Analytic Eclecticism in the Study of World Politics ......Rudra Sil ([email protected]) is...

Analytic Eclecticism in the Studyof World Politics: ReconfiguringProblems and Mechanisms acrossResearch TraditionsRudra Sil and Peter J. Katzenstein

This article defines, operationalizes, and illustrates the value of analytic eclecticism in the social sciences, with a focus on the fields ofcomparative politics and international relations. Analytic eclecticism is not an alternative model of research or a means to displaceor subsume existing modes of scholarship. It is an intellectual stance that supports efforts to complement, engage, and selectivelyutilize theoretical constructs embedded in contending research traditions to build complex arguments that bear on substantiveproblems of interest to both scholars and practitioners. Eclectic scholarship is marked by three general features. First, it is consistentwith an ethos of pragmatism in seeking engagement with the world of policy and practice, downplaying unresolvable metaphysicaldivides and presumptions of incommensurability and encouraging a conception of inquiry marked by practical engagement, inclu-sive dialogue, and a spirit of fallibilism. Second, it formulates problems that are wider in scope than the more narrowly delimitedproblems posed by adherents of research traditions; as such, eclectic inquiry takes on problems that more closely approximate themessiness and complexity of concrete dilemmas facing “real world” actors. Third, in exploring these problems, eclectic approachesoffer complexcausal stories that extricate, translate, andselectively recombineanalytic components—mostnotably, causalmechanisms—from explanatory theories, models, and narratives embedded in competing research traditions. The article includes a brief samplingof studies that illustrate the combinatorial potential of analytic eclecticism as an intellectual exercise as well as its value in enhancingthe possibilities of fruitful dialogue and pragmatic engagement within and beyond the academe.

T hree decades ago, Charles Lindblom and DavidCohen lamented that “suppliers and users of socialresearch are dissatisfied, the former because they arenot listened to, the latter because they do not hear much

they want to listen to.”1 Recent advances in theory andmethod do not seem to have remedied the problem, atleast if we are to believe Ian Shapiro: “In discipline afterdiscipline . . . academics have all but lost sight of what

Rudra Sil ([email protected]) is Associate Professor ofPolitical Science at the University of Pennsylvania. He isauthor of Managing “Modernity”: Work, Communityand Authority in Late-Industrializing Japan and Russia(University of Michigan Press, 2002) and co-editor (withDennis Galvan) of Reconfiguring Institutions AcrossTime and Space: Syncretic Responses to Challenges ofPolitical and Economic Transformation (Palgrave 2007).Peter J. Katzenstein ([email protected]) is the Walter S.Carpenter, Jr. Professor of International Studies at CornellUniversity. He has served as the President of the AmericanPolitical Science Association (2008–2009). His most recentsingle-authored books include Rethinking Japanese Secu-rity: Internal and External Dimensions (Routledge, 2008)and A World of Regions: Asia and Europe in the Ameri-can Imperium (Cornell University Press, 2005). Numerousindividuals and institutions have provided valuable feed-back and support at various stages in the writing of thisarticle. With apologies to those whom they may have failed

to mention here, the authors would like to acknowledgeAmel Ahmed, Jeffrey Checkel, Stephen Crowley, MichaelDoherty, Charlotte Epstein, Emiliano Grossman, PeterHaas, Gunther Hellman, Adam Humphreys, Jeffrey Isaac,Nicolas Jabko, Patrick Thaddeus Jackson, Robert Keohane,David Laitin, James Mahoney, Bruce Mazlish, JamesMoskowitz, Ido Oren, Christian Reus-Smit, Srirupa Roy,Benjamin Schiff, Vivien Schmidt, Ian Shapiro, KathleenThelen, Veljko Vujacic, Stephen Watts, Pan Wei, AlexanderWendt, Cornelia Woll, Wang Yizhou, Brigitte Young, andRuizhuang Zhang, as well as three anonymous reviewers.The authors are also grateful for suggestions provided byparticipants at colloquia and workshops at various institu-tions worldwide, including Beijing University, the ChineseAcademy of Social Sciences, Oberlin College, Oxford Uni-versity, Sciences-Po, Tianjin University, University ofFrankfurt, and University of Massachusetts. In addition,Emma Clarke, Stefan Heumann, and William Petti pro-vided invaluable research assistance.

| |�

�

�

Articles

doi:10.1017/S1537592710001179 June 2010 | Vol. 8/No. 2 411

they claim is their object of study.”2 This article is moti-vated by the suspicion that these concerns about academicscholarship are at least partially valid, and that a part ofthe problem is the lack of adequate space in social sciencedisciplines for what we call analytic eclecticism.

Analytic eclecticism does not constitute an alternativemodel of research. It is an intellectual stance a researchercan adopt when pursuing research that engages, but doesnot fit neatly within, established research traditions in agiven discipline or field. We identify analytic eclecticismin terms of three characteristics that distinguish it fromconventional scholarship embedded in research tradi-tions. First, it proceeds at least implicitly on the basis of apragmatist ethos, manifested concretely in the search formiddle-range theoretical arguments that potentially speakto concrete issues of policy and practice. Second, it addressesproblems of wide scope that, in contrast to more narrowlyparsed research puzzles designed to test theories or fill ingaps within research traditions, incorporate more of thecomplexity and messiness of particular real-world situa-tions. Third, in constructing substantive arguments relatedto these problems, analytic eclecticism generates complexcausal stories that forgo parsimony in order to capture theinteractions among different types of causal mechanismsnormally analyzed in isolation from each other withinseparate research traditions.

This is not the first call for something resembling eclec-ticism. In addition to Lindblom and Cohen, numerousscholars have issued pleas for a more practically usefulsocial science—or, following Aristotle, a “phronetic” socialscience—oriented more toward social commentary andpolitical action than toward inter-paradigm debates.3 Ininternational relations, prominent scholars, some even iden-tified with particular research traditions, have acknowl-edged the need for incorporating elements from otherapproaches in order to fashion more usable and more com-prehensive forms of knowledge. For example, KennethWaltz, whose name would become synonymous with neo-realism, argued in his earlier work: “The prescriptionsdirectly derived from a single image [of international rela-tions] are incomplete because they are based upon partialanalyses. The partial quality of each image sets up a ten-sion that drives one toward inclusion of the others . . .One is led to search for the inclusive nexus of causes.”4 Anardent critic of realist theory, Andrew Moravcsik, wouldhave to agree with Waltz on this point: “The outbreak ofWorld Wars I and II, the emergence of international humanrights norms, and the evolution of the European Union,for example, are surely important enough events to meritcomprehensive explanation even at the expense of theo-retical parsimony.”5 Similarly, in an important sympo-sium on the role of theory in comparative politics, severalprominent scholars emphasized the virtues of an “eclecticcombination” of diverse theoretical perspectives in mak-ing sense of cases, cautioning against the excessive “sim-

plifications” required to apply a single theoretical lens tograsp the manifold complexities on the ground.6

As far as programmatic statements go, these views are allconsistent with the spirit of analytic eclecticism. Whetherthese positions are readily evident in research practice, how-ever, is quite another matter. For the most part, social sci-entific research is still organized around particular researchtraditions or scholarly communities, each marked by its ownepistemic commitments, its own theoretical vocabulary, itsown standards, and its own conceptions of “progress.” Amore effective case for eclectic scholarship requires morethan statements embracing intellectual pluralism or multi-causal explanation. It requires an alternative understand-ing of research practice that is coherent enough to bedistinguishable from conventional scholarship and yet flex-ible enough to accommodate a wide range of problems, con-cepts, methods, and causal arguments. We have sought tosystematically articulate such an understanding in the formof “analytic eclecticism,” emphasizing its pragmatist ethos,its orientation towards preexisting styles and schools ofresearch, and its distinctive value added in relating aca-demic debates to concrete matters of policy and practice.

Below, we first offer a brief discussion of the benefitsand limitations of research traditions and consider howanalytic eclecticism complements existing traditions byseeking to leverage and integrate conceptual and theoret-ical elements in multiple traditions. In the next three sec-tions, we elaborate on three distinguishing features ofeclectic scholarship: its pragmatist ethos; its open-endedapproach to identifying problems; and its expansive under-standing of causal mechanisms and their complex inter-actions in diverse contexts. We then consider a small sampleof work in comparative politics and international rela-tions that illustrates the combinatorial potential of eclec-tic scholarship. The conclusion considers the risks andcosts of analytic eclecticism, but views these as acceptablein light of the potential gains of accommodating eclecticapproaches that complement and engage tradition-boundresearch in the social sciences.7

Research Traditions in the SocialSciencesThe evolution of fields and disciplines in the social sci-ences has been accompanied by the emergence of compet-ing approaches or schools, each of which relies on distinctivemetatheoretical postulates in establishing its boundaries.Such postulates frequently address questions such as: whatkinds of phenomena and questions are amenable to socialanalysis; what concepts and methods are best suited forinvestigating these questions; what kinds of observationsconstitute evidence in support of arguments; and whatfactors are most relevant in assessing progress in the field.The necessarily abstract responses to such questions reflectprogrammatic, if implicit, epistemic commitments that

| |�

�

�

Articles | Analytic Eclecticism in the Study of World Politics

412 Perspectives on Politics

frequently define and distinguish competing approachesor schools in a given field.

