APSA-CP Winter 1997 · 2018. 12. 19. · APSA-CP Newsletter 2 Winter 1997 tradition when those questions, issues, and problems are closely related, be-cause they have been chosen

APSA-CP Newsletter 1 Winter 1997

Comparative politics is a field insearch of theory. It should be insearch of a framework or orientation.

In my first job, I was the solecomparativist in a small faculty, allof whose members focused on theUnited States. In striking contrastwith my field, I discovered, theirspossessed a consensus as to whatconstituted meaningful research or animportant research finding. The rea-son was simple: they had democratictheory, while we did not.

Given their shared vision of poli-tics, the Americanists could engagein normal science. It became mean-ingful, for example, to seek moreprecise measurement of, say, incum-bency advantage: more decimalplaces and reduced errors obviouslymattered. For everyone would recog-nize the implications of the resultsfor the responsiveness of politicalleaders to those they governed.

The Americanists’ shared orien-tation toward politics – their commit-ment to the democratic paradigm –also informed their response totheory and method. The importanceof elections made investment in re-search into public opinion a “nobrainer,” as our youngsters might

say; survey research offered an obvi-ous means to mount scientific investi-gations into the behavior of citizens indemocracies. Americanists more rap-idly recognized the significance of for-mal theory as well; they were better po-sitioned than the rest of us to appreci-ate the political significance of seeminglyabstruse theorems regarding cyclicity insocial choice and the possibility of equi-libria under majority rule.

In comparative politics, we havelacked a similar intellectual frame-work. Indeed, to us, theAmericanists, with their shared vi-sion, appeared provincial: they actedas if democratic politics were theonly form of politics. For our part,our efforts to compensate for a lackof similar structure made us at timesappear ridiculous. Misperceiving thefoundations for the Americanists’ability to build a cumulative researchtradition, we sought to build ours onthin air. Thus our wholesale conver-sion to “systems theory,” “structural-functionalism,” the study of “thestate,” and – dare I say it – rationalchoice. But theory and method mustbe brought to bear on questions, is-sues, and problems; the research tra-dition becomes a cumulative research

APSA-CPAPSA-CPAPSA-CPAPSA-CPAPSA-CPNewsletter of the APSA Organized Section in Comparative Politics

Volume 8, Number 1 Winter 1997

Letter from the PresidentTheory in Comparative Politics?

Robert H. Bates, Harvard [email protected]

ContentsContentsContentsContentsContentsNews and NotesNews and NotesNews and NotesNews and NotesNews and Notes 3

Letter from the EditorLetter from the EditorLetter from the EditorLetter from the EditorLetter from the Editor 4

Notes from the Annual MeetingsNotes from the Annual MeetingsNotes from the Annual MeetingsNotes from the Annual MeetingsNotes from the Annual MeetingsFFFFFormal Theorormal Theorormal Theorormal Theorormal Theory in Comparativey in Comparativey in Comparativey in Comparativey in ComparativePPPPPoliticsoliticsoliticsoliticsolitics

Barry Weingast 6

Book PreviewsBook PreviewsBook PreviewsBook PreviewsBook PreviewsComparative PComparative PComparative PComparative PComparative Politics: Rolitics: Rolitics: Rolitics: Rolitics: Rationalityationalityationalityationalityationality,,,,,Culture and StructureCulture and StructureCulture and StructureCulture and StructureCulture and Structure

Mark Lichbach & Alan Zuckerman 8Margaret Levi 9Ira Katznelson 9Marc Howard Ross 10Samuel Barnes 11Charles McAdam, Charles

Tilly & Sidney Tarrow 12Peter Hall 13Joel Migdal 14Mark Lichbach 14Alan Zuckerman 15

Paradigms and SandcastlesParadigms and SandcastlesParadigms and SandcastlesParadigms and SandcastlesParadigms and SandcastlesBarbara Geddes 18

Solving the Ecological InferenceSolving the Ecological InferenceSolving the Ecological InferenceSolving the Ecological InferenceSolving the Ecological InferenceProblemProblemProblemProblemProblem

Gary King 21Democracy and DevelopmentDemocracy and DevelopmentDemocracy and DevelopmentDemocracy and DevelopmentDemocracy and Development

Adam Przeworski, Mike Alvarez, JoséAntonio Cheibub and FerrandoLimongi 23

Data Sets and ArchivesData Sets and ArchivesData Sets and ArchivesData Sets and ArchivesData Sets and ArchivesData, Archiving and the RealData, Archiving and the RealData, Archiving and the RealData, Archiving and the RealData, Archiving and the RealWWWWWorld of Comparative Scholarorld of Comparative Scholarorld of Comparative Scholarorld of Comparative Scholarorld of Comparative Scholar-----shipshipshipshipship

David Laitin 27

Book ReviewsBook ReviewsBook ReviewsBook ReviewsBook ReviewsPrivatizing RussiaPrivatizing RussiaPrivatizing RussiaPrivatizing RussiaPrivatizing Russia 30Challenging the StateChallenging the StateChallenging the StateChallenging the StateChallenging the State 31Internationalization and Domes-Internationalization and Domes-Internationalization and Domes-Internationalization and Domes-Internationalization and Domes-

tic Ptic Ptic Ptic Ptic Politicsoliticsoliticsoliticsolitics 32

Copyright 1997 American Political Science AssociationSubsidized by the College of Letters and Science, UCLA


tradition when those questions, issues,and problems are closely related, be-cause they have been chosen within acoherent and encompassing frame-work.

In the middle income countriesof the world, democratic theory mayprovide that framework; it may ful-fill a role in comparative politicssimilar to the role it played in thestudy of American politics. There areindeed rapidly mounting signs of itdoing so, with the comparative studyof electoral systems taking on newlife in the field and survey researchbeing employed in new and power-ful ways in formerly communist so-cieties.

Viewing development as the po-litical economy of growth may pro-vide criteria of relevance that trans-form research into the developing ar-eas into a cumulative research pro-gram. The accumulation of capital,the process of investment, the searchfor the political foundations for eco-nomic growth: investigated withinthe structure offered by growththeory – classical, Marxian, neo-clas-sical, or modern – these themes yieldresearch programs that offer mutualcommentary and criticism. Withinsuch a framework, “normal science”could grow.

Social theory offers a third intel-lectual framework. Social theoryhighlights the political significanceof culture and the producers of cul-ture: artists, priests, and intellectuals.It also highlights the role of rhetoricand symbolism. It provides a frame-work in which we can see the inten-

tionality that underlies the constructionof values, and the political purposes towhich culture is put. Within this frame-work, we should be able to address notjust the privileging of voices in the Westbut also contemporary appeals to reli-gion, ethnicity, and identity throughoutthe world. While social theory may pro-vide the framework, it will no doubt beothers who provide the theory andmethodology. Social theory is relativelybarren when it comes to showing howits claims can be systematically evalu-ated. Indeed, it may well be the practi-tioners of forms of science that socialtheorists themselves reject who showhow and why the insights of the socialtheorists are correct. Game theorists,for example, will surely be un-welcomebed fellows; but it is they who havebegun to formulate testable models ofthe strategic emission of meaningful ges-tures.

Students of comparative politicsare given to fads; we often seek frommethodology answers that it cannotgive. We are too often given to sci-entific posturing. We seek to becomescientific by acting like scientists,rather than by using the tools of sci-ence to pursue intellectual issues.Methodology cannot provide guid-ance to interesting and importantquestions. Nor can it provide a frame-work supportive of debate, cumula-tion, or self-correction. For guidancein these matters, we must turn to in-tellectual traditions. I have pointedto three. The contributors to this is-sue will point to others.

Section OfficersSection OfficersSection OfficersSection OfficersSection Officers

PresidentPresidentPresidentPresidentPresidentRobert Bates

Harvard [email protected]

Vice-President and President-Vice-President and President-Vice-President and President-Vice-President and President-Vice-President and President-ElectElectElectElectElectDavid Collier

University of California, [email protected]

SecretarSecretarSecretarSecretarSecretar yyyyy-----TTTTTreasurerreasurerreasurerreasurerreasurerJames Caporaso

University of [email protected]

1997 APSA Program Coordinator1997 APSA Program Coordinator1997 APSA Program Coordinator1997 APSA Program Coordinator1997 APSA Program CoordinatorNancy Bermeo

Princeton [email protected]

At-Large Committee MembersAt-Large Committee MembersAt-Large Committee MembersAt-Large Committee MembersAt-Large Committee MembersCatherine Boone

University of Texas, [email protected]

Robert KaufmanRutgers [email protected]

Edmond KellerUniversity of California, Los [email protected]

Frances RosenbluthYale [email protected]

Gabor TokaCentral European [email protected]

The Newsletter is published twice yearly.

Deadline for Winter issue: December 15Deadline for Summer issue: June 15


Newsletter Staff:Newsletter Staff:Newsletter Staff:Newsletter Staff:Newsletter Staff:

EditorEditorEditorEditorEditorMiriam Golden

University of California, Los [email protected]

Assistant EditorAssistant EditorAssistant EditorAssistant EditorAssistant EditorDavid Yamanishi

University of California, Los [email protected]

Regional Editors-at-Large:Regional Editors-at-Large:Regional Editors-at-Large:Regional Editors-at-Large:Regional Editors-at-Large:

Soviet Successor StatesSoviet Successor StatesSoviet Successor StatesSoviet Successor StatesSoviet Successor StatesRichard Anderson

University of California, Los Angeles

Middle EastMiddle EastMiddle EastMiddle EastMiddle EastLeonard Binder


Latin AmericaLatin AmericaLatin AmericaLatin AmericaLatin AmericaBarbara Geddes


AfricaAfricaAfricaAfricaAfricaEdmond Keller


WWWWWestern Europeestern Europeestern Europeestern Europeestern EuropeRon Rogowski


Eastern EuropeEastern EuropeEastern EuropeEastern EuropeEastern EuropeIvan Szelenyi


FFFFFormal Analysis and Methodologyormal Analysis and Methodologyormal Analysis and Methodologyormal Analysis and Methodologyormal Analysis and MethodologyJohn Londregan

University of California, Los AngelesHoover Institution

The Newsletter will be available on the Internet for viewing and downloading.

Beginning March 1, visit us at http://shelley.sscnet.ucla.edu/apsacp/index.html.

Send technical inquiries to the Assistant Editor.

The Nominating Committee, consistingof Valerie Bunce (Chair), Barry Ames,and Jeff Herbst, chose Susan Shirk, ofthe University of California at San Di-ego, and Ronald Herring, of CornellUniversity, as new members of the Ex-ecutive Committee.

In consultation with the Executive Com-mittee, Robert Bates appointed a newnominating committee, consisting ofKaren Remmer of the University of NewMexico (Chair), Geoffrey Garrett(Stanford and the University of Penn-sylvania), Torben Iversen (Harvard),Susan Stokes (Chicago), and KirenChaudhry (Berkeley). The committeewill nominate two new members of theExecutive Committee and select thePresident to succeed David Collier.

The Luebbert Award committee was thisyear chaired by James Alt and includedRuth Collier and Barry Weingast. Therecipients of the award for the best bookwere Stephan Haggard and RobertKaufman for The Political Economy ofDemocratic Transitions (Princeton:Princeton University Press, 1995).

The runners up for the book award wereBarbara Geddes for Politician's Di-lemma: Building State Capacity in LatinAmerica (Berkeley: University of Cali-fornia Press, 1994) and Sidney Tarrow,for Power in Movement: Social Move-ments, Collective Action, and Politics(New York: Cambridge UniversityPress, 1994).

(News & Notes continues on page 17.)

In response to spirited discus-sions at the Business Meeting, thePresident of the Section has sent thefollowing letter to Elinor Ostrom,President of the American PoliticalScience Association:

Dear President Ostrom:

As you are aware, the allocation of pan-els remains a contentious issue in the As-sociation. The governing criterion has longbeen attendance at previous panels offeredby the subfield.

Recent records indicate that member-ship in the Comparative Politics Section isincreasing, and that the subfield remainsthe largest within the Association. As aconsequence, we feel highly constrainedby the Program, in which we are allocatedmany fewer panels than our membershipcan productively fill. Indeed, we are ableto fulfill but 8% of our requests to offerpanels.

The result is that, in our field, our con-vention, that is supposed to be inclusive, isin fact more exclusive than our profes-sional journal.

This argument was vigorously advancedby the membership in the Section’s busi-ness committee, who urged that the prob-lem be brought to the attention of the As-sociation. I would be most grateful if youcould take it up with the Program Com-mittee, perhaps in consultation with KarenRemmer, who served as the Program Chairfor the Section at this year’s meeting.

Robert H. Bates

News & Notes


It has been five years since theAPSA-CP Newsletter’s editor has com-municated directly and formally withreaders, except when my predecessor,Ron Rogowski, did so three years agoin his dual capacity of Section Presidentand Newsletter Editor. Given recentchanges in policy and practices, nowseemed a timely moment to do so again.

