Cambridge University Press.

Risse-Kappen, Thomas. (1996). Exploring the Nature of the Beast: International Relations Theory and Comparative Policy Analysis Meet the European Union.� Journal of Common Market Studies, 34, 53-80.

Ruggie, John Gerard. (1998). Constructing the World Polity: Essays on International Institutionalization.� London: Routledge.

Sandholtz, Wayne and John Zysman. (1989).� 1992: Recasting the European Bargain.� World Politics, 42, 95-128.

Sandholtz, Wayne. (1993).� Choosing union: monetary politics and Maastricht.� International Organization, 47, 1-39.

Sbragia, Alberta M. (1993). The European Community: A Balancing Act. Publius, 23, 23-38.

Sbragia, Alberta. (Ed.). (1991). Euro-Politics: Institutions and Policymaking in the "New" European Community.� Washington, DC: Brookings Institution.

Scharpf, Fritz W. (1988).� The Joint-Decision Trap: Lessons From German Federalism and European Integration.� Public Administration, 66, 239-278.

Schneider, Gerald and Mark Aspinwall. (Eds.). (2001).� The Rules of Integration: The Institutionalist Approach to European Studies.� Manchester: Manchester University Press.

Sil, Rudra. (2000). The Foundations of Eclecticism: The Epistemological Status of Agency, Culture, and Structure in Social Theory. Journal of Theoretical Politics, 12, 353-387.

Sj�blom, Gunnar. (1993).� Some Critical Remarks on March and Olsen's Rediscovering Institutions.� Journal of Theoretical Politics, 5, 397-407.

Smith, Dale L. and James Lee Ray. (1993).� European Integration: Gloomy Theory versus Rosy Reality.� In Dale L. Smith and James Lee Ray (Eds.), The 1992 Project and the Future of Integration in Europe (pp. 9-44).� Armonk, NY: M.E. Sharpe.

Stigler, George J. and Gary S. Becker. (1977).� De Gustibus Non Est Disputandum.� American Economic Review, 67, 76-90.

Stinchcombe, Arthur L. (1967).� Constructing Social Theories.� NY: Harcourt, Brace, and World.

Streeck, Wolfgang and Phillippe C. Schmitter. (1991). From National Corporatism to Transnational Pluralism: Organized Interests in the Single European Market. Politics & Society, 19, 133-159.

Sunstein, Cass R. (1998).� How Law Constructs Preferences.� Georgetown Law Journal, 86, 2637-2652.

Taylor, Paul. (1975).� The Politics of the European Communities: The Confederal Phase.� World Politics, 27, 336-360.

Taylor, Paul. (1991). The European Community and the state: assumptions, theories and propositions. Review of International Studies, 17, 109-125.

Van Hees, Martin.� (1997).� Explaining institutions: A defence of reductionism.� European Journal of Political Research 32, 51-69.

Verdun, Amy. (1999). The Role of the Delors Committee in the Creation of EMU: An Epistemic Community. Journal of European Public Policy, 6(2), ???

Volcansek, Mary. (1986). Judicial Politics in Europe: An Impact Analysis.� New York, NY: Peter Lang.

Volcansek, Mary. (Ed.). (1992). Judicial Politics and Policy-Making in Western Europe. London: Frank Cass.

von Weizs�cker, Carl Christian. (1971). Notes on Endogenous Change of Tastes.� Journal of Economic Theory, 3, 345-372.

von Weizs�cker, Carl Christian. (1984). The Influence of Property Rights on Tastes.� Journal of Institutional and Theoretical Economics, 140, 90-95.

W�ver, Ole. (1998).� The Sociology of a Not So International Discipline: American and European Developments in International Relations.� International Organization, 52, 687-727.

Wallander, Celeste. (1999). Mortal Friends, Best Enemies: German-Russian Cooperation after the Cold War. Ithaca, NY: Cornell University Press.

Weber, Max (1947).� The Theory of Social and Economic Organization.�� NY: Free Press.

Wright, Sewell. (1960).� Path Coefficients and Path Regressions: Alternative or Complementary Concepts?� Biometrics, 16, 189-202.