In identifying these competing approaches, we employthe concept of a research tradition, as articulated by LarryLaudan.8 Following Thomas Kuhn and Imre Lakatos,9

Laudan emphasizes the significance of shared, enduring,foundational commitments related to the conduct and eval-uation of normal scientific research. These include: “(1) aset of beliefs about what sorts of entities and processes makeup the domain of inquiry; and (2) a set of epistemic andmethodological norms about how the domain is to be inves-tigated, how theories are to be tested, how data are to becollected, and the like.”10 Unlike Kuhnian paradigms andLakatosian research programs, however, Laudan’s researchtraditions can coexist and compete for long periods of time,generating substantive claims that may overlap with thoseproduced in other traditions. Laudan even acknowledgesthe possibility of a single scholar working in different tra-ditions even though the foundations of these traditions maybe considered by some to be incommensurable.11 Laudanthus offers us a view of social science that is more flexibleand consistent with the complicated histories of such fieldsas comparative politics and international relations.12 Moreimportantly, Laudan acknowledges the possibility thatdiverse scholarly practices and research products need notbe shoehorned into one of a handful of mutually exclusiveparadigms or research programs. This possibility, in turn,anticipates eclectic approaches that combine analytic ele-ments drawn from separate research traditions.

Research traditions have crucial advantages in generat-ing and organizing initial stocks of knowledge. Given theinfinite complexity of social reality and the limited resourcesavailable to scholars, it is helpful to establish some com-mon assumptions, parameters, vocabularies, and conven-tions to facilitate a focused examination of selected aspectsof that reality. These common understandings also enableadherents of research traditions to establish a consensuson what criteria might be appropriate for assessing thequality of research and for evaluating progress in a givenfield. Moreover, “creative confrontations”13 among rivalscholarly traditions serve as models or foils that prod adher-ents of research traditions to apply concepts and theoriesto new arenas of research, demonstrating their relevanceto substantive issues on which other traditions claim tohave more insights to offer.

For example, in international relations, realism initiallyprovided a common conceptual apparatus for framing andinvestigating problems related to the outbreak of war, theformation of alliances, and the distribution of capabilitiesamong states. Similarly, modernization theory in compar-ative politics provided a common framework for formu-lating questions and generating comparable data on therelationships between economic, social, and political changeacross vast expanses of time and space. In both instances,shared boundary conditions and theoretical vocabularies

employed by adherents of a research tradition facilitatedthe production and assessment of new knowledge claimsconcerning new phenomena. Later, these arguments invitedchallenges and became foils for newer research traditions,as in the case of neoliberalism and constructivism in inter-national relations or of rational-choice theory and histor-ical institutionalism in comparative politics. Each of thesenewer traditions distinguished itself by distinct sets of foun-dational assumptions that facilitated the creation of newproblematiques and new analytic frameworks that helpedto expand the range of substantive arguments and thestocks of empirical knowledge in its respective field. Tothe extent that this stylized process is a reasonable repre-sentation of the changes that have occurred in the twosubfields, it reveals why the emergence of, and competi-tion between, research traditions can expand the fund ofideas, concepts, observations, and theories for a field.

These intellectual benefits are valuable and should notbe forfeited. However, they come at a high price in theabsence of a counterweight in the form of eclectic modesof inquiry. Research traditions establish their identitiesand boundaries by insisting on a strong consensus onenduring and irreconcilable foundational issues. This, inturn, effectively privileges some concepts over others,rewards certain methodological norms and practices butnot others, and places great weight on certain aspects ofsocial reality while ignoring others. In fact, the battlesamong research traditions recur not because of hardeneddifferences over substantive issues but over preexisting epi-stemic convictions about what kinds of social phenomenaare amenable to social analysis, what kinds of questionsare important to ask, and what kinds of processes andmechanisms are most likely to be relevant. Research tra-ditions give themselves permission to bypass aspects of acomplex reality that do not neatly fit within the metatheo-retical parameters they have established by fiat. Theseaspects are either “blackboxed,” relegated to “context,” ortreated as “exogenous.” Such simplifying moves, while help-ful for the purpose of generating elegant knowledge claimsabout particular aspects of reality, are not independentlycapable of generating a more comprehensive understand-ing of complex, multi-faceted problems that interest schol-ars and policymakers alike. For this purpose, scholarlyanalysis needs to be more open-ended, proceeding fromontologies that, as Peter Hall notes, embrace “more exten-sive endogeneity and the ubiquity of complex interactioneffects.”14 This is where analytic eclecticism has a distinc-tive role to play alongside, and in engagement with, dif-ferent strands of scholarship embedded in multiple researchtraditions.

Why Eclecticism?Our defense of analytic eclecticism takes its cue from AlbertHirschman’s famous observation: “ordinarily, social

| |�

�

�

June 2010 | Vol. 8/No. 2 413

scientists are happy enough when they have gotten holdof one paradigm or line of causation. As a result, theirguesses are often farther off the mark than those of theexperienced politician whose intuition is more likely totake a variety of forces into account.” That is not to saythat paradigms are not “useful for the apprehending ofmany elements” in the unfolding of large-scale social trans-formations; but, for Hirschman, the paradigm-focusedsocial scientist tends to focus on only some forces andignore others, thereby running the risk of “a particularlyhigh degree of error.”15

Hirschman’s position is not without empirical backing.In a study of judgmental accuracy under different modesof decision-making, Philip Tetlock has suggested thatgrossly inaccurate forecasts are more likely to result whenexperts behave like “intellectually aggressive hedgehogs,”relying on a single parsimonious approach to explain manythings and depending excessively upon “powerful abstrac-tions to organize messy facts and to distinguish the possi-ble from the impossible.”16 Better forecasts are more likelywhen experts behave more like “eclectic foxes” who areable “to blend hedgehog arguments” and improvise adhoc solutions in a rapidly changing world rather thanbecoming “anchored down by theory-laden abstractions.”17

More recently, Scott Page has argued that long-termprogress and innovation are more likely when a society orgroup depends less on singular solutions offered by bril-liant individuals or like-minded experts and instead poolstogether a broader range of ideas generated by diversegroups of people. Based on his studies of a wide range ofsocial and institutional settings, Page contends: “collec-tions of people with diverse perspectives and heuristicsoutperform collections of people who rely on homo-geneous perspectives and heuristics.”18 In the context ofAncient Greece, Josiah Ober makes a similar observationin the process of analyzing how Athens emerged as the“preeminent Greek polis by a very substantial margin.”The key, Ober argues, was “the distinctive Athenianapproach to the aggregation, alignment, and codificationof useful knowledge . . . dispersed across a large and diversepopulation. . . .”19 What all of these authors are suggest-ing in quite different ways is that, whatever the immediateintellectual payoffs of employing a particular approach,reliance on any one perspective involves tradeoffs thatbecome increasingly costly in the absence of complemen-tary and countervailing efforts to draw upon multiple anddiverse approaches. Analytic eclecticism is such an effort,a means for social scientists to guard against the risks ofexcessive reliance on a single analytic framework and thesimplifying assumptions that come with it.

Importantly, the accommodation of analytic eclecti-cism does not imply the marginalization of scholarshipembedded in research traditions. The value added by ana-lytic eclecticism depends after all upon demonstrating howdifferent sorts of findings and mechanisms emerging from

existing research practices can be reconceptualized andintegrated as elements of more complex explananda. Ana-lytic eclecticism’s distinctive utility stems from its aware-ness of the strengths and tradeoffs of the approachesemployed by existing traditions, and from its recognitionof the particular intellectual gains generated by these tra-ditions in relation to substantive problems. In fact, whatkeeps analytic eclecticism from devolving into a perspec-tive in which “everything matters” is the presumption thatthe analyses produced within research traditions are valu-able for the purpose of identifying many of the factorsthat are likely to matter most. The objective of analyticeclecticism is to uncover how these factors matter in rela-tion to specific research questions, not to generate an ever-expanding list of all imaginable causal factors that caninfluence world politics. Eclectic scholarship that is inatten-tive to theories embedded in research traditions runs therisk of missing important insights, reinventing the wheel,or producing analyses that appear idiosyncratic or unintel-ligible to other scholars. The distinctiveness of analyticeclecticism arises from its effort to specify how elementsof different causal stories might coexist as part of a morecomplex argument that bears on problems of interest toboth scholars and practitioners. This requires engagingand utilizing, not displacing, the well-organized researchefforts undertaken by committed adherents of varioustraditions.