To recount for our newer members:this Newsletter was founded, initially asan annual publication, with the estab-lishment of the Organized Section inComparative Politics in 1990. TheNewsletter moved from the Universityof Washington to UCLA in 1993, at thesame time becoming a semiannual pub-lication. Funded in part by members’dues, production of the Newsletter de-pends crucially on a local subsidy. Weare currently in the first of a three-yearcommitment of funding on the part ofUCLA’s College of Letters and Science,for which we are highly grateful. For theprior three years, we received fundsfrom UCLA’s International and Over-seas Programs, which we also gratefullyacknowledge. UCLA’s funding coversthe costs of hiring an Assistant Editorto produce the Newsletter; the portionof Section dues that reverts to the Sec-tion pays for printing and mailing.

We currently have 1,550 members,making the Organized Section in Com-parative Politics twice as large as theAPSA’s second largest organized sec-tion. Not only are we an extremely largeSection, we are intellectually highly di-verse. With the exception of the Ameri-can Political Science Review, this News-letter is perhaps the most widely readpublication among students of compara-tive politics in the English-speakingworld. As the practice of comparativepolitics becomes increasingly frag-mented and intellectually balkanized,the need for a central forum for debatehas become that much greater. The maingoal of the APSA-CP is to contribute to

the intellectual cohesiveness of the com-parative field. By this I don’t mean thatwe expect to impose an artificial degreeof intellectual uniformity on our dispar-ate readers, but only that the Newsletteraims to serve as a site for debate of thecentral issues animating our corner ofthe discipline. At the same time, we alsoseek to inform our readers about trendsand developments in scholarship, to pro-vide a place for leading figures in thecomparative field to speak and to en-gage issues (as well as each other), andto alert our readers to scholarly re-sources of which they otherwise mightremain unaware.

Under the earlier editorial leader-ship of my colleague, Ron Rogowski,policy was adopted to focus each issueof the Newsletter on a single centraltopic. Ron’s goal was to make the News-letter a place for substantive debate, notmerely a site for professional announce-ments. I have followed his lead. My ownaim has become that of focusing eachissue on some very general, perhapseven meta-theoretical, topic. There aremany places for us to present the resultsof our research, but far fewer where wecan step back and consider how our ownresearch speaks to the comparative fieldas a whole, how it intersects with otherkinds of research or other approacheswithin the comparative field, and thetrends and developments characterizingour subfield more generally. The News-letter aims to be the place where we canreflect on the state of comparative poli-tics. Of course, there’s no substitute forthe real thing, and in my own view, do-ing comparative politics is in the endmore interesting and certainly morechallenging than thinking and talkingabout what we do, but the occasionalself-conscious foray onto meta-theoreti-cal territory is nonetheless a useful de-tour. In particular, I hope that some self-reflection on the part of practitioners isuseful to younger scholars who may still

be looking to define their place in the fieldand the discipline. Suggestions for themetopics for future issues are always wel-come.

Let me now turn to matters of prac-tical concern to readers, and providesome information regarding policies andprocedures. First, the APSA maintainsour membership lists and we receive ourmailing labels directly from the Asso-ciation for each issue. If you move, yourchange of address will be handledthrough the Association as part of yourmore general change of address as amember. There is thus no reason to in-form the Newsletter directly of anychange of address. Similarly, when youjoin the Section, there may be some lagin getting you on the rolls, but again,you should communicate any problemsdirectly to the APSA. At the APSA,Sheilah Mann ([email protected]) andJun Yin ([email protected]) are the per-sons responsible for Organized Sections.

Second, in an effort to impose a con-sistency of tone to the Newsletter, I haveenacted a five-footnote policy. Con-tributors, in other words, are limited tono more than five footnotes in any ar-ticle. The Newsletter is a professionalbut not a scholarly publication. We re-port on research trends but we do notreport a great deal of original researchitself. Hence, it seemed appropriate tolimit footnotes. In instances where thelimitation is unrealistic, we will provideinformation allowing readers to commu-nicate directly with the author in orderto obtain references. Underlying thispolicy, of course, have been seriousspace limitations.

Third, we have had serious spacelimitations, which, as the current issuetestifies, have now been substantially re-laxed. Until now, with Section dues asthey are, printing and mailing costs al-lowed us to publish issues generally of16 and occasionally of 20 pages. Ournew Assistant Editor, David Yamanishi,

Letter from the EditorMiriam Golden, University of California, Los Angeles

[email protected]


has used his background in the graphicsindustry to achieve a major reduction inproduction costs. Not only is this issueof the Newsletter offset rather than pho-tocopied, but we have consolidatedprinting, labelling and shipping fromone location rather than dividing thework between our local copy shop andthe UCLA mail room. This has two po-tential benefits. First, we hope it will de-crease the technical errors in productionthat we have encountered (frequent mis-prints, extremely slow mailing) and thathave justifiably so annoyed readers. Sec-ond, it has so dramatically reduced ourcosts that we have been able to doublethe size of the Newsletter, increasing itfrom 16 to 32 pages. The only poten-tially negative side-effect is that we arenow printing on lower weight paper forforeign subscribers, in order to keep theextremely high costs of foreign postageunder control. If this results in damagedNewsletters arriving overseas, we defi-nitely want to know.

Our new enlarged format shouldtransform the Newsletter into an evenmore serious and substantive forum fordebate. With 32 pages per issue, we canbe much more flexible in the page limi-tations we establish for individual con-tributors. In addition, we can substan-tially increase the range we cover invarious debates, requesting many morepersons to contribute. Otherwise, how-ever, while we welcome brief announce-ments of matters of interest to our read-ers, I cannot promise that we will haveadequate space to print them. I considerthe thematic articles and the book re-views more important, and we fit an-nouncements in only as we can. We areespecially unlikely to have room forannouncements that contain alreadywell-publicized information.

Fourth, we welcome unsolicited re-search reports on the construction of da-tabases or archives accessible and use-ful to others. If you are assembling a dataset that will shortly become publiclyavailable, a summary description wouldbe welcome. Research and data reportsmay be submitted at any time directlyto the Editor or Assistant Editor, pref-

erably by email, using whatever elec-tronic format to which you are accus-tomed. If we have to postpone publica-tion because of reasons of space, we willlet you know.

Fifth, our policy for book reviewshas evolved over time, and is now en-tering a new phase. Our book reviewsare written by graduate students, whoselect the books to review themselves,according to their own interests. Thegreat advantage of this policy is that stu-dents pick books that they find espe-cially helpful or inspirational in theirresearch. When the Newsletter was firstmoved to UCLA, largely as a matter ofconvenience we solicited reviews onlyfrom our own graduate students. Gradu-ally, we began soliciting them from stu-dents working with officers of the Or-ganized Section teaching at other insti-tutions, so now our book reviews aredrawn from universities across the coun-try (although a disproportionate numberprobably still come from UCLA gradu-ate students). We are now a stableenough body that I have decided to openthe book review section to graduate stu-dents generally. Please encourage yourstudents to submit proposals. In the firstinstance, this should consist simply ofan inquiry regarding a particular title,which should be sent to myself and theAssistant Editor (preferably by email).We will let the student know whethersomeone is already reviewing the bookproposed, whether we think we will havespace in the next issue and when to getthe review in. When the review arrives(again, we urge electronic submission),we may require the author to edit it, ei-ther for length or for other consider-ations. (We will handle minor copy-ed-iting ourselves.) In extreme cases, wemay reject a review which is somehowwildly inappropriate, but I certainlyhope this will occur only in extraordi-nary circumstances. We suggest thatbook reviews run 1,000 to 1,500 wordsin length.

Increasingly, I receive books frompresses with requests to review them inthe Newsletter. Since students selecttheir own books to review, there is no

mechanism for handling unsolicited re-view materials. If you publish a book,you would be better advised to look fora graduate student to review it (I am sureit goes without saying that the studentshould be at an institution other thanyour own!) than to have a copy sent tome for review purposes. Likewise, a stu-dent who wishes to review a book shouldplan on procuring his or her own copy,since we lack the staff resources to so-licit copies of books from publishers.

Finally, I encourage you to usewhatever portions of the Newsletter arerelevant in the classroom. The APSAhas recently agreed to allow reproduc-tion of articles in the Newsletters pro-duced by Organized Sections for peda-gogical use without requiring any fee.(The same is not the case for multiplesgreater than 10 of articles taken fromeither the American Political ScienceReview or Political Science & Politics.)Your graduate students are our futuremembers. We hope you will share yourNewsletter with them, and help profes-sionalize them by encouraging them tosubmit book review proposals. And forour colleagues at institutions which donot grant graduate degrees, we hope theNewsletter will help keep you abreastof debates and publications in the field.

A last word to our out-going Assis-tant Editor, Terri Givens, who has donean outstanding job of laying out theNewsletter, getting it reproduced andmailed, and generally staying on top ofthe administrative details over the pastfew years. Under Terri’s care, the lay-out of the Newsletter was improved,making it easier and more pleasant toread. She has also carried a huge admin-istrative burden invisible to the readerbut extremely visible to the Editor. Nofaculty member could take on a News-letter like this without a large amountof assistance, and Terri has done a su-perb job. While I am sorry to see hergo, she will have more time to devote toher own research without the responsi-bilities of the Newsletter. And I ampleased to welcome our new AssistantEditor, David Yamanishi, in her place.


Formal Theory andComparative PoliticsBarry R. Weingast

Stanford [email protected]

A remarkable event occurred twosummers ago when Tim Fedderson andRoger Myerson ran a conference atNorthwestern University on the formaltheory of political institutions. This wasa large conference, involving over sev-enty people. Although the official em-phasis of the conference was on institu-tions, a large portion of the participantsspoke about their work in comparativepolitics. Each participant had previouslyunderstood that his or her work includeda comparative dimension. What few hadunderstood was that nearly all of us wereworking in comparative politics. Inshort, comparative politics has becomethe program among formal theorists.

This seems to suggest that formaltheory has a bright future in compara-tive politics. Other signs point in thesame direction. Formal theory has longhad a presence in comparative politics,and many comparative scholars havecome to draw on elements of rationalchoice. Importantly, within every areaand topic of comparative politics, I havefound comparativists willing to engagein a dialogue with formal theorists.

Nonetheless, formal theory’s future incomparative remains uncertain. All isnot sweetness and light. At the abstractlevel, the reigning way of understand-ing the interaction of formal theory andtraditional approaches is captured bytwo metaphors, Gabriel Almond’s sepa-rate tables and Thomas Kuhn’s compet-ing paradigms. These metaphors suggestthat formal theory and traditional meth-ods are incompatible and competingapproaches. They point scholars toward

confrontation and the need to choosesides.

Extremists of both stripes reinforcethe view suggested by these metaphors.Some rational choice proponents seemto believe that “we’re science, you arenot, and that we do it you don’t.” Andmany traditional scholars appear to re-ject the possibility of a social science ofcomparative politics. These extreme,“get lost” attitudes are hardly designedto elicit sympathetic cooperation.

A New MetaphorI believe that the metaphors of com-

peting paradigms and separate tables arenot useful ways to think about the inter-action of formal theory and compara-tive politics. I propose instead anothermetaphor, that formal and traditionalapproaches are complementary ratherthan competing paradigms. Each ap-proach has what economists call a “com-parative advantage,” something it doeswell relative to the other. Each hassomething to bring to the same table,thus providing the basis for cooperationrather than competition and controversy.

A potentially useful model for coop-eration between formal and traditionalapproaches reflects the way in whichcooperation emerged between formaltheorists and those studying the politi-cal culture and behavior in Congress inthe 1970’s and 80’s. Comparativists arelikely to be skeptical that anything fromAmerican politics might be useful forcomparative politics. Nonetheless, asformal approaches emerged in Ameri-can politics, the seeds of the same con-flict were present. Traditional scholarsdominated the study of Congress. For-mal methods were greeted with consid-erable skepticism, in part because theyabstracted from the details that provedcentral to the traditionalists’ approach.

Traditional scholars, emphasizingpolitical behavior and culture, focused

on careful observation of congressionalpractice, such as congressional norms.They provided detailed descriptions ofthe norms as embedded in particularcontexts and reflecting particular mean-ings for individuals. Formal theoriststended to take these findings as given,seeking to provide explanations forthem; for example, by providing a modelof why a particular norm could emergeas an equilibrium in a game among leg-islators.

In this way, the two approaches inter-acted positively; each provided insightsfor the other. Both studied the same sub-ject, but from different perspectives.And scholars from both camps came tolearn from and enjoy each other’s work.Thus, the two approaches proved not tobe competing in Kuhn’s sense, butcomplementary ones emphasizing dif-ferent aspects of the same subject.

The same type of complementarity ap-pears possible in comparative politics.Consider, for example, the burgeoningliterature on ethnic politics. One obvi-ous lesson from the literature is the in-credible variety of ways in which eth-nic conflict arises. Formal theory is un-likely to predict the specific circum-stances of particular conflicts; tradi-tional approaches have a comparativeadvantage in the study of particular eth-nic conflicts.