Of course, when drawing upon theories or narrativesdeveloped in competing research traditions, there is thedanger of theoretical incoherence linked to the problemof incommensurability across traditions. The incommen-surability thesis, as articulated by Paul Feyerabend amongothers, argues that the concepts, terms, and standards usedin one theoretical approach, because they are formulatedon the basis of distinct assumptions about knowledge inthe context of distinct theoretical vocabularies, are notinterchangeable with those used in another theoreticalapproach.20 Thus, an eclectic theory drawing upon researchtraditions founded on competing ontological and episte-mological principles can produce an artificial homogeni-zation of incompatible perspectives along with a host ofunrecognized conceptual problems that subvert the aimsof the theory.21

We recognize that much care needs to be taken to ensurethat the relevant concepts, terms, and indicators employedin different research traditions are properly understoodand translated before they are brought into an integratedanalytic framework. The problem of incommensurabilityrepresents a challenge, but there are two reasons we seethis challenge as less serious than assumed. First, as Don-ald Davidson, Hilary Putnam and others have argued,there do exist possibilities for translation and redefinition.Putnam employs the parallel to languages to argue: “if thethesis were really true, then we could not translate otherlanguages—or even past stages of our own language—at

| |�

�

�



all.”22 In fact, Feyerabend himself was primarily con-cerned with the idea of neutral testing protocols that couldbe invoked to compare different types of theories; he nei-ther viewed incommensurability as implying untranslat-ability nor assumed that translatability was a preconditionfor theory comparison.

Second, when social science theories embedded in par-adigms take on substantive research questions, they ulti-mately rely on empirical referents to operationalize variousconcepts, variables, and mechanisms. This provides oneavenue through which specific elements of a causal storywithin one research tradition can be juxtaposed, reconcep-tualized, and possibly combined with elements of a causalstory in a different tradition. It is also possible to breakdown competing explanatory logics into elemental seg-ments in such a way that they become “abstractly compat-ible, such that we could imagine a world in which all wereoperating while we debate how much variants of eachcontributed to any given action.”23 This implies that itmay be possible to temporarily separate metatheoreticalpostulates from specific substantive claims or interpreta-tions so as to enable direct comparison between, and greaterintegration across, the entire range of causal stories thataim to address similar or related empirical phenomena.24

In any case, the challenge of establishing equivalence amongconcepts and mechanisms across research traditions is onlysomewhat greater than that of doing so across diversestrands of research traditions, once we consider that manyof the key metaphysical divides in the social sciences—forexample, between objectivism/subjectivism; nominalism/realism; materialism/idealism; or agency/structure—haveproven to be “fractal distinctions,”25 structuring narrowerdebates within competing traditions. The challenge maybe greater when traveling across traditions, but it is notinsurmountable if proper care is taken to consider thepremises upon which specific analytic components are oper-ationalized in relation to the empirical world.

At the same time, analytic eclecticism should not beconfused with unified synthesis. Although some do usethe term “synthesis” to refer to what we call eclecticism,26

we view synthesis as a more ambitious project. It requiressomething extraordinary and unprecedented: a markeddeparture on the part of most scholarly communities fromtheir original epistemic commitments, followed by a vol-untary convergence upon a new, uniform set of founda-tional assumptions and analytic principles to guide research.In the absence of such a convergence, attempts at synthe-sis are likely to devolve into hegemonic projects in whicha single metatheoretical framework generates substantivetheories concerning diverse social phenomena, while mar-ginalizing or subsuming the insights offered by preexist-ing traditions about many of these same phenomena.27

Analytic eclecticism is more modest and pragmatic. Itis intended to generate diverse and flexible frameworks,each organized around a concrete problem, with the under-

standing that it is the problem that drives the construc-tion of the framework. Moreover, the value added byeclectic scholarship depends to a large extent on the con-tinued success of existing research traditions. Neither aspir-ing to uncover universal laws, nor content with statisticalassociations or interpretations of specific phenomena, ana-lytic eclecticism is best thought of as operating at the levelof what Robert Merton called “the middle range.” Mid-range theories are designed to be portable within a boundedset of comparable contexts where certain cause-effect linksrecur. The task for a mid-range theorist is to recognize theconditions under which some of these links become morecausally significant while others do not.28 Such an effort isquite different from the construction of a grand theory orgeneral law that is intended to be portable not only acrossspatio-temporal contexts but also across a wide range ofsubstantive problems.

Analytic eclecticism may utilize but is not synonymouswith methodological triangulation or multi-methodresearch.29 Any attempt to investigate the interactionbetween general macro- and micro-level processes and spe-cific contextual factors would benefit from attention todifferent kinds of approaches employing different tech-niques of empirical analysis. Yet, it is important to notethat analytic eclecticism does not require the acquisitionor use of multiple methodological skills; it simply requiresa broad understanding of the relative strengths and trade-offs of different methods and an openness to consideringcausal stories presented in different forms by scholarsemploying different methods. The combinatorial logic ofanalytic eclecticism depends not on the multiplicity ofmethods but on the multiplicity of connections betweenthe different mechanisms and social processes analyzed inisolation in separate research traditions. In principle, sucha project can be advanced by the flexible application of asingle method—be it formal modeling, multiple regres-sion, historical case studies, or ethnography—so long asthe problem and the explanandum feature efforts to con-nect theoretical constructs drawn from separate researchtraditions.

This combinatorial logic of analytic eclecticism is evi-dent in, among other fields, the study of institutional change.The first point to note is that the path towards more eclec-tic styles of analysis typically begins with the relaxation ofmetatheoretical postulates and the broadening of analyticboundaries among discrete research traditions. Generallytreated as competing alternatives, economic, historical,and sociological variants of the “new institutionalism”have all sought to explain a wider range of phenomenaemploying a wider range of analytic constructs.30 Histori-cal institutionalists have moved away from the emphasisthey initially placed on institutional persistence linkedto path dependence. They now seek to trace more incre-mental or gradual processes of change that can either gen-erate novel institutional forms over long time horizons or

| |�

�

�

June 2010 | Vol. 8/No. 2 415

produce unexpected breakdowns at critical thresholds.31

Economic institutionalists have gone beyond the treat-ment of institutions as emergent self-enforcing equilibriaproduced by individual-level preferences. They now seekto make sense of institutional change by considering theimplications of shifting parameters and iterated games, andby exploring how social norms affect the supply of infor-mation and the expectations of actors engaged in bargain-ing.32 In addition, sociological institutionalists havegenerated more complex understandings of how sharedworldviews, cognitive scripts, and normative templates mayinteract with discursive or symbolic practices to influenceinstitutionsand institutional actors ingivencontexts.33 Theseshifts do not themselves constitute a full-blown embrace ofeclecticism. However, by stretching the analytic bound-aries initially established by each of the new institutional-isms, they open the door for more self-consciously eclecticapproaches to the study of institutional change.

One example is Mark Blyth’s Great Transformations, acomparative analysis of dramatic institutional change inthe U.S. and Sweden.34 Blyth challenges both historicaland rational-choice institutionalism by demonstrating howthe emergence and appeal of particular economic ideasunder conditions of uncertainty alter existing preferencestructures, coalitions, and institutional orders. Parting wayswith sociological institutionalists who treat ideas and sym-bols as mechanisms of stability, Blyth emphasizes the trans-formative potential of ideational mechanisms in situationsof extraordinary uncertainty. At the same time, his analy-sis draws from all three of the new institutionalisms inexamining the complex interplay of old and new ideas,individual preferences, coalition-building processes, andconstraints related to preexisting institutional orders. ForBlyth, individual interests certainly influence the construc-tion and maintenance of institutions. However, economiccrises create conditions of “Knightian uncertainty” duringwhich existing institutions are ineffective and preferencesare difficult to articulate, let alone act upon. In such anenvironment, new economic ideas have an opportunity toprovide simplifying blueprints for stabilizing and coordi-nating expectations, helping to build new coalitions forchange and providing scientific and normative founda-tions for this change. This approach enables Blyth to showhow the United States and Sweden, typically viewed assharply distinct types of capitalism, responded to crises byshifting policies in the same direction: by promotingembedded liberalism and accommodating the interests oflabor following the 1930s’ crisis, and then dismantlingthese arrangements in response to the demands of coali-tions dominated by business following the 1970s’ crisis.