Formal theory’s potential contributionresides in the ability to answer questionsto which traditional methods are lesssuited. For example, how do we explainthe common pattern across different eth-nic conflicts of long periods of peace,punctuated by periods of intense ethnicviolence? The promise of formal ac-counts of ethnic conflict is not in ex-plaining details and meanings, but inunderstanding general mechanisms thatapply to contexts across time and space.

The reaction to Jim Fearon’s paper,“Commitment Problems and the Spread

Notes from the Annual MeetingsThe article below is the first in what we hope will be a continuing series linking the Newsletter and Section panels atthe APSA meetings in a more explicit fashion. Weingast’s reflections are drawn from the paper of the same titlepresented on a panel on “Formal Theory and Comparative Politics” at the 1996 APSA meeting.


of Inter-Ethnic Conflict,” illustrates thepotential cooperation and open-mindedness among both kinds of schol-ars. Although this paper has not become“the way” to understand ethnic conflict,traditional scholars have been willing totake the insights of this approach seri-ously in a range of contexts. And Fearoncould not have produced his idea with-out first embedding himself in the rich,traditional literature.

Cooperation between formal theoristsand traditional comparativists is not in-evitable, however. A central limit to theacceptance of formal approaches incomparative is found in formal theory’straditional focus on the institutions ofrepresentative government in developedsocieties. By emphasizing legislatures,elections, and voting typically studiedin stable democracies with stable insti-tutions formal theory’s traditional focusignores a vast set of important politicalphenomena associated with less devel-oped societies. To mention a few: cyclesof authoritarianism and democracy;problems of failed democracy; absenceof the rule of law, stable property rights,and political rights; dramatic instancesof discontinuous political change, as re-flected by coups and revolutions; andfinally problems of ethnic strife.

To many formal theorists, phenomenaof this sort represent the attraction ofcomparative politics: a vast frontier ofphenomena to be modeled. Of course,there have long been rational choice ef-forts in these areas (such as Bates andPopkin). In the last few years, formaltheorists have begun to study a muchbroader range of comparative questions.Yet it is fair to say that formal theory’spromise to provide a systematic ap-proach to a range of comparative phe-nomena remains unproven. Formaltheory’s traditional emphasis leadsmany comparativists to be skepticalabout whether techniques developed tostudy highly institutionalized societieswill yield equally high payoffs for phe-nomena central to less developed soci-eties. Until formal theory begins to pro-vide serious answers to phenomena out-side the developed societies, rational

skepticism will remain. Put another way,the proof is in the pudding, and formaltheory has only begun to provide it.

Strengths and Weakness of FormalTheory

I end this article by raising anotherconcern about formal theory’s role incomparative politics. Consider two ofthe impressive programs of formal re-search in comparative politics, AdamPrzeworski’s study of democratic sta-bility and Michael Laver and KennethShepsle’s study of government forma-tion in the European democracies.

These programs are emblematic ofrecent formal work in comparative poli-tics. Both programs are rightly regardedas showcase applications of cross- na-tional, formal methods. Both combinenew theory with systematic empiricalanalysis. The texture of phenomena in-formed these scholars’ theory building.Finally, the striking empirical power ofPrzeworski’s and Laver and Shepsle’smodels demonstrates the value and in-sight of formal theory.

Inevitably, these programs’ successinvites the question: will abstract meth-ods become the way to study compara-tive politics, pushing aside other ap-proaches? I raise this question becauseI believe that the success of formal meth-ods in comparative depends on how weanswer it. Here, too, the literature onCongress provides some hints. Just astraditional congressional scholars careto learn about formal methods that de-bate, for example, whether members ofCongress maximize votes or the prob-ability of reelection, so too did manyformal theorists start to immerse them-selves in the close observation of Con-gress. The same pattern ofcomplementarity and cross-fertilizationis possible in comparative. The analyti-cal power of formal methods will begreater if they become integrated withtraditional approaches.

By way of summary, I have arguedthat formal methods and traditional ap-proaches should be viewed as comple-mentary approaches to the same subjectrather than competing ones. They eachhave their place in comparative politics.

Nonetheless, the future of formal theoryremains uncertain. I have suggested thatformal theory’s long term place in com-parative politics depends on two factors:first, how and whether it moves beyondthe institutions of representative democ-racy to provide new insights into theproblems of less developed societies;second, the degree to which formaltheory becomes integrated with existingcomparative approaches.

I’m optimistic about formal theory’sfuture. I believe that formal theory willprovide novel and enlightening ap-proaches to most problems in compara-tive politics, that its theoretical perspec-tive will help integrate a series of dis-parate sub-literatures, and finally that itwill suggest a major reorganization ofthe field. Formal theory, as I’ve argued,holds the potential of fitting alongsidetraditional approaches. Whether it doesso or whether it continues to be seen asalien by many comparativists, I cannotpredict. I can predict, however, that con-frontation rather than cooperation ismore likely if most comparativists andformal theorists continue to treat theirapproaches as separate tables and com-peting paradigms.

References

Almond, Gabriel A. 1990. “SeparateTrials: Schools and Sects in Politi-cal Science,” ch. 1 in A DisciplineDivided: Schools and Sects in Po-litical Science. Newbury Park: SagePublications.

Kuhn, Thomas. 1962. The Structure ofScientific Revolutions. Chicago:University of Chicago Press.

Laver, Michael, and Kenneth A.Shepsle. 1994. Cabinet Ministersand Parliamentary Government.New York: Cambridge UniversityPress.

Laver, Michael, and Kenneth A.Shepsle. 1996. Making and Break-ing Governments: Cabinets and leg-islatures in Parliamentary Democ-racies. New York: Cambridge Uni-versity Press.


authors of the essays in this volume ex-amine the research schools in compara-tive politics, they assess knowledge andadvance theory, seeking to direct re-search in the coming years.

We have organized this volumearound the themes of theory and re-search schools in comparative politics.Because so much analysis in compara-tive politics is guided by the expecta-tions, assumptions, methods, and prin-ciples of rational choice theory,culturalist analyses, and structuralistapproaches, assessments of the state oftheory and prospects for advancingtheory need to focus on these researchschools. What explains the imperialistexpansion of these schools and the dis-appearance of other approaches? AsLichbach’s and Zuckerman’s essays inthis volume demonstrate, these schoolsshare an ontological and epistemologi-cal symmetry. They offer – indeed force– choices along the same dimensions.Furthermore, at a more fundamentallevel, the themes of the research schoolsrest at the heart of the human sciences.Reason, rules, and relations are uniqueto social theory. Focusing on thesethemes sets research in the social sci-ences apart from the physical sciences,providing a fundamental basis on whichto theorize about political phenomena.Rationalist, culturalist, and structuralisttheories are embedded in strong researchcommunities, scholarly traditions, andanalytical languages. As they dominate

comparative politics, they provide thelocus for assessments of theory in thisarea of knowledge.

Because research in comparative poli-tics centers around distinctive topics, wehave selected four themes to examinethe interplay between theory and thethree schools: the analysis of mass poli-tics, especially regarding electoral be-havior, and social movements and revo-lutions; political economy; and state-so-ciety relations. Why did we choose thesetopics? Taken together, they encompassmuch of the research done in compara-tive politics. Each displays a history ofsophisticated theoretical and empiricalwork that stretches over several decades.The comparative study of voting behav-ior begins in the inter-war years. Be-cause most people who engage in po-litical activities do so only at the ballotbox, this research examines the politi-cal behavior of the largest set of people;here the study of politics moves its fo-cus away from politicians and bureau-crats, government agencies and politi-cal parties; and the abstractions of stateand society. The systematic analysis ofsocial movements and revolutions de-scends directly from Marx and Weber.It also links to studies of regime trans-formations and the bases of stable de-mocracies. Beginning with Keynes’theories, the analysis of the politicaleconomies of advanced industrial soci-eties has become the focus of the larg-est part of research on the political in-

Comparative Politics: Rationality, Culture and Structureedited by Mark I. Lichbach, University of Colorado, Boulder

[email protected] and Alan S. Zuckerman, Brown University

[email protected]

In Press, Cambridge University Press, 1997

Comparative Politics:Rationality, Culture,and StructureMark I. Lichbach

University of Colorado, [email protected]

Alan S. ZuckermanBrown [email protected]

Theory advances through explicit dis-cussions of strengths and weaknesses,detailing areas of success and failure thatestablish the ability of analyses to ad-vance knowledge in a discipline ofknowledge. Theory is a collective en-terprise. While scholarship is the workof single scholars, knowledge accumu-lates as groups of scholars accept andreject claims about the world. The ex-traordinary range and importance of thetopics examined in comparative politics,the powerful and competing researchschools, and the scholarly pedigree in-vite periodic assessments. In the early1960’s, Harry Eckstein and David Apteredited a collection of essays that estab-lished the field’s questions. In 1970,Robert Holt and John Turner gatheredtogether a set of contributions that raisedthe level of theoretical sophistication.Both volumes guided research in com-parative politics and influenced the con-tests among the research schools. As the

Book PreviewsThe focus of the current issue is on four forthcoming books which we believe will often be used in introductorygraduate seminars in comparative politics. Two – those by Geddes and King – are methodological in orientation, one– the volume edited by Lichbach and Zuckerman – tries to give a sense of the scope and types of work currently donewithin the field, and one – by Przeworski et al. – considers a fundamental substantive problem within comparativepolitics. We intend these book previews to alert readers to some of the most exciting, innovative, and important work


Alan S. Zuckerman, returning the focusto the theme of advancing theory incomparative politics.

A Model, A Methodand a Map: RationalChoice in Compara-tive and HistoricalAnalysisMargaret Levi


Empirical rational choice in compara-tive analysis is in its relative infancy. Itis still in the process of making the tran-sition from analytics to analytic narra-tive. Even so, it has become one of theleading paradigms in the field and pro-duced some major and influential workon a range of subjects, places, and peri-ods. Although the divide between ratio-nalists and many of those in this vol-ume may be shrinking as rationalistsbecome more concerned with contextand non-rationalists recognize collectiveaction problems, voting cycles, andother insights from rational choice, thedivide remains nonetheless. Structural-ists and rationalists are in the same con-versation but persist in very differentviews of the origins and affects of insti-tutions and preferences. Many rational-ists are taking on board the concerns ofculturalists with providing a more com-plete account of preferences and strate-gies, but they continue to disagree onthe uses of those accounts. What dividesrationalists from culturalists and struc-turalists is not method in the sense ofmathematics versus statistics, fieldwork, observation, and archival re-search; there are many rationalists whorely on precisely these tools. What di-vides them is method in the sense of howto construct theory, organize researchfindings, and address the issues of fal-sifiability and plausibility.

Rational choice will continue to haveits serious detractors in comparative

politics. It simplifies the world and hu-man psychology more than suits thetastes of many comparativists, espe-cially those committed to area studiesor interpretivist explanations. Its posi-tivist ethic may be unpalatable to post-modernists and others. The very com-mitment of rationalists to scientificprogress by means of fact-finding, test-ability, and partial universalism will re-main repugnant to some critics and animpossible goal to others. Rationalistsmust continue to refine and clarify theirmodels so as to increase their explana-tory power, and they must find moresatisfying means for arbitrating amongcompeting accounts. These are the tasksincumbent on all comparative socialscientists committed to explanation.

Structure and Con-figuration in Com-parative PoliticsIra Katznelson

Columbia [email protected]

At just the moment Comparative Poli-tics was founded in 1968 as a new jour-nal devoted to the reorientation of thesubdiscipline in a scientific, behavioraldirection from its older, more country-by-country institutional qualities, an-other scholarly tendency – structural,macroanalytical, configurative, histori-cal, and institutional – had begun to con-vene a research program that soon trans-formed the potential scope, ambition,and content of comparative politics.Turning the study of post-feudal moder-nity away from the realms of descrip-tion and metaphysics, the treatments ofimmense historical change by scholarsincluding Perry Anderson, ReinhardBendix, Shmuel Eisenstadt, SamuelHuntington, Barrington Moore, SteinRokkan, Theda Skocpol, Charles (withLouise and Richard) Tilly, andImmanuel Wallerstein, for all their dif-ferences, broadly came to share a com-mon mode of inquiry combining onto-logical and methodological commit-

stitutions and public policies of estab-lished democracies. As comparativistsstudy state-society relations that followa path first marked by Marx, Weber,Mosca, Michels, and Pareto. As theystudy the formation of states, they blendabstract theorization and detailed em-pirical studies. In addition, each displaysother vital characteristics. Examiningthe successes of the research schoolswith regard to each of these topics alsocasts light on the utility of various ana-lytic techniques: electoral analyses typi-cally use quantitative techniques tostudy survey results and work on state-society relations includes the results ofqualitative studies, while both researchon social movements and revolutionsand political economy vary in the useof quantitative and qualitative modes ofanalysis. Finally, these research themesalso elucidate the relations betweentheories developed in comparative poli-tics and those that characterize relatedfields in political science and the othersocial sciences, such as the utility ofhypotheses devised to explain electoralbehavior and social movements in theUnited States and methods and argu-ments drawn from economists in thestudy of political economy and from an-thropologists for the analysis of state-society relations. As we analyze theseresearch topics, we examine central is-sues of theory in comparative politics.

We have divided the essays into threeunits. The first, containing the chapterswritten by Margaret Levi, IraKatznelson, and Marc Howard Ross,offers briefs for each of the researchschools. The essays summarize eachschool’s core principles, noting varia-tions within the approach and present-ing recent work that points to new com-binations. The next unit contains thechapters written by Samuel H. Barneson mass politics, Doug McAdam,Sidney Tarrow, and Charles Tilly onsocial movements and revolution, PeterHall on the political economy of estab-lished democracies, and Joel S. Migdalon state-society relations in newlyformed states. The concluding unit con-tains essays by Mark I. Lichbach and


ments geared to the specification ofmacro-foundations for human action.

These scholars developed a probabi-listic approach to structure, wageringthat the most significant processes shap-ing human identities, interests, and in-teractions are such large-scale featuresof modernity as capitalist development,market rationality, state-building, secu-larization, political and scientific revo-lution, and the acceleration of instru-ments for the communication and dif-fusion of ideas. ‘Society’ in this orien-tation is replaced by the structured con-catenation of processes. These, while notdetermining of behavior in any strictsense, establish in specific times andplaces a calculus of cognitive and be-havioral probabilities by creating situ-ational orders in which individualsthink, interact, and choose. Persons, inthis view, are embedded agents operat-ing within relational structural fieldswhich distinguish the possible from theimpossible, the likely from the lesslikely.

In the 1990’s, neither the originalproject of Comparative Politics nor thestructural microanalysis of the 1960’sand 1970’s are giving powerful direc-tion to comparative political studies.Rather, as the 1995 World Politics sym-posium on “The Role of Theory in Com-parative Politics” signified, much of thedrive in the subfield now belongs to ra-tional choice scholarship and to variouspostmodern currents, leading some ofthe participants, including Peter Evansand Theda Skocpol, to worry whetherthe traditional core of comparative poli-tics risks being overwhelmed.

In truth, however, most current workstill resides in the space between thepoles of microeconomics and significa-tion, but much, alas, is rudderless andlacking in self-confidence. My paper forthe Lichbach-Zuckerman volume isgeared to ask how the macroanalyticaltradition might again come to play adefining role in comparative politics,less as an alternative to other currentsthan as a way of working capable of en-gaging and incorporating them confi-dently.

Put differently, I am searching for aprogram and way of working more am-bitious than the trajectory of the lineageof work growing out of the grandmacroanalytical scholarship of the1960’s and 1970’s that has produced thenew historical institutionalism of the1970’s and 1980’s which mainly hasfocused on comparative interest repre-sentation, public policy, and politicaleconomy. Viewed against the work oftheir predecessors, such scholars as Pe-ter Hall, Ellen Immergut, Sven Steinmo,and Paul Pierson who have producedsuperb studies have shortened their timehorizons, contracted their regime ques-tions, and narrowed the range of con-sidered options. This smaller-scale his-torical institutionalism has proved stron-ger as a skeptical response to other dis-ciplinary trends than as a sharply-etchedproject of the kind the earlier and moreadventurous macroanalytical scholar-ship appeared to propel. The result hasbeen something of a loss to the élan andpotential of comparative politics, espe-cially at the core of the enterprise. Inpart this contraction is the result of acertain lack of theoretical and method-ological self-consciousness at a timewhen Marxism, which in fact did muchof the intellectual work formacroanalysis, has been called intoquestion by global events and by chal-lenges to its essentialism, functionalism,and teleology.

In discussing how structuralmacroanalysis could reclaim its leader-ship position in comparative politics, thepaper surveys the macroanalytical hey-day, focusing primarily on Moore andSkocpol, and elaborates on five key is-sues central to the revival ofmacroanalysis: the qualities of compari-son, the relationship of history to ana-lytical social science; structural theoryafter Marxism; the special status of thestate; and the question of behavioral andstrategic microfoundations. This en-deavor is sustained, in part, by freshscrutiny of three pivotal texts: Alexis deTocqueville’s Democracy in America,John Stuart Mill’s A System of Logic andMax Weber’s The Methodology of the

Social Sciences; and by a perspectiveon institutions linking their configurationand design to the formation and exist-ence of political agents who possess par-ticular clusters of preferences, identities,and interests. Some of the best workalong these lines has been appearing instudies of the United States under therubric of American Political Develop-ment (APD): a genre of work that hasbegun to recover the dimensions of in-vention and surprise that attracted manyof us to political science and to com-parative politics two and three decadesago.

Culture and Identityin Comparative Po-litical AnalysisMarc Howard Ross

Bryn Mawr [email protected]

Two distinct, but not unrelated, fea-tures of culture are relevant to compara-tive politics. First, culture is a systemof meaning which people use to man-age their daily worlds, large and small;second, culture is the basis of social andpolitical identity which affects howpeople line up and how they act on awide range of matters. The effects ofculture on collective action and politi-cal life are generally indirect, and tofully appreciate the role of culture in po-litical life, it is necessary to inquire intohow the impact of culture interacts withinterests and institutions.

Culture is not a concept with whichmost comparativists are comfortable.For many, culture complicates issues ofevidence, transforming hopes of rigor-ous analysis into “just so” accountswhich fail to meet widely held notionsof scientific explanation. Culture vio-lates canons of methodological indi-vidualism while raising serious unit ofanalysis problems for which there areno easy answers. Culture to many, neo-Marxists and non-Marxists alike, seemslike an epiphenomenon offering a dis-course for political mobilization and


demand-making while masking moreserious differences dividing groups andindividuals. Finally, employing the con-cept of culture puts political scientistsinto a series of controversies over whichproponents of cultural analysis in an-thropology themselves are deeply di-vided. Each of these objections is ad-dressed in this article, and while I donot argue that they are unimportant, Ido not view them as sufficiently dam-aging to warrant throwing the baby outwith the bath water.

Cultural analysis of politics takes se-riously the postmodern critique of be-havioral political analysis and seeks tooffer contextually rich intersubjectiveaccounts of politics which emphasizehow political actors understand socialand political action. In cultural analy-ses, for example, interests are contextu-ally and intersubjectively defined andthe strategies used to pursue them areunderstood to be context dependent. Iargue that this view of culture can becompatible with rigorous comparison(while not denying its complexities). Atthe core of cultural analysis is the con-cept of interpretation. The interpreta-tions of particular political significanceare built from the accounts of groups andindividuals striving to make sense oftheir social and political worlds and in-terpretation refers both to the sharedintersubjective meanings of actors andto the explicit efforts of social scienceobservers to understand and to presentthese meanings to others. Shared inter-pretations of actors – world views– areimportant in any cultural analysis andoffer an important methodological tool,along with an examination of rituals andsymbols, for examining both systems ofmeaning and the structure and intensityof political identity.

This article first discusses five con-tributions which the concept of culturedefined as a system of meaning andidentity makes to comparative politicalanalysis: culture frames the context inwhich politics occurs, culture links in-dividual and collective identities, cul-ture defines the boundaries betweengroups and organizes actions within and

between them, culture provides a frame-work for interpreting the actions andmotives of others, and culture providesresources for political organization andmotivation. I then examine five centralthemes in cultural analyses of politics:culture and personality studies, the civicculture tradition, culture and politicalprocess (an approach which originatedin anthropology), political ritual, andculture and political violence. Third, Iidentify five critiques of cultural stud-ies of politics: unit of analysis issues,the problem of within-culture variation,the difficulty of distinguishing culturefrom social or political organization, thestatic nature of culture in explaining po-litical change, and the need to identifyunderlying mechanisms which suggest“how culture works.” The fourth sectionexamines the role of interpretation incultural analysis as an effort to link thecontextually rich political details foundin particular political settings (be theysmall communities or countries) to gen-eral domains of political life such asauthority, community, and conflict. Idiscuss the concept of psychocultural in-terpretations, and their methodologicalrelevance in the comparative study ofculture and politics for understandingprocesses such as ethnic and nationalidentity construction. I conclude thatculture is a too-often ignored as a do-main of political life and that culturalanalyses can enrich how we conceptu-alize areas such as political economy,social movements, and political institu-tions in a number of useful ways, oftencomplementing the insights derivedfrom interest and institutional ap-proaches.

Electoral Behaviorand ComparativePoliticsSamuel H. Barnes

Georgetown [email protected]

The study of electoral behavior hasprogressed in an ad hoc manner, largely

unconcerned with the grander theoreti-cal issues of political science as a disci-pline or with the great “isms” of thetwentieth century. Most research hasbeen country oriented; comparative poli-tics as a subdiscipline has not greatlyinfluenced this field and has not beengreatly influenced by it. Electoral stud-ies have formed a theoretical “island,”developing a data rich research traditionthat has been influenced by many con-ceptual and methodological trends with-out being monopolized by any singleone. This chapter focuses on the surveytradition and evaluates culturalist, insti-tutionalist, and rational choice contribu-tions to its study.

A review of studies of partisan choicesuggests that the bases of electoralchoice have varied according to spaceand time. An isomorphism exists be-tween theories of partisan choice andtheories of changing patterns of the mo-bilization of mass publics in democra-cies. Mobilization has shifted from be-ing based on social cleavages such asclass, religion, ethnicity, and the like;to political mobilization based on at-tachments to specific political objects;and – increasingly today – to cognitivemobilization reflecting individual deci-sions based on knowledge of issues, per-ceptions of interest, and “preferences,”including values (see Dalton, 1984).

Such an interpretation assumes that inthe earlier period attachment to groupsdefined by societal cleavages and insti-tutions, of which unions and churcheswere the most significant, was the basisof partisan identity. In the second, it wasattachment to particular political partiesand movements – which may have origi-nated in the older cleavage structures butthen acquired independent identities andloyalties. Advanced democracies today,with their highly educated populations,access to information through the me-dia and elsewhere, and declining in-volvement in associational life, are mov-ing toward cognitive mobilization.

Cognitive mobilization involves in-creasing individual processing of infor-mation, calculation of interests, and per-haps even the individual “construction”


of political identities that were ascribedat birth in earlier times. It reflects emerg-ing trends in advanced societies towarddemassification, higher education, ac-cess to information, privatization, ratio-nal egoism, and, especially, the domi-nant role of television.

There is strong empirical evidence forthe progression described above, butresearch findings are not conclusive.Some components, such as a decline inthe importance of many cleavages, therise in educational levels, and the roleof the media, are noncontroversial. Theassumed decline in involvement in or-ganizations and attachments to partiesseems generally valid but not for allcountries (see especially Kaase, New-ton, and Scarbrough, 1995).

Early academic studies in the UnitedStates focused on sociological variablessuch as occupation and group member-ships; investigators were surprised thatthese variables proved not to be veryimportant. However, here is little doubtthat social cleavages were important de-terminants of the vote in Europe. Reli-gion, class, and – in countries such asSpain and Belgium – region all fit intothe category of social partisanship.While Marxist and leftist parties madeextensive efforts at political educationof their members – and also sought tocreate large membership organizations– many voters possessed little sense ofa uniquely political identity. The votewas largely an expression of religious,class, and regional identities.

The weakness of social partisanshipin the United States led researchers atthe University of Michigan in the1950’s, in a widely influential series ofelectoral studies and publications, to de-velop the concept of partisan identifi-cation. Underlying the concept is alearning model (Converse, 1969) thathas proved useful in explaining the re-lationship between age and participa-tion, the strength of partisan attach-ments, volatility in voting patterns, per-ceived legitimacy of institutions, andother matters. Partisanship, like otheraspects of culture, is largely learnedthrough socialization. However, the

learning model has flaws. Early social-ization may prove especially inadequatein periods of change, in which the pastis a poor guide to the present. The dis-integration of major parties in estab-lished party systems in recent years islikewise damaging. The learning modeldoes not build in intensity, strong affect,charismatic leadership, and similar fac-tors that can accelerate or block change.

There is no argument among scholarsas to whether institutions make a differ-ence. The question is “how much?” Ithas been demonstrated in numerousstudies that institutional differences leadto differences in behavior – when all elseis held constant. But it also appears truethat institutions are greatly affected bythe cultures within which they operateas well as by the internal cultures thatthey generate over time. The questionof the importance of institutions givesrise to additional questions concerningthe question of what constitutes an in-stitution – how much is structure andhow much is culture – for similar struc-tures function quite differently in dif-ferent political systems.

Rational choice is at present the chiefcontender as a general theory of poli-tics. However, while it has contributedto the study of electoral behavior in theUnited States as well as in other indi-vidual countries, its contributions tocross-national research are largely non-existent. The empirical nature of elec-toral research has not meshed well withthe data-less analysis and model build-ing favored by some rational choicescholars. Few have invested heavily inthe costly and time-consuming task ofgenerating cross-national data. The el-egance and parsimony of formal theoryare quite promising for dealing withcomplexities of many countries andmany variables. But the work has notyet been carried out on a substantialscale.