John Campbell’s Institutional Change and Globalizationis also cognizant of the limitations of each of the threenew institutionalisms, particularly in specifying the differ-ent configurations of mechanisms responsible for evolu-tionary and revolutionary processes of institutional change

along different dimensions.35 Significantly, Campbell iscareful to note that he is not interested in a grand synthe-sis that will dissolve distinctions among paradigms andprovide a unified approach to studying institutional change.Instead, he engages in a process of “bricolage” that involves“selecting various ideas from different places and combin-ing them in ways that yield something new.”36 Campbell’sconception of agency severs the connection between self-interest and intentional action and makes room for ide-ational mechanisms, both cognitive and normative, thatshape actors’ preferences and enable them to creativelyframe problems that emerge in various arenas of institu-tional life. Moreover, in Campbell’s open-ended concep-tion of institutional change, path-dependency in somedimensions coexists with processes in which old institu-tional characteristics are modified or recombined in otherdimensions. The emergence or reproduction of specificinstitutional features reflects the diffusion of particularpractices as well as the actions undertaken by institutionalentrepreneurs in response to a given problem. This frame-work provides the basis for Campbell’s substantive inves-tigation of tax levels and tax policies. Whereas conventionalaccounts predicted a worldwide reduction in taxation lev-els in response to globalization, Campbell finds that theaverage tax burden actually increased and that neoliberaltax reforms proved to be politically less sustainable thanreforms in other institutional arenas. Campbell’s analysisaccounts for this by tracing the processes through whichcomplex interactions among ideational, regulative, evolu-tionary, and diffusion mechanisms mediate the effects ofglobal economic forces on institutional change.

Further below, we consider a broader sample of eclecticscholarship drawn from comparative politics and inter-national relations. For now, we elaborate on the signifi-cance of the three markers we employ to identify analyticeclecticism in practice. Each is defined in flexible terms soas to preclude specific injunctions, but each is also clearenough for the purposes of distinguishing eclectic fromtradition-bound scholarship. The first is a broadly prag-matist ethos, whether implied or proclaimed; the secondis an effort to formulate problems in a manner that seeksto trace rather than reduce complexity; and the third isthe construction of causal stories focused on the complexprocesses through which different types of mechanismsinteract. The next three sections address each of thesefeatures.

The Pragmatist Ethos of AnalyticEclecticismMuch research in the social sciences is founded on a pos-itivist view of social knowledge. While disagreeing on cer-tain ontological and epistemological issues,37 positivistsgenerally embrace a view of inquiry in which patterns ofhuman behavior are presumed to reflect objective laws or

| |�

�

�



law-like regularities that exist above and beyond the sub-jective perceptions of actors and scholars. Various kinds ofinterpretivist and subjectivist approaches also differ in termsof specific assumptions, objectives, and methods,38 butthey generally share a common skepticism about the pos-sibility of inferring generalizations about human behaviorand instead commit to an understanding of meaningfulaction among actors within the context of their immedi-ate social environments. In view of the continuing hetero-geneity of perspectives on the most fundamental questionsconcerning the goals, premises, and methods of scholarlyinquiry, eclectic approaches tend to (at least implicitly) setaside metatheoretical debates in favor of a pragmatist viewof social inquiry.39

Pragmatism has its roots in the late nineteenth andearly twentieth century writings of American philoso-phers, most notably “the canonical trinity”40 of JohnDewey, Charles Pierce, and William James. Challengingthe Kantian tradition that continued to guide Europeanphilosophy, pragmatists held that “philosophy should con-cern itself with the messiness of human meaning.”41 Thisimplied both a rejection of the positivist “dualism of aknowing subject and a known object” and a “refusal toembrace skepticism or subjectivism.”42 Following a periodof marginalization precipitated by the behavioral revolu-tion and the ascent of analytic philosophy, pragmatismhas experienced a resurgence, carried forward by RichardRorty, Hilary Putnam, and others who have offerednuanced critiques of positivism without surrendering torelativism or subjectivism.43 Although multiple strands ofpragmatism and neopragmatism have emerged from quitedifferent intellectual milieux,44 they share certain basictenets that we consider to be a loose philosophical ground-ing for eclectic research practice.

One of these tenets is seen in pragmatists’ aversion toexcessively abstract or rigid foundational principles infavor of a focus on the consequences of truth claims inrelation to different strategies for addressing social prob-lems. Following James, pragmatists seek to bypass “meta-physical disputes that otherwise might be interminable,”and instead to “try and interpret each notion by tracingits respective consequences” in concrete situations.45 Inthe analysis of world politics, a pragmatist perspectivethus implies that competing approaches need to be refor-mulated to facilitate reflections on both how a problemis constituted and how it is to be solved.46 Pragmatismalso speaks to how scholars contend with the multiplicityand diversity of historical narratives, encouraging contin-uous engagement with different strands of historiogra-phy for the purpose of building a tentative consensus on“facts” that can be deployed to cope with contemporaryproblems.47

A second principle concerns the creative process throughwhich aspects of knowledge production are utilized andadapted in different situations. For Dewey, this process of

“reconstruction” involves a rearrangement and updatingof beliefs, habits, and practices in relation to efforts tomaintain or restore equilibrium in social life.48 As Deweyput it: “We take a piece of acquired knowledge into aconcrete situation, and the results we get constitute a newpiece of knowledge, which we carry over into our nextencounter with our environment.”49 In the context ofsocial scientific research, this suggests that knowledgeclaims, however produced and defended, are always inneed of reconsideration and reconstruction on the basis ofengagement with the experiences of actors seeking to copewith real-world problems. In this sense, pragmatism notonly requires a “spirit of fallibilism”50 on the part of thescholar, but is also consistent with the aims of middle-range theorizing because it seeks to “clip the wings ofabstract concepts in order to ground philosophy in theparticularities of everyday life.”51

A third pragmatist notion has to do with the relation-ship between inquiry and dialogue. Pragmatists under-stand that competing knowledge claims are frequentlyconstructed within the context of scholarly communitieswith their own rules and methods. However, they placegreater emphasis on the ensuing process of dialogue andreflection within a more open community in which par-ticipation and deliberation are counted upon to legitimizewhatever consensus emerges in relation to specific prob-lems.52 The barriers that separate academic debate andpublic discourse are not always easy to break down incontemporary settings, but there do exist possibilities for“theorizing that is attentive to practical difficulties andlatent possibilities” in society and in public life.53 In theanalysis of world politics, this implies that open delibera-tion is not merely a procedural commitment to improvethe extent and quality of information available to policy-makers. It is a reflection of the fact that the production ofknowledge about world affairs is fundamentally a socialand discursive activity, linked to the negotiation of con-sensual norms and the legitimation of institutions govern-ing international life.54

A fourth tenet, based largely on the work of GeorgeHerbert Mead, is its open-ended ontology. Mead offeredan extensive account of the processes through which theevolution of the “mind” depends on its relationship withmeanings shared with other “minds” in a social environ-ment, and in which the “self ” is constructed and recon-structed in continuous dialogue with others in thatenvironment.55 Mead’s pragmatist social psychology servesas a point of departure for quite different intellectual per-spectives in contemporary social science, including struc-turationist social theory, rational-choice approaches as wellas constructivist approaches stressing the process and sig-nificance of collective identity-formation in internationalarenas.56 It is likely to have even more appeal for eclecticscholarship, suggesting that how and why some agentschoose to reproduce, while others redefine or transform,

| |�

�

�

June 2010 | Vol. 8/No. 2 417

existing material and ideational structures are questions ofempirical inquiry that cannot be settled by fiat.57

These basic notions, while distilled in a manner thatmay appear oversimplified to philosophers of pragma-tism, capture the ways in which analytic eclecticism rep-resents a kind of pragmatist inquiry. As such, it may bedifferentiated from, but is not fundamentally opposed to,scholarship embedded in contending research traditions.In effect, pragmatism offers a reasonable basis upon whicheclectic scholarship can meaningfully utilize elements ofdiverse research traditions while engaging substantive issuesof policy and practice without becoming trapped in unend-ing and unresolvable debates over epistemic commit-ments and analytic principles.

Widening the Scope of ProblemsProblem-driven research is frequently opposed to method-driven research, although even the staunchest advocates ofmethodological rigor are increasingly eager to demon-strate the problem-driven character of their work.58 Theissue, in our view, is not whether the social sciences oughtto be problem-driven or method-driven but rather howproblems are identified and formulated. Projects embed-ded in different research traditions frequently address sim-ilar or related substantive issues but frame these issues inorder to focus on specific aspects in keeping with theirtheoretical priors.59 Such simplification of social reality iscertainly understandable, even necessary. However, theextent to which and the manner in which large parts ofthat reality are simplified in the formulation of problemsmatter for the purpose of generating insights that bear onthe choices and actions of actors coping with complexsubstantive problems. A pragmatist conception of analyticeclecticism invites us to consider how the problems asdefined within research traditions might (or might not)relate to each other and to concrete dilemmas related topolicy and practice.