This chapter argues that, rather thanall-purpose grand theories, it is islandsof theory, theories of the middle range,which have emerged largely inductively,that have thus far proved most useful inthe cross-national study of mass behav-

ior, including electoral behavior.

References

Philip E. Converse. 1969. “Of timeand Partisan Stability.” ComparativePolitical Studies. 2:139-171.

Russell J. Dalton. 1984. “CognitiveMobilization and PartisanDealignment in Advanced IndustrialDemocracies.” Journal of Politics.46:264-84.

Max Kaase, Kenneth Newton, andElinor Scarbrough, eds. 1995.Beliefs in Government. (Fivevolumes.) Oxford: Oxford Univer-sity Press.

Towards an Inte-grated Perspective onSocial Movementsand RevolutionDoug McAdam


Sidney TarrowCornell [email protected]

Charles TillyColumbia [email protected]

The protest cycles of the West in the1960’s stimulated renewals of thinkingabout contentious politics in both west-ern Europe and the United States in therationalist, structuralist, and culturalisttraditions. Not only that: there was alsoa separate development of models forsocial movements, revolutions, andother forms of contention. This led to aflourishing of research on contentiouspolitics but had the consequence of lead-ing to declining communication andsynthesis between traditions and acrossfields. Sociologists wrote of “move-ments,” political scientists of “protests”and “violence” and few scholars at-tempted to connect various forms ofcontention to one another or to routine


politics. To make matters worse, muchof the work on contention in non-West-ern, nondemocratic countries was beingdone in isolation from Western models.

In the last decade or so, some of theselacunae began to be filled and bridgesconstructed across traditions, areas anddifferent types of contention. Close ex-amination of work on social movementsand revolutions reveals numerous op-portunities for analogy and synthesisacross phenomena and theoretical tra-ditions. The paper begins with an out-line of the main lines of the structural-ist, rationalist and culturalist traditionsand of the potential interfaces amongthem. It then distinguishes opportunitiesfor synthesis with respect to analysis of(a) conditions generating contention, (b)mobilizing structures, and (c) framingprocesses. A political-process accountof the American civil rights movementillustrates the promise and problems ofsynthesis among rationalist, structural-ist, and culturalist models. A closing ses-sion alludes to the analogies amongmovements, cycles of protest and revo-lutions.

The Role of Interests,Institutions and Ideasin the ComparativePolitical Economy ofthe IndustrializedNationsPeter Hall

Harvard [email protected]

In keeping with the themes of the vol-ume, this essay compares approacheswithin the literature on comparativepolitical economy that emphasize, re-spectively, interests, institutions andideas. In broad terms, these approachescorrespond to three questions that havelong animated the field. Whose inter-ests are served by a given set of eco-nomic arrangements? How do institu-tions affect the operation of the

economy? How do specific conceptionsof the economy and its effects arise?

Interest-based approaches to theeconomy come in two dominant vari-ants. The first emphasizes the way inwhich changes in the internationaleconomy shift the material interests ofproducer groups so as to form new coa-litions behind particular economic poli-cies. The second emphasizes the way inwhich broader electoral pressures leadpoliticians to pursue particular kinds ofpolicies, so as to generate theories of thepolitical business cycle, retrospectivevoting, and the impact of coalition gov-ernment. These are powerful approachesthat speak directly to the intuition that,if a pattern of policy is to be sustained,it must advance the interests of broadsegments of society and of the politi-cians who implement it. However, mostsuch analyses are highly sensitive to theeconomic theories used to specify thematerial interests of the relevant actorsand some are relatively insensitive to thecollective action problems associatedwith coalition formation.

Institution-oriented approaches em-phasize how the institutional structuresthat organize the political economy af-fect the character of economic policyand performance. Initially rooted in theliterature on neo-corporatism, such ap-proaches now emphasize the way inwhich a variety of institutional features,ranging from the character of wage bar-gaining or the organization of the finan-cial system to the independence of thecentral bank, interact to generate nation-ally distinctive economic outcomes.Such analyses provide us with a way ofunderstanding why common economicchallenges may not produce convergentresponses across all nations or regionsand, in recent years, they have begun torestore the firm to a central place in theanalysis of the political economy. How-ever, as more interaction effects are ob-served, it becomes increasingly difficultto put nations into neat categories andthose who take this approach face in-creasing pressure to devise general theo-ries of institutional determination, in-cluding theories capable of explaining

the resilience or mutability of institutions.Idea-oriented approaches stress the

role of ideas in the determination of eco-nomic policy and come in several vari-ants. One privileges interest-based ex-planations but incorporates ideas tocomplete the causal chain, for instance,as focal points specifying one fromamong several competing equilibria.Another emphasizes the way in whichideas about appropriate policy in therelevant communities of experts acquireinfluence over policy and become insti-tutionalized in standard operating pro-cedures. A third goes farther to explainpolicy in terms of the cultural outlookscharacteristic of particular nations. Allof these approaches remind us that theeconomy is not visible to the naked eyeand can be modelled in multiple ways.However, it proves especially difficultto disentangle the effects of ‘ideas’ fromthose of other variables and to explainhow ideas can be persuasive in them-selves or at least partially independentof the power of their proponents.

Few scholars emphasize interests, in-stitutions or ideas to the exclusion of theothers; and there is considerable poten-tial for integration among these perspec-tives. However, the field also faces amore basic methodological divide be-tween those who view political economyas an effort to apply the methods of eco-nomics to politics and those who em-phasize the way in which noneconomicfactors, associated with politics or cul-ture, influence the course of events thatthe former attempt to explain in largelyfunctional or rational terms.

Proponents of institution-oriented ap-proaches find themselves in the middleof such debates and, perhaps as a con-sequence, some of the most interestingdevelopments in the field are takingplace at the interface between this andthe other two approaches. At one inter-face, the “new economics of organiza-tion” and “endogenous growth theory”have opened up new ways of reconcil-ing institutional approaches with tradi-tional economic theory. At the other, aseries of debates about the character of“social capital” and the “new institution-


alism” in organization theory providesites for fruitful dialogue with those whostress the importance of culture to eco-nomic outcomes.

Contemporary economic develop-ments are bringing the field back tosome of the core issues that inspired itsfounders, including, most notably, ques-tions about the relationship between thestate and the market, given that eachprovides a different kind of mechanismwith the potential to advance the “pub-lic interest”. However, there are still atleast three significant gaps in the litera-ture.

Paradoxically, the field as a whole haspaid little attention to the problem of ex-plaining the distributive inequities thatnational patterns of policy and perfor-mance can generate. It has not yet tiedthe study of economic policy to the ques-tions about representation, which havelong been central to political science.And none of the three approaches sur-veyed here have yet captured very wellthe contribution that political conflictand debate can make to the construc-tion of interests. These issues provideresearch frontiers on which recent workhas only now begun.

Studying the StateJoel S. Migdal


Political science has most oftentreated states as firmly bounded entities.Within the state’s boundaries, citizenshave also been seen in clear, formalterms. Both realist and liberal ap-proaches in international relations, forexample, have assumed the notion ofstates exercising sovereignty withintheir existing borders. And, in compara-tive politics, the notion of universal citi-zenship has been key in terms of states’claims to legitimacy as representativeof fairly homogeneous populations (ex-pressed as the nation or society).

This paper will re-examine the con-ceptions of states’ fixed boundaries andof a single society, represented by the

state, within those borders. WesternEuropean unification, masstransnational migration, state disintegra-tion in eastern Europe and Africa, spi-ralling international capital flows andmore impel political scientists to viewstate boundaries, control, and sover-eignty in different terms. Varying clas-sifications of citizens – either formally,as in Indonesia, or informally, as in Ger-many – also demand re-thinking the re-lationship of the state to society and ofsocial groups to each other. Some pre-liminary work – e.g., Yoav Peled’s ar-ticle in the APSR, “Ethnic Democracyand the Legal Construction of Citizen-ship,” and Yasemin Soysal’s Limits ofCitzenship – begins the process of as-saying the mix of state, ethnicity, andcitizenship. Now, political science needsto make state-formation and society-for-mation, as well as the relationship be-tween them, a central part of its agenda.

Much of our existing thinking stemsfrom earlier conceptions of politics asthe clash of established interest groupswithin a given territory. Such thinkinggives rise to methodologies and ap-proaches, such as rational choice or stan-dard structuralist theories, which assumefixed preferences from which politicalscientists can deduce action.

The fluidity of state and social bound-aries that we are witnessing at the endof the twentieth century must lead us toquestion our old approaches. We mustnow begin to treat states and socialgroups not as established, but as vari-able. How and why do they form andre-form? Politics cannot simply be un-derstood in terms of well-defined struc-tures with given preferences, as ratio-nal choice or structuralism would haveit. We must work towards an under-standing of politics given the indeter-minacy of the boundaries of states andsocial forces.

What is the Current

State of Theory inComparative Politics?Mark I. Lichbach

University of Colorado, [email protected]

Compared to twenty-five years ago,self-conscious theoretical reflectionfinds almost no home within our field.We do not take our theories nor our theo-rists seriously. Evidence to support thisharsh judgment comes from our lead-ing journal’s recent symposium on “therole of theory in comparative politics”(World Politics, October 1995). Theparticipants minimized the value of de-ductive, a priori theorizing of the sortthat is done within strongly defined re-search communities. Moreover, in spiteof the fact that the symposium includedwidely acknowledged experts in specificresearch traditions, apparently no oneviewed, for example, today’s rational-ist/culturalist divide as theoretically in-teresting, exciting, and productive.Structural or institutional analysis wasnot even recognized as a theoreticallydistinctive enterprise but rather wasthought of as a part of the field’s “messycenter.” Most participants feared that thefield might return to the sort of Marx-ist/functionalist debate that character-ized it in the 1950’s and 1960’s. Conse-quently, method – prediction, compari-son, counterfactuals, history, quantita-tive and qualitative data, explanation, in-terpretation, causation, and generaliza-tion – was on everyone’s minds. The“nomothetic” vs. “ideographic” dividewas what really animated discussion.The consensus was that mostcomparativists are part of the consen-sus: today’s comparativists practice“theoretically informed empirical politi-cal analysis” and adopt “diverse concep-tual lenses” (2) and “eclectic combina-tions” (5). They are interested in “ques-tions” and “empirical puzzles” (10).Hence, “comparative politics is verymuch a problem-driven field of study”and comparativists are mostly interestedin solving “real-world puzzles” (46).


The flaw of this pragmatist, means-oriented heaven was perceptively rec-ognized by the symposium’s organizer.He concludes that “if the problem ori-entation of the field tends to relegate therole of theory mainly to that of a tool ofempirical research, the quest for causalgeneralizations, by contrast, moves itsrole to the forefront” (47). Similarly, theconclusion from a methods symposiumon comparative (small-n) studies in an-other leading journal (American Politi-cal Science Review, June 1995) may bestated here as paraphrase of Kant: goodtheory without good research design isempty; good research design withoutgood theory is blind (454). AsRogowski’s (1995) important essaymakes clear, one cannot begin inquirywith “evidence” derived from and usedto test “theory”; one must begin withtheoretically-embedded observations.The inevitable conclusion is that onewill eventually reflect on the nature ofthat theory – which leads one to ques-tions broadly defined as “social theory”or “philosophy of social science.”

World Politics’s symposium did notcontribute to the cause of theory in com-parative politics because its picture oftheory in our field as dominated by a“messy center” is inaccurate and self-defeating. My essay seeks to refute thatperspective and advance theory in com-parative politics in three ways.

First, I recognize that three researchtraditions – the rationalist, theculturalist, and the structuralist – areactive in contemporary comparativepolitics. Section 2 thus begins the analy-sis with three exemplary comparativists.Each thinks of himself or herself as amember of a strong research commu-nity. Robert Bates (1989) argues that heis a rationalist, James Scott (1985) iden-tifies with the culturalists, and ThedaSkocpol (1979) places herself within thestructuralist school. While each recog-nizes the value of synthesis and thecross-fertilization of ideas, each is prin-cipally concerned with advancing a par-ticular intellectual tradition and theoreti-cal agenda that transcends comparativepolitics.

Second, I set the dialogue among theschools within the historical context ofthe development of social theory. Sec-tion 3 thus attempts to understand thethree research communities by tracingthem back to Talcott Parsons’s (1937)effort to systematize the classic socialtheorists and thereby integrate socialtheory. I have modified his approach totake account of the structure-actionproblem of reconciling individuals andcollectivities. I call this modified ap-proach the socially embedded unit act.Using this meta-framework to provideinsight into the individual frameworks,I explore the differences among the threetraditions with respect to core assump-tions, explanatory strategies, and histori-cal developments.

Finally, I set the dialogue within theframework of the historical situationconfronting the contemporary world.Section 4 thus seeks an underlying unityin rationalist, culturalist, and structur-alist thought by delving even furtherback to Max Weber’s master problemof a century ago. Weber studied the dia-lectic of modernity in world historicaland comparative perspective: how rea-son and irrationality manifest them-selves at individual and societal levelswith great normative and empirical sig-nificance. The dialectic is important tocontemporary politics in the West. Dueto the West’s influence on the globe, thedialectic is equally important to the en-tire world community of nations.

Section 5 summarizes my theme aboutthe problem situation of contemporarycomparative politics: there are funda-mental difficulties with a field that con-sists only of a “messy center” and basicvirtues with a field that embraces cre-ative confrontations, which can includewell-defined syntheses in particular re-search domains among strongly definedresearch communities. Comparativistsshould explore the rationalist-culturalist-structuralist debate andthereby appreciate the different struc-ture-action combinations of interests,identities, and institutions that guide in-quiry. Even self-described “problem-oriented” comparativists – those who

think of themselves as part of a “messycenter” – should be aware of the com-peting research traditions that have his-torically been a part of our field. Wecannot remain theoretically challenged– a field of theoretical Philistines – andactually solve substantive problems.Contemporary comparative politicstherefore will be greatly enriched by adialogue among the traditions, espe-cially one that is informed by self-con-scious reflection about the enduring is-sues of social theory. Comparative poli-tics needs strong and yet mutually sym-pathetic intellectual communities: be-lievers who raise questions and nonbe-lievers who appreciate answers.

References

Bates, Robert H. 1989. Beyond theMiracle of the Market: The PoliticalEconomy of Agrarian Developmentin Kenya. Cambridge: CambridgeUniversity Press.

Parsons, Talcott. 1937. The Structureof Social Action. New York: FreePress.

Rogowski, Ronald. 1995. “The Roleof Theory and Anomaly in Social-Scientific Inference.” AmericanPolitical Science Review 89 (June):467-70.

Scott, James C. 1985. Weapons of theWeak: Everyday Forms of PeasantResistance. New Haven: YaleUniversity Press.

Skocpol, Theda. 1979. States andSocial Revolutions: A ComparativeAnalysis of France, Russia andChina. Cambridge: CambridgeUniversity Press.

Reformulating Scien-


tific Understandingand AdvancingTheory in Compara-tive Politics

Alan S. ZuckermanBrown [email protected]

Explanations based on covering lawsand causal accounts have long definedthe set of acceptable forms of scientificunderstanding in comparative politics.The principles of the research schoolsthat guide analysis in comparative poli-tics reflect this expectation. Rationalchoice theory draws heavily on the epis-temology of logical positivism, center-ing explanations around covering laws.Structuralist analyses combine nomo-logical and causal explanations. At theheart of the divisions among culturalistsis a fundamental disagreement over thenature of explanation. While someinterpretivists seek causal accounts andothers apply nomological principles,many seek only to understand an actor’sgoals, eschewing the need for theory orcausal mechanisms, moving their schol-arship outside the realm of scientificexplanations. In comparative politics, asin political science and all disciplinesthat claim to be science, covering lawand causal explanations have long stoodas the standards for scientific under-standing.

The widespread appreciation of thesestandards to the contrary notwithstand-ing, there is reason to expand the mod-els of scientific understanding used incomparative politics. Each of the stan-dard forms of explanations has deficien-cies. Some of these are long-standing,appearing in the work of philosophersof science. Others derive from recentadvances in various sciences, that high-light the limited conception of realityinherent in the standard forms of scien-tific understanding. Causal and nomo-logical explanations apply only to simpleand determinate patterns, and they as-

sume that reality rests on smaller, morebasic phenomena. They envision a worldcomposed of linear relationships amongvariables; parity in the size of cause andeffect; recurrent, determined and closed-ended patterns; and the fundamental in-significance of chance happenings.Analyses of complex dynamic systems,however, call attention to the limitationsof this ontology. Nomological and causalexplanations are not useful for the analy-sis of a world of nonlinear relationshipsamong phenomena; no necessary paritybetween size and effect; sensitive de-pendence on initial conditions; the possi-bility of change at any point in time; open-ended processes; and the presence ofchance as a substantive part of processesand their explanations – all of which arecharacteristics of an ontology associatedwith the chaos theory and other modesof analyzing complex dynamical sys-tems. Covering law and causal explana-tions apply to a relatively limited set ofphenomena.

Although the precise utility of chaosand complexity theory in comparativepolitics still remains to be demonstrated,there are obvious parallels between po-litical phenomena and the world de-scribed in these new fields of study.Consider the following summary state-ments about recent research:

1. Political and social structures per-sistently display fluid and com-plex patterns. Ethnic, social class,and political diversity character-ize European and North Ameri-can societies over the past cen-tury.

2. Political attitudes are complex,variable, and probabilistic. Hence,explanatory questions need to ex-amine the probabilities that ac-company phenomena.

3. Nonlinearities characterize inter-actions among citizens and thepeople around them. The influ-ence of social contexts and dis-cussion networks is not simpleand straightforward but complexand interactive.

4. Formative characteristics of politi-cal organizations and decisions

constrain subsequent processesand events.

5. Political protest, revolutions, andcabinet crises display unstablepatterns, making their emergenceunpredictable.

6. More generally, chance factorsinfluence social and political pat-terns. Chance is neither “statisti-cal noise,” nor “unmeasured vari-ables.” Chance is an inherent partof the political world.

In comparative politics, research ex-plores a political world that is not en-compassed by the simple ontology ofcausal and nomological explanations.

Reformulating the epistemology andontology of comparative politics affectsthe research schools. Rationalists needto reduce the domain of what they studyor expand their theoretical principles.Classic rational choice theory applies tolimited types of political phenomena:the simpler the process, the more de-fined the rules and the meanings of win-ning and losing, the more likely are per-sons to use the cognitive processes ofinstrumental rationality, and the morelikely are rationalist principles to be use-ful. The more complex the circum-stances, the more likely are people todraw on other sorts of knowledge andreasoning, in which instrumental calcu-lation plays a limited role. In turn, struc-tural analysis needs to jettison its real-ist assumptions. State, ethnicity, socialclass, political cleavage and other con-cepts are not natural types. They do notsignify cohesive sets of persons; aggre-gate patterns are not easily predicted,and they are certainly not determined.This mode of analysis needs to combineseveral distinct principles: (a) the per-sistent effect of formative patterns anddecisions; (b) the expectation of unpre-dictable events that have major conse-quences; (c) the divergence of like sys-tems over time; (d) aggregate charac-teristics that may not be reduced to thedecisions of individuals; and (e) the in-herently probabilistic effect of struc-tured relationships on individuals. Sub-stantial changes in the principles of struc-tural analysis follow from changing the


understanding of the underlying politicalreality.

There is reason to alter the standardsof scientific understanding used in com-parative politics. General theoreticalpropositions may include law-likepropositions, not necessarily laws. Theymay apply to specific sets of cases; theyneed not display an unlimited scope. Theeffort to establish covering principlesstands in the way of other kinds of ex-planations. It directs research to exam-ine the accuracy, reliability, and the do-main of general laws. It directs atten-tion away from the analysis of more trac-table problems. It is impossible to es-tablish general laws and causal mecha-nisms with absolute certainty. All ex-planations require assessments of theirrelative certainty. All benefit from teststhat eliminate nuisance factors and as-sess the power of rival plausible hypoth-eses. Note as well that all explanationsrequire theories; in their absence, theselection of explanatory variables is ar-bitrary.

Comparativists need to combine gen-eral claims and particular details. For-mal models in comparative politics re-quire bridges that link the abstract math-ematical claims to the explanation ofparticular cases and sets of cases. Sta-tistical models should not be bound bythe assumptions inherent in linear mod-els. “Why questions” need to include theprobability of the emergence of particu-lar events, not only their absence or pres-ence. Analysis may include processmodels that respond to “How ques-tions,” moving the analytic focus awayfrom questions about emergence andcross-national variation. Chance shouldfind a central place in political analysis.The complexities of the political worldneed to be incorporated into the theo-ries of comparative politics.

As these changes occur, fresh theo-retical combinations emerge. Rational-ists, culturalists, and structuralists ex-amine rationality as one element in asystem of meaning. Scholars from allthree schools examine the relationshipbetween individual decisions, social con-texts, and institutions. Blending theoreti-

cal positions risks obliterating distinctionswithout necessarily leading to theoreti-cal gains. Rigorous adherence to thestandards of explanation in comparativepolitics enables theory to advance, dis-

tinguishing the path of scholarship fromthe swamp of the messy center.

News & Notes(News & Notes began on page 3.)

The committee also conferred the awardupon George Tsebelis for his article,“Decision Making in Political Systems:Veto Players in Presidentialism,Parliamentarism, Multicameralism, andMultipartism,” British Journal of Politi-cal Science 25, 3 (July 1995): 289-325.

The runners up for the article awardwere Jonas Pontusson for “Explainingthe Decline of European Social Democ-racy: The Role of Structural EconomicChange” in World Politics (v. 47, no. 4(July 1995): pp. 495-533), SylviaMaxfield for “Financial Incentives andCentral Bank Authority in Industrializ-ing Nations” in World Politics (v. 46,no. 4 (July 1994): pp. 556-589), andMichael Bratton and Nicolas van deWalle for “Neopatrimonial Regimes andPolitical Transitions in Africa” in WorldPolitics (v. 46, no. 4 (July 1994): pp.453-489).

Next year, the selection of awards willbe divided among three committees.Two will confer prizes in the name ofGreggory Luebbert; one for the bestbook and the other for the best articlepublished in the field of comparativepolitics in the last two years.The committees will consist of:

Luebbert Book Committee: Sam Popkinof UCSD (Chair), Jeff Frieden (HarvardUniversity), and D. Michael Shafer(Rutgers).

Luebbert Article Committee: Gary Coxof UCSD (Chair), Susan Whiting (Uni-versity of Washington), and ArunAgrawal (Yale).

A third committee will deliver the SageAward, endowed by Sage Publications,for the best paper presented at the an-

nual convention. The committee will bechaired by Desmond King (Oxford), andwill include Allan Kornberg (Duke) andPeter Gourevitch (UCSD).

Nancy Bermeo (Princeton) is in chargeof organizing panels for the AnnualMeetings of 1997. In a new departure,the Executive Committee has recom-mended that she work with MiriamGolden, editor of the Newsletter, to or-chestrate debates over central issues inthe field, the debates to originate in theNewsletter and to culminate in roundtables or panels at the Convention.

The first appears in this issue; the Juneissue will focus on rational choice andpolitical culture.

Change of address for the Newsletterwill automatically take effect for Sec-tion members when a change of addressis filed with the APSA. Please do notsend change of address information tothe Newsletter.

Use the Newsletter in the classroom!The APSA has authorized universityteachers to reproduce articles from theNewsletter for use in the classroom atno charge. Take advantage of this policy,and introduce your graduate students tothe latest research, issues and debatesin comparative politics.

How to subscribe: Subscriptions to theAPSA-CP Newsletter are a benefit tomembers of the Organized Section inComparative Politics of the AmericanPolitical Science Association. Subscrip-tions are paid for out of members’ dues.To join the APSA, call (202) 483-2512.

(News & Notes continues on page 23.)


The last twenty years have been un-kind to students of politics in countriesoutside the North Atlantic core. At pre-cisely the moment when the shift toauthoritarianism had “been fully ex-plained by a variety of converging ap-proaches and [was] therefore understoodin its majestic inevitability and perhapseven permanence” (Hirschman 1979,98), democratization swept throughlarge numbers of countries. In a secondequally unexpected development, nu-merous governments began to abandonstate interventionist economic policiesin favor of greater market orientation.On top of everything else, the Sovietempire collapsed. Though scholars havegreeted many of these events with de-light, they did not predict them and, eventoday, could if they wished explain morepersuasively why these events shouldnot have taken place than why they have.

Confronted by these compelling andexciting events in the world, scholarsquickly turned their attention to tryingto understand them. One of the firstfruits of these investigations was the rec-ognition that few of the theories dear tothe hearts of comparativists offeredmuch leverage for explaining recentevents. The inability to explain, muchless foresee, these transitions has madeus more conscious than ever of the pal-try accumulation of theoretical knowl-edge in the comparative field. Imagina-tive theories and sweeping paradigmsarise to the accompaniment of excite-ment and fanfare, but they disappearwith equal frequency and rapidity, leav-ing scarcely a trace. Why?

I begin the book with the argumentthat our limited success in accumulat-ing theoretical knowledge deriveslargely from methodological and re-search design norms in the comparativefield. Because of these norms, we havefailed to test rigorously the arguments

we advance and have accepted the ar-guments of others without demandingthat they be supported by strong evi-dence. The book aims to show in a com-pelling manner why standard method-ological practices in the comparativefield should be improved. It does thisby demonstrating the consequences ofsome of the methodological pitfalls mostcharacteristic of large parts of the sub-field.