For the sake of simplicity, we rely here on a distinctionbetween substantive and analytic problems. The formerrelate to issues that exist apart from academic discourseand that constitute practical dilemmas facing social andpolitical actors. Scholars are often interested in these issuesbut focus only on certain aspects in the course of formu-lating their own analytic problems. The latter are posed insuch a way as to reflect the ontologies, epistemic princi-ples, and theoretical vocabularies embraced by adherentsof a given research tradition. What is worth problematiz-ing for adherents of a research tradition may be a functionof gaps in the existing literature within that tradition or ofthe particular features of a substantive problem that canbe represented within its analytic confines. In the process,dimensions of the problem that have equal or greater sig-nificance in the realms of policy and practice may be brack-eted out.

Following Lindblom and Cohen,60 we see no reason toprivilege exercises driven by such analytic problems overproblems aimed at tackling substantive problems in polit-ical life. At the same time, the investigation of differentlyformulated analytic problems within contending researchtraditions frequently offer relevant insights for the pur-poses of solving substantive problems. The challenge is tocompare and selectively integrate these insights so thatthey can be more practically useful in relation to substan-tive problems. This is where analytic eclecticism comes in.Eclectic scholarship does not provide “better” answers toproblems articulated within a given research tradition. Itsutility lies in recasting problems so that they have widerscope and can thus incorporate related aspects of morenarrowly circumscribed analytic problems that adherentsof research traditions prefer to tackle. Given their expandedscope, the kinds of problems addressed by eclectic schol-ars are more likely to have concrete implications for themessy substantive problems facing policymakers and ordi-nary social and political actors.

One area where we see efforts to broaden the scope ofproblems is the study of social movements. In the past,relative deprivation theory (e.g., Ted Gurr), macro-structuralist accounts (e.g., Theda Skocpol), resource mobi-lization theory (e.g., Charles Tilly), rational choice theory(e.g., Dennis Chong), opportunity-structure arguments(e.g., Sidney Tarrow), and “framing” approaches (e.g.,Robert Benford and David Snow) have been presented asdiscrete traditions in the analysis of social movements.61

Although usually presented as competing perspectives,these approaches are organized around related but differ-ent questions, each focused on a different segment of theprocess whereby grievances ultimately lead to a transfor-mation of the status quo. Relative deprivation and ratio-nal choice theories are primarily concerned with the processthrough which individual grievances are aggregated intocollective protest. They focus, respectively, on the psycho-logical dynamics and cost-benefit calculations that spurindividuals to commit to risky forms of collective action.What antecedent conditions created the choice situationsfacing these actors and what factors are necessary for amovement to succeed are not problematized. Structural-ist approaches focus on the question of the preconditionsthat make existing institutional and social structures vul-nerable to contentious politics. In Skocpol’s approach,for example, international competitive pressures and pres-sures applied by a restive peasantry combine to weakenthe state and facilitate social revolution. The issue of whygrievances arose among peasants is not pertinent to theproblem as she formulates it. Approaches focused onresource mobilization and political opportunity centeron the processes through which a social movement iscapable of sustaining effective collective action within agiven set of institutional constraints. These efforts do notoffer an answer to the question of what structural condi-

| |�

�

�



tions or individual motivations permit the emergence ofcollective action. In other words, none of these approachesis intended or able to address the broader problem ofidentifying where, whether, when, and how grievancesproduce transformations of the status quo.

More recently, a number of scholars have sought tobuild more integrated frameworks intended to bridge dif-ferent types of perspectives in order to generate a morecomprehensive understanding of social movements.62 Thismove is not being driven by a desire for a synthetic grandtheory. It is being driven by a recognition of the multi-plicity of processes, at the structural and individual levels,that need to come together in order for grievances to beconverted into protest behavior, and for protest actions tobe aggregated into a social movement that can carry outits transformative agenda. Karl-Dieter Opp’s “structural-cognitive” approach, for example, reformulates variousmacro approaches (e.g. stressing environmental condi-tions or political opportunity structures) and microapproaches (e.g. stressing incentives or cognitive framing)related to the process through which individual protestactions emerge, coalesce into collective protest, and trans-form the macro-structural conditions, including shiftingopportunity structures.63 In effect, the scope of the prob-lem has been widened to encompass more of the segmentsof the extended process through which initial conditionsconducive to protest give rise to a transformation of thestatus quo. This does not suggest that the earlier studies ofsocial movements are inconsequential or incorrect. Farfrom it. Precisely because these earlier studies yielded pow-erful insights into particular aspects and processes relatedto social movements, it makes sense to explore whetherthese insights can be integrated which, in turn, requiresbroader problems that subsume the specific aspects or pro-cesses targeted by existing traditions.

Of course, a research tradition may make assumptionsand impose boundary conditions that make sense for sometypes of questions but not for others. Indeed, the compe-tition among traditions is often as much about whichquestions are more important as about the relative utilityof particular concepts, mechanisms, or methods. For exam-ple, neorealism trained its sights on questions related tohow the balance of power in a given setting affects thelikelihood of conflict or stability. It did not initially con-cern itself with the main questions on which liberalismconcentrated its attention: the extent of economic inter-dependence and its effect on the prospects for greater insti-tutionalized cooperation among self-interested state actors.At the same time, realist principles can be reformulated increative ways to shed light on how the distribution ofpower affects states’ economic policies and the conduct ofinternational economic relations.64 Similarly, the trade andconflict literature suggests that concepts and measures ini-tially developed within neoliberal institutionalism to cap-ture interdependence can be refined, reconstructed, and

deployed to analyze the prospects of conflict in the inter-national arena.65 Thus, there need not be a one-to-onecorrespondence between research traditions and the ana-lytic problems that initially inspired them. In fact, theseproblems can be reformulated and combined insofar asthey are connected to a common substantive dilemma inthe world of policy and practice.

Causal Complexity and theMultiplicity of MechanismsWhile research traditions generate quite varied researchproducts—ranging from formal models and causal infer-ences to historical narratives and ethnographies—we fol-low Andrew Abbott in viewing all of these as offeringcausal stories based on particular “explanatory programs.”66

These programs may differ in their fundamental aims andassumptions, but as Ernst Haas and Peter Haas note, it ispossible to intersubjectively approach substantive phenom-ena with the aim of generating a “causal understanding ofthe interplay between forces typically analyzed by discreteschools.”67 To explore this possibility and the barriers toit, we rely on the concept of mechanism, a key feature incausal stories cast at the level of the middle range.

Although eclectic research need not be framed in the lan-guage of mechanisms, the concept has heuristic value in dis-tinguishing causal stories presented in eclectic accounts fromthose constructed within research traditions.The latter, pro-ceeding on the basis of particular ontological and episte-mological assumptions, implicitly or explicitly focusattention on certain types of mechanisms while ignoring ordefining away others. Analytic eclecticism, by contrast, offerscomplex causal stories that incorporate different types ofmechanisms as defined and used in diverse research tradi-tions. That is, rather than privilege any specific conceptionof causal mechanism, analytic eclecticism seeks to trace theproblem-specific interactions among a wide range of mech-anisms operating within or across different domains andlevels of social reality.68 This section considers alternativeconceptions of mechanisms that insist on particular defin-ing characteristics, then offers a more open-ended defini-tion that is consistent with the goals and requirements ofeclectic scholarship.

Contemporary treatments of mechanisms challenge notonly the quest for universal laws and general theories butalso the empiricist position struck by Gary King, RobertKeohane, and Sidney Verba (KKV). In their view, causaleffects are logically prior to and more reliable than unob-servable mechanisms; the latter are significant mainly inso-far as they induce new observations or influence the levelof confidence in causal inferences.69 Alexander Georgeand Andrew Bennett, however, emphasize that mecha-nisms operate at the ontological level and should not beconflated with or subsumed under hypothesized causaleffects.70 Others see KKV as reducing scientific inference

| |�

�

�

June 2010 | Vol. 8/No. 2 419

to a specific type of inference based on a “quantitativeworldview,” resulting in a conception of mechanisms asmere “servants of inferences.”71 Thus, if the concept ofmechanism is to have a distinctive analytic function, itneeds to be understood as a process link within, but poten-tially independent from, a more general explanatory modelor descriptive narrative.72 This suggests that all causal sto-ries, however abstract or bounded, may be viewed as con-figurations of mechanisms that explain how some set ofinitial conditions in one or more contexts generates someset of outcomes or variations. Beyond this common viewof how mechanisms are to be positioned vis-à-vis empiri-cal observations and general laws, however, there is littleagreement on what constitutes a causal mechanism.73 Cur-rent debates over the nature and operation of mechanismsreveal at least three contentious issues.