The best argument that methodologi-cal issues need to be taken seriously isthe demonstration that inappropriatemethodological choices lead to wrongsubstantive conclusions. Much of thebook is taken up with retests using moreappropriate research designs of argu-ments in important and respected workwith which most comparativists are fa-miliar. By using, where possible, graphi-cal demonstrations of the logical flawsinvolved in the original research de-signs, and by drawing examples of whatcan go wrong from well-known and re-spected literature, I make clear the un-desirable consequences of methodologi-cal practices usually considered accept-able by comparativists.

All the studies discussed in the bookare intelligent, plausible, insightful, andpossibly correct in their knowledgeclaims. All have been advanced byhighly respected social scientists. Theeffort here is not to discredit argumentsor belittle authors – who are, after all,working within accepted conventions –but to demonstrate the deficiencies ofthe conventions themselves. These con-ventions affect not only authors butreaders of comparative politics. Authors,including some of those discussed be-low, are frequently aware of the tenta-tiveness of the evidence supporting theirarguments and indicate their awarenessin the caveats they attach to them. Read-ers, however, tend to ignore the caveats

and give greater to weight to unsystem-atic evidence than it deserves. Manystudies in which authors have carefullyhedged their explanatory claims are dis-cussed in seminars, cited in literature re-views, and summarized in qualifyingexams as though the tentative argumentsadvanced were actually supported bysolid evidence. One of the purposes ofthe book is to decrease the credulity ofreaders.

Some comparativists may find thebook controversial, since it criticizespractices widely accepted and self-con-sciously defended within the subfield. Ibelieve, however, that methodologistswill see it as an accessible exposition ofwell-known ideas. All but one of themethodological ideas presented simplyshow the implications for non-quanti-tative work of ideas about the logic ofresearch design that have been carefullyworked out and accepted without reser-vation among those who do quantitativework. I expect the book to be useful tosocial scientists with little training in sta-tistics and to be assigned in introduc-tory graduate classes on research designand methods.

Chapter I: Rise and Decline of Para-digms in Comparative Development

This section discusses the rise and fallof theories of development from WorldWar II to the present, making the casethat the scholars who advanced thesetheories failed to make full use of avail-able evidence. As a result, later research-ers rapidly discovered that these theo-ries could not account for events andprocesses in the world. In consequence,the theories, and eventually the para-digms in which the theories were em-bedded, were swept away likesandcastles.

Though all theories may be “born inideological sin rather than scientific vir-tue,” to use Dick Sklar’s words, the rea-

Paradigms and Sandcastles: Research Design in Comparative PoliticsBarbara Geddes, University of California, Los Angeles

[email protected]

Forthcoming, University of Michigan Press


using evidence drawn from samples ofcases which, unlike those used in theoriginal arguments, are not correlatedwith the outcome on the dependent vari-able. In each retest, an unbiased sampleyields conclusions at odds with thoseoriginally advanced.

Chapter 3: Case Studies and Path De-pendence

The comparative case study method,though useful in some kinds of studies,precludes serious tests of theories whenpaired with the complex, path dependentarguments often made bycomparativists. Path dependent argu-ments begin with an initial causal hy-pothesis about an outcome at time one,followed by a series of interveningcausal hypotheses to explain a sequenceof later outcomes. At each interveningpoint, new independent variables enterthe argument to explain different out-comes, and whichever outcome occurslimits what may happen in the future.

Each link in a path dependent argu-ment could in principle be tested rigor-ously by identifying the universe ofcases in which a particular hypothesisshould hold (that is, cases having theinitial conditions specified in the earlierstages of the argument), randomlychoosing a sample from the universe ifit is large, and examining the outcomesin those cases to see if they are consis-tent with the hypothesis. Reliance oncomparative historical case studies of alimited number of countries as the pri-mary source of evidence, however,eliminates the possibility of this kind oftesting procedure. Instead of testing eachhypothesis on the universe of cases towhich it, though perhaps no other hy-pothesis in the sequence should apply,a small number of cases are selected atthe beginning and followed in detailthroughout the relevant historical period.Consequently, all hypotheses are testedon a few, often three, cases. By the endof the analysis, the number of indepen-dent variables almost always exceedsthe number of cases.

To demonstrate the possibility of find-ing cases that fit the initial conditionsrequired to test later stages of path de-

pendent arguments, I retest Lipset andRokkan’s “freezing” hypothesis on a setof Latin American countries that, thoughnot included among the cases they ex-amined, meet the initial conditions setout at the relevant stage in their argu-ment. As in the selection bias examples,the retest fails to confirm the originalargument.

Chapter 4: Fluid Definitions of Vari-ables

No one would argue against the needto define concepts and variables pre-cisely. Nevertheless, the meaning of keycausal variables often seems to vary asan author moves from one element ofthe argument to another, case to case,or time period to time period. The con-cepts most subject to this kind of slip-page are those with no clear empiricalreferents, such as state autonomy, threat,power, and insulation. I demonstrate theeffect of loose variable definition bychoosing a few key variables from im-portant studies, operationalizing each toreflect different definitions availablewithin a single work by a single author,conducting tests of the argument ad-vanced using differentoperationalizations, and showing thedifferences in conclusions that result.

Chapter 5: “Big Structures, Large Pro-cesses, and Huge Comparisons”

In this chapter I argue that we canmore fruitfully focus research on under-lying processes than on final outcomeswhen trying to understand major eventsin history. The point of this argument isnot that there is anything methodologi-cally wrong with trying to explain ma-jor events, but that our ability to do sois severely limited by the interaction oftwo problems: the large role of non-sys-tematic factors in determining any par-ticular outcome; and the very primitivetechniques we rely on, which give us noway of determining when we have iden-tified systematic factors that explainsome but not all of the variation in out-comes. Perhaps the best example of thisproblem can be found in contemporaryefforts to explain democratization. Mostare frustratingly atheoretical, in part

son for the failure to accumulate theo-retical knowledge in the field of com-parative development is not ideologicalbias per se, but that the methodologicalnorms of the subfield do not compen-sate for individual scholars’ biases. Inprinciple, the systematic testing of hy-potheses and replication of studies con-ducted by other scholars should gradu-ally eliminate the biases introduced intoanalyses by individual scholars. Normsin the comparative politics field, how-ever, legitimize research designs thatpreclude even minimally effective testsof hypotheses and set standards of evi-dence that are too low to overcome ideo-logical bias.

In the book, I deal with several char-acteristic features of research design andthe use of evidence to support knowl-edge claims that have wide acceptancein the comparative field. The specificpractices discussed at length include:

• Selection on the dependent vari-able

• The comparative case studymethod in complex, path depen-dent explanations

• Fluid definitions of key variables• The attempt to explain large com-

plex outcomes such as democra-tization and revolution

In chapters dealing with each of these,I lay out the logic underlying what theresearcher does when he or she use thepractice I criticize. I use extended ex-amples drawn from well-known authorsand schools of thought to illustrate eachmethodological problem.

Chapter 2: Selection BiasOne of the most durable conventions

in comparative politics is the selectionof cases for study on the dependent vari-able. That is, if we want to understandsomething, for example, revolution, weselect one or more occurrences and sub-ject them to scrutiny to see if we canidentify antecedent events as causes.This chapter demonstrates the conse-quences of this research strategy by re-testing arguments made in ThedaSkocpol’s States and Social Revolutionsand in the literature on successful eco-nomic performance in the Asian NIC’s,


because it can be shown that virtuallyany cause identified as important insome set of cases was not important inone or more other cases, and in part be-cause analysts who have observed theprocess closely know that serendipitousfactors such as deaths of key leaders,personality traits, and natural disastershave influenced outcomes. In my view,these are insoluble problems as long aswe choose major events as dependentvariables.

In choosing such complex dependentvariables, we inadvertently paint our-selves into a methodological corner inwhich inductive research strategies pre-vail, and the implicit model of explana-tion always turns out to be an enormouskitchen sink regression. While simpledeductive explanations of these out-comes never seem to be confirmed byevidence, the long lists of hypothesesgenerated by inductive research strate-gies cannot be tested because variablesalways end up outnumbering observa-tions.

At its best, this approach to the studyof big questions is analogous to that ofmedical researchers who try to under-stand the onset of cancer by amassingdata on all the dietary, hereditary, andenvironmental factors that can be imag-ined to increase the likelihood of thedisease. This is useful research. It re-sults in the accumulation of factualknowledge and leads to some inductivegeneralizations. It does not, however,lead to an understanding of the processthrough which cancer develops. For that,researchers must step back from thecomplex outcome, the diseased person,and focus instead on basic mechanisms,for example, the nature of cells and thegenetic mechanisms that regulate celldivision. They must concentrate on theunits within which the process occurs,the cell and the chromosome, rather thanon the overall outcome that results: thediseased organism.

In a similar manner, I think studentsof comparative politics need to seek tounderstand underlying processes ratherthan “explaining” complex outcomes.To do that, we need to focus on the fun-

damental unit of politics, in most casesthe individual. We need to break up thetraditional big questions into more pre-cisely defined questions that are moretheoretically accessible. This would in-volve a redefinition of the questions ofinterest so that the construction and test-ing of theories becomes possible.

To demonstrate this approach in prac-tice, in this chapter I examine the pro-cess underlying one component of re-gime transition, the breakdown of au-thoritarian regimes. I use simple gametheory to model the relationship betweendictators and their supporters and rivalsin three types of authoritarian regime,showing the effects of different authori-tarian regime types on the probabilityand manner of regime dissolution. I thentest a few implications drawn from themodels, using evidence from a data setthat includes all authoritarian regimesin existence at any time between 1946and 1990.

The purposes of this exercise are two-fold. First, it is a demonstration of whatI mean by a focus on process rather thanoutcome. Using game theory is a wayto emphasize the incentives facing de-cision makers and the logic of the situa-tion they face. It simplifies reality in auseful way, allowing the analyst to con-centrate on particular interactions with-out being distracted by all the otherthings that are simultaneously occurringin the world. The models do not seek toexplain regime transition; rather, theyaim to explain a process that contrib-utes to regime transition.

Second, this example illustrates theresearch strategy of testing the implica-tions of an argument rather than the ar-gument itself. Many arguments cannotbe tested, for either practical or logicalreasons. Nevertheless, we want to havesome assurance that they are not false.Often some of the implications of anargument can be tested, even though theargument as a whole cannot. I seek topersuade readers that a wide-rangingsearch for testable implications shouldbe a part of every conscious researchstrategy.

Chapter 6: Contending Approaches to

Theory BuildingThis chapter considers what ap-

proaches to explanation are most likelyto result in the accumulation of theoreti-cal knowledge. It discusses the charac-teristics that explanations need to havein order to be useful in the social enter-prise of knowledge building, and assessesseveral of the approaches now advocatedby comparativists in light of these needs.A review essay on the uses and limita-tions of the rational choice approach il-lustrates the argument. The point here isnot to advocate further imperialism byrational choice but rather the creation ofother, possibly more realistic, approachesthat have most of the characteristics thathave given rational choice its fruitfulness,stamina, and reach.

ConclusionA new theory is like a river in spring.

Rushing down from the high ground, itcuts a narrow channel through the wil-derness of complexity. When it encoun-ters factual obstacles too large to sweepalong, it should be diverted into a new,equally rapid and narrow course. In thecomparative field, however, old theories,modified by many collisions with incon-venient facts, are like rivers that havereached the delta after crossing a broadplain. They dissipate into numerous smallchannels meandering through swampuntil merging gradually and impercepti-bly into the sea of thick description. Wehave now reached the swamp stage onmultiple fronts. This is not, in my view,a temporary state of affairs. Theoreticalinadequacy will continue to plague us aslong as we remain loyal to the time-hon-ored methodological traditions of com-parative politics.

References

Hirschman, Albert O. 1979. “The Turnto Authoritarianism in Latin Americaand the Search for Its Economic De-terminants,” in David Collier, ed., TheNew Authoritarianism in LatinAmerica. Princeton: Princeton Univer-sity Press.

Tilly, Charles. 1984. Big Structures,Large Processes, Huge Comparisons.New York: Russell Sage.


Given these marginal per-centages from Louisiana in1990, what are the cell en-tries?

Vote No Vote

Black ? ? 26.6%

White ? ? 73.4%

68.5% 31.5%

My special thanks to MiriamGolden for encouraging me to writethis article.

For several years prior to publication,I ran a “virtual seminar” on the subjectof this book with faculty and studentsfrom other universities. The arrange-ment was that participants would emailme while reading the manuscript withanything they found unclear, unhelpful,or untrue, and in addition to revisingthe manuscript I tried to return the fa-vor with immediate explanations viaemail of anything that was holding themup. As the book is now in production,the seminar has ended, but I would stillbe happy to respond to related inquir-ies.