First, there is the question of observability. Peter Hed-ström and Richard Swedberg emphasize the primacy of“unobserved explanatory mechanisms” in social scientificexplanation.74 Although George and Bennett acknowl-edge that new technologies may make “unobservable” ele-ments “observable” at a later time, they nonetheless assert:“No matter how far down we push the border between theobservable and the unobservable, some irreducibly unob-servable aspect of causal mechanisms remains.”75 On theother side, Barbara Reskin insists on observability as a defin-ing criterion for all but intra-psychic mechanisms, whetherinterpersonal, societal, or organizational.76 An even strongerposition is staked out by Mario Bunge, who insists on a mate-rialist definition of mechanism along the lines of what isassumed by natural scientists and engineers: “mechanismsare processes in concrete (material) systems, whether phys-ical, social, technical, or of some other kind.”77 This posi-tion dismisses a priori the possibility of mechanisms locatedwithin conceptual or semiotic systems that could directlyaffectoutcomeswithout requiring the interventionofobserv-able actions or processes.78

Scholars also differ over whether mechanisms mustnecessarily exist at a level of generality that transcendssingular spatio-temporal contexts. For Jon Elster, mech-anisms, while less general in scope than laws, are “fre-quently occurring and easily recognizable causal patternsthat are triggered under generally unknown conditionsor with indeterminate consequences.”79 Gudmund Hernesgoes further, arguing that a mechanism is no less generalthan a causal law since it is essentially “an abstract repre-sentation that gives the logic of a process” unfolding indifferent contexts.80 For others, mechanisms not onlyexist independently of theories and laws, but they maybe unique to, and embedded within, a single temporaland spatial context. That is, the absence of portabilitydoes not make a mechanism any less causally significantsince social action is often the result of individual calcu-lations in which contextual factors play crucial roles.81

Colin Wight emphasizes that, since the point of the con-

cept is to capture a “sequence of events and processes(the causal complex) that lead to the event,” it is entirelypossible to arrive at a configuration of mechanisms thatis unique to a particular time- and space-bound context.82

Perhaps the most heated arguments over the nature ofmechanisms stem from the long-standing debate over meth-odological individualism/holism. For some, the search formechanisms assumes an unequivocal commitment to reduc-tionism. Thomas Schelling, for example, sees a mechanismas a set of “plausible hypotheses” that explain social phe-nomena so long as they incorporate an “interpretation, interms of individual behavior, of a model that abstractly repro-duces the phenomenon that needs explaining.”83 In morerecent years, methodological individualists have intro-duced different kinds of aggregation mechanisms to copewith the disjuncture between individual intentions andactual outcomes. This is evident, for example, in TimurKuran’s treatment of the social outcomes resulting fromindividual-level decisional dissonance, and in RogerPeterson’s analysis of the role of tipping mechanisms thatcapture how the choices of a critical mass of individuals alterthe pay-off matrices of larger groups of individuals.84

At the same time, as George and Bennett note, even ifagents are assumed to be necessary for mechanisms to takeeffect, uniformities in individual behavior suggest the pos-sibility of mechanisms operating at the level of social struc-ture.85 These structural mechanisms are distinct from theaforementioned aggregation mechanisms. Their putativeeffects on outcomes do not require the intervention ofindividual agents. According to Charles Tilly, “relationalmechanisms (such as brokerage) and environmental mech-anisms (such as resource depletion) exert strong effects onpolitical processes without any necessary connection toindividual-level cognitive mechanisms.”86 Relational mech-anisms do not only operate through the structure of observ-able social interactions and networks, but can also takethe form of collective ideas and symbols that directly engen-der or influence macro-level transformations.87 HermanSchwartz also identifies system-level mechanisms drawnfrom economic geography that bypass agency in directlylinking the spatial distribution of productive factors tomacro-level developmental outcomes.88 In all these cases,mechanisms posited at the collective level are viewed asgenerating outcomes at the macro-level without the medi-ation of individual choices or actions.

In spite of their differences over how mechanisms oughtto be defined, many of the aforementioned scholars con-verge on the point that multiple mechanisms combine togenerate social phenomena. From a materialist perspec-tive, for example, Bunge argues: “Highly complex systems. . . have several concurrent mechanisms. That is, theyundergo several more or less intertwined processes at thesame time and on different levels.”89 From a very differentperspective, Tilly also notes that the cumulative effectsof mechanisms “vary considerably depending on initial

| |�

�

�



conditions and on combinations with other mecha-nisms.”90 The problem lies less in the principle than inhow it is translated into research practice given the episte-mic commitments of research traditions. By focusing onlyon certain kinds of mechanisms and ignoring or definingaway others, adherents of research traditions risk missingthe complex processes through which diverse mechanismsrelate to one another.

To offset this risk, analytic eclecticism is predicated ona definition of mechanism that is more open-ended thanmost. This is not because we are hedging our bets or refus-ing to take sides. It is because eclectic analysis requires adefinition that minimizes a priori constraints on inquiryand leaves the door open to consideration of a wider rangeof causal processes operating in different levels and domainsof social reality. Accordingly, we define mechanisms as allentities—whether individual actions or choices, social rela-tions or networks, environmental or institutional character-istics, specific events or contextual factors, individual cognitivedispositions or collectively shared ideas and worldviews—that generate immediate effects through processes that may ormay not recur across contexts and that may be, but often arenot, directly observable.

Such a definition still covers the analytic function attrib-uted to mechanisms as process links within causal stories.However, it would also permit traveling across diverse theo-retical terrain to enable consideration of multiple causalforces operating in different domains of reality and acrosslevels of generality.This definition allows for the possibilitythat unobservable mechanisms often account for the regu-larities we observe, but also sees no reason to insist on observ-ability as a defining attribute of mechanisms, especially sinceit is always possible to “observe” analytical constructs viathe indicators operationalizing them.91 Our definition alsodoes not insist that mechanisms be limited in scope vis-à-vis laws or that they must recur across multiple contexts.We grant that many interesting mechanisms operate acrosstime and space, but we need to leave open the possibility ofsingular phenomena having effects in the same manner thatrecurrent phenomena can. Neither generality and con-stancy, nor specificity and complexity need to be intrinsicattributes of a mechanism.92 On the issue of methodolog-ical individualism/holism, too, we resist the temptation toprivilege either mechanisms that operate through individ-ual cognition and action or ones that exert direct macro-level effects without the intervention of agents.For most substantive problems that have greater scope thanthose formulated within research traditions, useful causalstories are likely to incorporate the interactive effectsofmech-anisms operating across levels of social reality.

In sum, the pragmatist ethos and epistemological agnos-ticism of analytic eclecticism inclines it towards a morecomplex view of causality in which different types of mech-anisms interact to generate outcomes of interest in differ-ent contexts. As noted above, this does not require an

ever-expanding list of all imaginable causal factors. Giventhe time, energy, and attention that research traditionshave implicitly or explicitly devoted to demonstrating theeffects of particular mechanisms, it makes sense to focusone’s attention on how to frame problems and analyses inways that allow these diverse mechanisms to coexist andinteract within comparable contexts. The result is likely toinvolve a more complex configuration of mechanisms thanis typical in most social scientific research. But greatercomplexity is precisely what policymakers and ordinaryactors contend with as they address substantive problemsin the course of everyday politics. Scholars who wish tohave their research speak to such problems must also bewilling to contend with complexity.

Toward Eclecticism in the Study ofWorld PoliticsAlthough still comparatively rare in the social sciences,eclectic scholarship is beginning to make an impression incertain fields. Such eclecticism may be identified in rela-tion to distinct strands within a broadly defined researchtradition.93 In the interest of brevity, we focus here oneclectic scholarship that cuts across research traditions inthe fields of international relations and comparative poli-tics. We neither pretend to offer an adequate summary ofthe arguments considered, nor assess their substantive accu-racy or explanatory power. We do, however, view theseworks as meeting our three criteria for analytic eclecti-cism: they take on problems of broad scope, they developcomplex causal stories at the level of middle-range theory,and they implicitly seek pragmatic engagement within andbeyond the academe. To this extent, we regard these worksas reasonable approximations of analytic eclecticism.

In the study of international security, Robert Jervis’sAmerican Foreign Policy in a New Era represents a creativemove in the direction of eclecticism.94 While the study isfocused on the United States’ policies in the post-ColdWar era, the analysis is predicated on the assumption thata revolutionary transformation has taken place in the inter-national system: A distinctive kind of security communityhas emerged, consisting of the most powerful and devel-oped states in the world, each of which has forsaken theuse of force in its dealings with other members (as evidentin the absence of official war plans). Although many takethis state of affairs for granted, Jervis points out that it is anovel phenomenon that needs to be problematized andexplained. Even when security communities had emergedin the past, they did not include the most powerful anddeveloped states in the international system. For Jervis,the current security community constitutes “proof by exis-tence of the possibility of uncoerced peace without centralauthority,”95 and thus requires scholars and policymakersto adjust their theoretical assumptions about states’ per-ceptions, interests, and behavior.

| |�

�

�

June 2010 | Vol. 8/No. 2 421

For this purpose, Jervis notes the strengths and limita-tions of theories embedded in the constructivist, liberal,and realist traditions. These include: constructivist theo-ries emphasizing the norm of nonviolence and an emer-gent identity shared by capitalist democracies; neoliberaltheories stressing the pacifying effects of democratic poli-tics, economic interdependence, and joint membership ininternational organizations; and realist theories focusingon the presence of external threat, American hegemony,and the logic of nuclear deterrence. Noting that none ofthese theories can independently explain the emergenceor dynamics of the new security community, Jervis pro-ceeds to adopt an eclectic analytic framework that refor-mulates and combines several causal factors: the beliefthat territorial conquest is difficult and unnecessary; therecognition of the costs of war, particularly in a nuclearage; and, rooted in the spread of democracy, shifts in iden-tity that reflect a sharp decline in militarism and nation-alism as well as a growing compatibility in values amongthe most advanced major powers. Interestingly, the signif-icance of these factors and the complex manner in whichthey interact depends on ongoing historical processes. Forexample, the evolution of the international economy hasbeen marked by a disassociation between territoriality andnational prosperity, which has increased the costs of terri-torial acquisition in relation to potential material benefits.Similarly, the high degree of cooperation among the mem-bers of the security community is in part a function ofenduring legacies of the Cold War when these states, asmembers of a common alliance, were socialized to behaveas “partners” and set aside conflict as a means to settletheir grievances vis-à-vis one another.

Significantly, Jervis does not treat his analysis as a purelyacademic exercise. He is also concerned about the practicalimplications of his eclectic analysis, specifically in relationto American foreign policy and the responses of other mem-bers of the security community. Jervis argues that, as a resultof the Bush doctrine, “[w]e are headed for a difficult world,one that is not likely to fit any of our ideologies or simpletheories.”96 While pessimistic, the latter prediction is notsimply a polemical statement. It derives from Jervis’s recon-sideration of the contours of the present international envi-ronment, which requires an urgent updating of conceptionsof national interest and of the present course of Americanpolicy. In particular, Jervis cautions that unilateral actionsby the US since 9/11 have begun to undermine the trust ofmembers of the security community who are increasinglyconcerned about American hegemony. However, he alsonotes that other members of the security community, to theextent that they wish to check US hegemony, are adoptingnew styles of balancing that involve subtle, coordinatedefforts to socialize and entrap the US to keep its behavior“within acceptable bounds.”97 Whether or not one con-curs with Jervis’ implied prescriptions, his analysis enablesa more open-ended discussion among scholars and policy-

makers about the foreign policy implications of the multi-ple dimensions of a new, evolving international order.

Michael Barnett and Martha Finnemore note in Rulesfor the World that, until recently, even those who paidattention to international organizations (IOs) framed theirdebates in terms of the capabilities and interests of mem-ber states.98 For Barnett and Finnemore, IOs are neitherpassive instruments of states nor mere facilitators of coop-eration. They are Weberian bureaucracies and, as such,possess a bureaucratic culture that often spurs them tobehave in ways that are not sanctioned by their membersand that are clearly unanticipated by their founders. Thisbehavior is not adequately problematized in the conven-tional literature. This is why Barnett and Finnemore launchtheir own efforts to analyze IOs in terms of their auton-omy, power, dysfunctionality, and extent of change.

The eclectic character of Barnett and Finnemore’sapproach is evident in their effort to layer a bureaucratic-centered perspective over conventional state-centeredaccounts in order to generate a more complex analysis ofIO behavior. They recognize that the power and interestsof states are important and that IOs are rarely able tocompel powerful states to act against their interests. Inmany circumstances, however, IOs act independently from(even if not in opposition to) states, generating unantici-pated effects and sometimes shaping state preferences bytaking the lead in agenda-setting. Barnett and Finnemoremake another eclectic move by linking regulative mecha-nisms stressed in rationalist theory to constitutive mech-anisms stressed in constructivist theory. Building on Weber’sconception of bureaucratic power as “control based onknowledge,”99 Barnett and Finnemore suggest that IOslearn to employ their authority, knowledge, and rules bothto regulate the world and to constitute a world that appearsto require further regulation. By creating categories, affix-ing meanings to these categories, and diffusing new normsand rules to guide political practice, IOs are in a positionto create the political world in which they operate and inwhich states form their preferences.

To illustrate the distinctive utility of their approach,Barnett and Finnemore analyze and compare the Inter-national Monetary Fund (IMF), the Office of the UnitedNations High Commissioner for Refugees (UNHCR), andthe UN Secretariat. The three cases cover substantivelydiverse issue areas, each of them associated with well-developed statist explanations that can be used as foils forBarnett and Finnemore’s more eclectic approach. The threeIOs also feature different types of authority claims, withthe IMF relying more on expert authority, the UN Secre-tariat more on moral authority, and the UNHCR on amore evenly balanced mixture of expert and moral author-ity. By demonstrating common patterns across such diversecases—for example, the tendency of IOs to steadilyexpand—Barnett and Finnemore are able to provide atentative defense of their argument.

| |�

�

�



As with Jervis’ book, Barnett and Finnemore’s eclecticapproach has practical implications for the world of policyand practice. In fact, Barnett, who has worked at the UnitedNations, and Finnemore, who has done research on theWorld Bank, note that their academic training and expe-rience had not prepared them for “what international orga-nizations were really like.”100 Their analysis of how IOsgrow in size and take on broader missions sounds a cau-tionary note about the power of IOs to reconfigure inter-national and domestic social spaces, resulting in benefitsfor some and domination for others. Barnett andFinnemore also fear that the procedural legitimation ofIO rules and the substantive legitimation attached tosocially liberal values can reduce the scope for democraticparticipation and accountability as IOs grow in reach andpower. These concerns prompt Barnett and Finnemore topoint to the dangers of an “undemocratic liberalism” thatcan emerge within even the most well-intentioned IOs.

Leonard Seabrooke’s book, The Social Sources of Finan-cial Power, weaves together concepts and mechanisms drawnfrom research traditions in international political econ-omy, comparative politics, the new institutionalism, andeconomic sociology.101 Seabrooke sees the standard liter-ature on financial power as excessively focused on the inter-actions among global economic forces, state economicpolicies, and big business. Little attention is paid to theeveryday economic struggles of ordinary actors, especiallythose in lower-income groups, and to the ways in whichtheir norms and practices can affect a state’s financial capac-ity in the international arena. Seabrooke defines his prob-lem in a manner that does not neatly conform to theparameters of typical approaches to state-society relationsor international political economy; instead, his study self-consciously aims “to examine the mechanisms that linkstate and social groups in broadening or narrowing a state’ssocial source of financial power.”102

For this purpose, theories emphasizing either factorendowments or institutional logics are static; they recog-nize the opportunities generated when the circulation ofcredit is increased but fail to consider social responses tothe manner in which this credit is allocated among differ-ent groups. Theories focusing on the constraints posed byfinancial globalization and different state structures offerpartial explanations for variation in financial systems andpolicies, but they cannot explain why similar states mayhave quite different capabilities in projecting financialpower and shaping the international financial order. Con-structivists consider social norms and identities, but onlya few studies consider the social foundations of financialinstitutions, and even these tend to limit their focus toelite actors or ideational entrepreneurs when discussingthe locus of ideas and norms that pertain to financial andfiscal institutions.

Seabrooke employs the Weberian concept of legitimacyto build a complex analytic framework that incorporates

mechanisms from multiple traditions in multiple fields.Specifically, he argues that the extent to which lower-income groups view patterns of credit access, propertyownership, and tax burdens as fundamentally fair ulti-mately affects state influence in international finance inways that conventional treatments of political economycannot appreciate or explain. Especially important in thisregard is the extent of positive state intervention on behalfof lower-income groups—the lowering of tax burdens andthe expansion of access to credit and property—whichserves to broaden and deepen the domestic pool of capitaland subsequently to boost the international financial capac-ity of states. Historical comparisons of England and Ger-many (at the end of the nineteenth century) and the UnitedStates and Japan (at the end of the twentieth) provideempirical support for this argument. Whereas England’sstrong position was eventually weakened by challengesfrom lower-income groups responding to negative statepolicies, a similar form of contestation in the United Statesgave rise to positive state policies after the mid-1980s thatincreased the legitimacy of national financial policies andenabling the United States to extend its hegemony in theinternational financial order.

Seabrooke concludes: “if a state intervenes positively tolegitimate its financial reform nexus for lower-incomegroupings, it can provide a sustainable basis from whichto increase its international financial capacity.”103 Whilethis may not be the norm, particularly in rentier states,the analysis suggests that the legitimacy and financial capac-ity of states in the international arena are directly andsignificantly affected by the fundamentally domestic choicesstate elites make with regard to lower income groups.Seabrooke’s eclectic analysis not only integrates mecha-nisms drawn from diverse research traditions to link domes-tic social policy and international finance but also providesa useful bridge between academic discussions of politicaleconomy and those involved in social policy, particularlyin democratic settings, where lower-income groups haveregular institutionalized opportunities to punish state elites.

Phineas Baxandall’s Constructing Unemployment alsoaims to trace the operation of multiple mechanisms indifferent domains of social reality.104 Rather than takingthe concept and measurement of unemployment forgranted, Baxandall seeks to historicize its meaning andpolitical significance. Most students of political economyfocus on what kinds of policies, institutional structures,and economic conditions might account for higher orlower unemployment, and most students of labor assumeincreases in unemployment everywhere to be a trigger forprotest and contestation. Instead, Baxandall tackles themore complex problem of explaining how popular expec-tations about appropriate levels of unemployment areinfluenced by elites and policies, and how such expecta-tions then produce different reactions among differentsocial groups to similar increases in unemployment levels.

| |�

�

�

June 2010 | Vol. 8/No. 2 423

As Baxandall puts it, the “threat of unemployment appearsto constrain governments in some countries far morethan others; and the same unemployment rate can arousewildly different degrees of concern among the citizens ofdifferent countries.”105

Focusing on Hungary, Baxandall describes the emer-gence of an “unemployment taboo” in the post-Stalin era,as the Soviet Union and other Eastern Bloc countries beganto offer guarantees of full employment as a basis for a new“social contract.”This taboo survived under Hungary’s NewEconomic Mechanism and made it difficult for factories toshed core workers. But during the last decade of commu-nism, the Hungarian government sought to partially shiftits responsibility for full employment to an informal econ-omy in which individuals could hold second or third jobs.Hungarian communist leaders even allowed large-scale dis-missals for many previously protected categories of employ-ees while developing social policies intended to deal withthe “unemployed.” The postcommunist Hungarian gov-ernment thus had a relatively easy time in linking employ-ment levels to the growth of entrepreneurship rather thanto state policies, with surprisingly little social protest in theface of unemployment rates that skyrocketed to over 20 per-cent by 1993. In Poland, by contrast, the continued com-mitment to full employment through the official state sectormeant that the unemployment taboo survived until the veryend of the communist era, making rising unemployment asignificantly more contentious issue in the postcommunistera. Baxandall also considers responses to unemploymentinadvanced industrial economies todemonstrate thebroaderapplicability of his analytic framework.

Baxandall’s case studies and comparisons suggest thatthe political salience of unemployment depends largely onthree factors: existing ideas about what counts as successregarding unemployment; the sharpness of institutional-ized distinctions between being employed and beingunemployed; and the extent to which core workers areaffected by unemployment. But these factors in turn dependon broader institutional, socioeconomic, and ideationalmechanisms, including political calculations among elitesin certain political settings, the effects of policy discourseson different segments of the population, and historicallegacies related to the treatment of unemployment underprevious regimes. Only by considering the interplay ofthese various mechanisms is it possible for Baxandall toexplain the significant variation in popular responses tounemployment across similar groups of states and overdifferent time periods.

Finally, a more self-conscious deployment of analyticeclecticism is evident in Rudra Sil’s analysis of institu-tional borrowing among late-industrializers.106 The prob-lem Sil tackles concerns the prospects for legitimizinginstitutions designed on the basis of models originallydevised abroad while enlisting the cooperation of subordi-nates whose expectations and behaviors are conditioned

by more familiar local norms and practices. Sil argues that,in the context of belated industrialization, this challengeis exacerbated by the compressed timeframe for the pro-mulgation of new institutions. In distinguishing the rangeof approaches to institutional borrowing, Sil emphasizesthe extent to which elites count on the ability of novelinstitutions to incrementally transform the values, habits,and practices of recruited subordinates, and on the strengthof narratives and symbols that tout either a seamless con-tinuity or radical break with the past.

This problem is approached through a systematic “pairedcomparison”107 of the large-scale industrial firms in twen-tieth century Japan and Soviet Russia. The comparisonreveals that pre-World War II elites in both countries, inspite of their radically different worldviews and normativecommitments, systematically adopted core elements of theTaylorist-Fordist model originally devised in Americanindustry. This proved to be more problematic than expectedin light of contradictions between, on the one hand, pro-duction practices focused on individual skills and perfor-mance and, on the other hand, official and populardiscourses stressing a supposedly revolutionary ethos ofegalitarian collectivism (in Soviet Russia) or a more tradi-tionalist narrative of paternalistic benevolence (in Japan).These tensions, magnified by workers’ economic anxi-eties, contributed to a crisis of legitimacy manifested insurprising instances of labor protest as well as everydayforms of alienation and resistance. In the course of post-war reconstruction, however, Japanese (but not Soviet)elites engaged in coordinated national efforts to creativelyadjust production and employment practices in the inter-est of greater labor commitment and productivity. In theprocess, there emerged significantly greater congruencebetween standardized company practices (including “life-time” employment and small-group work organization)and a reformulated ideology of company paternalism thatincorporated norms and ideals informally articulated byworkers and unions. This congruence provides a partialexplanation for the increase in the legitimacy of employ-ment practices in large-scale Japanese firms, at least rela-tive to pre-war Japan and to Soviet Russia.

Sil’s eclectic framework thus highlights the interactionof several causal mechanisms drawn from a wide range ofresearch traditions. These include the competitive pres-sures of catch-up industrialization linked to the motiva-tions of ambitious state elites and the transnational diffusionof institutional models. The long-term legitimacy of theemergent institution, however, depends heavily on thecoherence between ideology and organization and onthe extent to which both are congruent with the norms,practices, and social relations of subordinates recruitedfrom a given social environment. The effects of these twodimensions of institutional congruence on subordinatesare mediated by cognitive mechanisms, identified on thebasis of research in occupational psychology, that link the

| |�

�

�



familiarity of organizational environments to the cooper-ation and commitment of members. Finally, there areseveral contextually delimited mechanisms linked to theparticular historical legacies and developmental objectivesof specific countries insofar as these reinforce, diffuse, ordeflect the effects of the more general mechanisms. Thiscomplex causal story does not lend itself to a general model,but it surfaces some of the less obvious challenges of insti-tutional borrowing and lays out the practical utility of asyncretist approach in managing the global diffusion ofideas and practices. Such an approach, approximated bythe post-war Japanese case, is one that features (i) thecreative modification of borrowed institutional modelsalongside (ii) an abstract reconfiguration of preexistinglocal norms, values, and practices with the result that (iii)a novel institutional form emerges that is both capable ofrealizing standard developmental objectives and familiarenough to permit meaningful involvement in specific localcontexts.

We discuss these examples of eclectic scholarship forthe purposes of illustration only. Our hope is simply tohave established here, in no more than a preliminary man-ner, the distinctiveness and potential utility of analyticeclecticism in fields in which there already exist estab-lished research traditions. The works discussed above takeon socially important problems formulated so as to bypassor transgress the theoretical boundaries established by spe-cific research traditions. Moreover, each of the works makesa conscious effort to transcend restrictive assumptions aboutwhich aspects of social reality are more fundamental, choos-ing to develop more complex causal stories featuring theinterplay of a wide range of mechanisms. And each fea-tures a clear, if implicit, pragmatic engagement with sub-stantive dilemmas that extend beyond the academe intothe world of policy and practice.

Conclusion: The Challenges andPayoffs of Analytic EclecticismEclectic approaches are not without costs and risks. Wehave already considered above the potential dangers oftheoretical incoherence linked to the possible incommen-surability of research traditions. These dangers, we realize,are serious. But, as we noted above, they are also ubiqui-tous in political science even when practiced more con-ventionally. Incommensurability can, in principle, existacross theories within research traditions as well as acrossapplications of the same theory in different contexts.More importantly, as we noted above, the problem is notentirely insurmountable, at least in the context of prag-matist inquiry. Although caution is required, there arepossibilities for the intersubjective translation of specifictheoretical constructs once these are detached from themetaphysical principles or epistemic commitments asso-ciated with contending research traditions.

There remains the problem of how eclectic scholars candemonstrate the quality and utility of their work to thoseworking in diverse research traditions. By its very defini-tion, eclectic scholarship lacks a Lakatosian “protectivebelt” that can shield substantive analyses from questionsabout core premises and assumptions. It also lacks thekinds of epistemic norms and uniform standards that enableresearch traditions to evaluate individual contributions andproclaim some degree of internal progress. Equally prob-lematic is the fact that an eclectic approach is likely todraw a wider range of criticism informed by the variedstandards and practices of varied research traditions.

This does not justify forgoing eclectic inquiry, how-ever. Rather, it puts the onus on eclectic researchers todemonstrate their attentiveness to standards and expecta-tions associated with different research traditions. Whilethe particular criteria employed by any one research tra-dition may not be appropriate for evaluating explicitlyeclectic approaches, eclectic scholars do need to be clearabout their own evidentiary standards

Analytic Eclecticism in the Study of World Politics ......Rudra Sil ([email protected]) is...

Documents

Transcript of Analytic Eclecticism in the Study of World Politics ......Rudra Sil ([email protected]) is...