IntroductionThis article summarizes a book manu-

script that gives a solution to the eco-logical inference problem, a method ofinferring individual behavior from ag-gregate data that works in practice. Be-low, I describe the problem in somewhatmore detail, give an example of how themethod works, and discuss how the so-lution can be of use in comparative poli-tics research. To save space, I do notdescribe the details of the statistical pro-cedures introduced. (The title of thisarticle is the tentative title of the book.Until publication by Princeton Univer-sity Press in April 1997, the completecurrent draft of the book is availablefrom my home page on the World WideWeb, http://GKing.Harvard.Edu; alsoavailable there is an easy-to-use com-puter program that implements all thestatistical and graphical methods intro-duced.)

Ecological inference, as traditionallydefined, is the process of using aggre-gate (i.e., “ecological”) data to infer dis-

crete individual-level relationships of in-terest when individual-level data are notavailable1 . The best existing methodsusually lead to very inaccurate conclu-sions about the empirical world; indeed,frequently, they give impossible answers,such as -20% of Israeli Labor Party vot-ers remaining loyal between the last twoelections. The ecological inference prob-lem is to develop a method that givesaccurate answers in practice. Ecologi-cal inferences are required in political sci-ence research when individual-level sur-veys are unavailable (e.g., local or com-parative electoral politics), unreliable (ra-cial politics), insufficient (political geog-raphy), or infeasible (political history).They are also required in numerous ar-eas of public policy (e.g., for applyingthe Voting Rights Act) and other aca-demic disciplines ranging from epidemi-ology and marketing to sociology andquantitative history. The ecological infer-ence problem has been among the long-est standing, most actively pursued, andconsequential in all of quantitative socialscience. It was originally raised over 75years ago as the first statistical problemin the nascent discipline of political sci-ence (Ogburn and Goltra, 1919) and hasheld back research agendas in mostempirical subfields of the discipline eversince.

The method introduced gives accurateestimates as well as reliable assessmentsof the uncertainty of all inferences. It isrobust to numerous data problems suchas high levels of aggregation bias. Be-cause the ecological inference problemis caused by the lack of individual-levelinformation, all methods of ecologicalinference, including those introduced inthe book, will always entail some risk.These risks are minimized in the ap-proach taken by new models that incor-

porate far more available information,intuitive graphics and diagnostics toevaluate when assumptions need to bemodified, and easy methods of modify-ing the assumptions. To verify that themethod works in practice, I use a vari-ety of new data sets for which the true,individual-level answer is known. Thismakes possible over 16,000 compari-sons between estimates of individual-level relationships from aggregate dataand the known individual-level answer.(This compares to 49 such comparisonsin the history of research on this sub-ject.) The method works in practice.

ExampleThe goal of ecological inference in

this example is to fill in the unobservedcell entries given only the observed ag-gregate percentages from the columnand row totals in this table (from Loui-siana in 1990) and the correspondingtables from each of Louisiana’s 3,262precincts (i.e., without survey data). Forexample, the upper left cell entry is the(normally unobserved) percent of blacks

A Solution to the Ecological Inference Problem: Reconstructing Indi-vidual Behavior from Aggregate Data

Gary King, Harvard [email protected]

In Press, Princeton University Press, 1997


who voted in 1990. Since the cell entriesin this case are known from publicrecords, I can report that the true valueof this cell is 64%. The estimate fromthe method reported in the book, basedonly on aggregate data, is only a frac-tion of a percentage point under thismark. In this example, like numerousothers reported in the book, the methodgives accurate statewide estimates,which has been the goal of past research.

But the solution to the ecological in-ference problem turns out to providemuch more interesting information thanaccurate statewide estimates. It also pro-vides, in these data for example, accu-rate estimates of the fraction of blackswho vote (and fraction of whites whovote) for each of Louisiana’s 3,262 elec-toral precincts. For example, the follow-ing figure compares estimates from ag-gregate data of the percent of blacks whovote to the true percent of blacks whovote in all Louisiana precincts (witheach precinct represented by a circlesized proportional to its black popula-tion). The center of almost all the circlesfalls on or near the diagonal line, indi-cating that the estimated percent ofblacks voting is very close to the trueindividual-level percentage. This figureis not merely a plot of the observed val-ues of a variable by the fitted values ofthe same variable used during the esti-mation procedure: it is instead a muchharder test because the true fractions ofblacks voting (the vertical dimension inthe figure) were not used during the es-timation procedure.

Here’s to you, Mr. RobinsonNearly half a century ago, a single ar-

ticle by William Robinson (1950) di-verted the stream of academic research.By popularizing “the ecological fal-lacy,” greatly clarifying the ecologicalinference problem, and (appropriately)convincing generations of scholars thataggregate data should never be used(with methods then available) to inferinformation about individuals, he fun-damentally changed the nature of socialscience research. Vibrant fields of schol-arship that relied on aggregate data with-ered. Once-important traditions of po-

litical geography in France, Germany,and the U.S. largely collapsed. Thescholars who continue to work in someof these fields – such as those attempt-ing to explain who voted for the NaziParty, or which social groups supporteach political party in the newly emerg-ing democracies – do so because of thelack of an alternative to ecological data,but they toil under a cloud of great sus-picion.

Research based on aggregate data wassucceeded at mid-century with the thenemerging methodology of survey re-search. Surveys have not only taught usa great deal; they may well represent thesingle most important methodologicalcontribution of the social sciences of thisera. Because better data beat more so-phisticated statistics every time, the eco-logical inference problem does not ariseif accurate survey data are available.(Although in almost no cases is exist-ing survey data sufficient to obviate eco-logical inferences at local geographiclevels.) Yet, an exclusive focus on avail-able surveys has had some unavoidable

costs. Surveys necessarily force analyststo study recent periods, and thus to misslong-term trends and some large-scalepatterns. Moreover, when we choose oursubject to study based on available sur-vey data, we study random collectionsof isolated individuals from essentiallyunknown geographic locations. We thusnecessarily lack the ability to gatherknowledge about local communities,contextual effects, or geographic pat-terns. Creative combinations of quanti-tative and qualitative research are alsomuch more difficult when the identityof and rich qualitative information aboutindividual respondents cannot be re-vealed to readers. Indeed, in most cases,respondents’ identities are not evenknown to the data analyst. If “all poli-tics is local,” political science is miss-ing much of politics.

In contrast, aggregate data analysiscan almost always be based on exten-sive, detailed, and local, geographic pat-terns; it can extend over very long timeframes; and it can be supplemented withqualitative information at any level of

Estimated Proportion of Blacks Voting

Tru

e Pr

opor

tion

of B

lack

s V

otin

g

0 .25 .5 .75 1

0

.25

.5

.75

1


richness or detail. In fact, with a work-ing method of ecological inference,there is little reason for any division be-tween quantitative and qualitative ap-proaches to a problem. Systematic quan-titative analyses can be easily joinedwith richer qualitative data at the localgeographic level and any study will beimproved as a result. Going into the fieldand visiting villages and communitiesfrom which data are available becomesa central task once again. In fact, if somelimited survey data are available, it toocan be used with the method proposedto improve ecological inferences.

Moreover, the quantity and quality ofunanalyzed aggregate data sets waitingaround for enterprising political scien-tists to take notice is staggering. In theU.S. alone, there are political data ondozens of electoral offices and 3000+census variables available from each of190,000 electoral precincts. Most ofthese data have been untouched by so-cial scientific hands. The situation is notmuch different in most other parts of theworld. When the Berlin Wall fell, schol-ars of Eastern Europe and the formerSoviet Union found that behind it allthese years was an ocean of data, andthose standing nearby were inundatedwhen the flood began. With these andother massive unanalyzed aggregatedata sets pouring in from all over theworld, and a solution to the ecologicalinference problem, much opportunity re-mains.

References

Ogburn, William F. and Inez Goltra.1919. “How Women Vote: A Study ofan Election in Portland, Oregon,” Po-litical Science Quarterly, 3, XXXIV:413–433.

Robinson, William S. 1950. “Ecologi-cal Correlation and the Behavior of In-dividuals,” American Sociological Re-view, 15: 351–57.

1 The name ‘ecological data’’ dates atleast to the late 1800’s. It stems fromthe word ecology, which is the scienceof the interrelationship of living thingsand their environments. Statistical mea-

sures taken at the level of the environ-ment, such as summaries of geographicareas or other aggregate units, are knownas ecological data. Ecological inference

Transformation”Hieronim Kubiak - “Hopes, Illusions

and Deceptions: Half a Century ofPolitical Sociology in Poland”

Dipankar Gupta - “Engaging withEvents: The Specifics of PoliticalSociology in India”

Habibul Haque Khondker - “Sociol-ogy of Political Sociology in South-east Asia and the Problem of De-mocracy”

Elisa P. Reis - “Political Sociology inBrazil: Making Sense of History”

Anthony M. Orum - “Almost a HalfCentury of Political Sociology:Trend in the United States”

Baruch Kimmerling - “ChangingMeanings and Boundaries of ‘Po-litical’: Some Conclusions”

is the process of drawing inferencesfrom these ecological data about the in-dividuals within the aggregates or areas.

(News & Notes continued from page 17)

At the end of 1996 Current Sociologypublished a special issue entitled “Po-litical Sociology at the Crossroads", ed-ited by Baruch Kimmerling. The volumeincludes:

Erik Allardt - “Is There a Scandina-vian Political Sociology Today?”

Birgitta Nedelmann - “Between Na-tional Socialism and Real Social-ism: Political Sociology in the Fed-eral Republic of Germany”

Christopher Rootes - “Political Soci-ology in Britain: Survey of the Lit-erature and the Profession”

Victor Voronkov and ElenaZdravomyslova - “Becoming Politi-cal Sociology in Russia and Russian

News & Notes

Book Reviewers Welcome

Doctoral students at any institution are welcome to submitproposals for book reviews. To do so, contact the editorand/or assistant editor with the name of the book you wishto review and a short (approximately 200 word) discussionof the book’s importance and why you wish to review it.Books must have been published no more than two yearspreviously. We will let you know if the title is still availablefor review, as well as the deadline for the next issue of theNewsletter. Reviewers are responsible for procuring theirown copies of books. Reviews should run 1,000 to 1,500words, must be submitted electronically, and must be inEnglish.


IntroductionThe central question is whether democ-

racy in the political realm fosters or hin-ders economic development.

We first dichotomize political regimesas democracies and dictatorships and ask:under which regime are economies morelikely to develop? Which is more likelyto generate miracles and which disasters?Which is more likely to assure that thecountry does not go backwards, that itsdevelopment is sustained? Which is moreapt to exploit advantageous conditionsand which is more adept at coping withadversities? Then we delve deeper, dis-tinguishing different types of democra-cies and dictatorships, and posing a newset of questions: can the observed differ-ences in performance be explained by theparticular institutional arrangements thatdifferentiate democratic systems? Canwe predict which democracies will de-velop and which stagnate? Can we tellex ante whether a dictator will turn outto be a statesman or a crook? Are theresome observable features of dictatorshipsthat allow us to predict their perfor-mance?

Yet even if our question concerns theimpact of political regimes on economicdevelopment, for methodological reasonswe must first examine whether economicdevelopment affects the rise and fall ofpolitical regimes. The problem faced inany analysis of an impact of institutions,policies, or programs on performance isthat if some factors, observed or not bythe researchers, affect both the choice ofinstitutions and the performance, then in-

ferences based on the observed sampleare invalid. A bias may emerge if verypoor countries tend to have dictatorialregimes and to stagnate (exogenous se-lection on observables), or if democra-cies are less likely than dictatorships tosurvive economic crises (endogenousselection on observables), or if countriesendowed with enlightened leaders tendto have democratic regimes and to de-velop (selection on unobservables). Un-der such conditions, we must study therelation between institutions and perfor-mance as simultaneous, using models ap-propriate to the mechanisms of regimeselection. Indeed, one impetus for thisbook is that the numerous previous stud-ies of the relation between political re-gimes and economic performance all failto consider selection bias.

This methodology underlies the orga-nization of the book. Even though ourprimary question concerns the impact ofdemocracy on development, in order tostudy this question we must first learnhow countries happen to have particularregimes – the impact of development ondemocracy. Thus, part one is devoted tothe analysis of the impact of developmenton the rise and fall of political regimes,while part two examines the impact ofregimes on development.

The book is written at three levels: forthose who are interested only in what theworld is like, for those who want to knowwhy we think that it is this way, and forthose who wish to delve into methods ofanalysis. More important findings aresummarized in an extensive preview that

opens the book. This part is self-containedand it includes the conclusions. The mainbody of the book presents the argumentsand the statistical analyses upon whichthese conclusions are based. Finally, tech-nical materials are presented in the appen-dices to the particular chapters and to thebook as a whole.

The analysis is based on 135 countries(we exclude six countries which derivemore than one-half of their revenues fromoil) observed between 1950 or the yearof independence or the first year forwhich the data are available (“entryyear”) and 1990 or the last year for whichdata are available (“exit year”). Our unitof analysis is a particular country duringa particular year, for a total of 4,318country-years, of which economic dataare typically available for 4,126 obser-vations.

Part One: Development and DemocracyObviously, our first task is to define

democracy and dictatorship, distinguishvarious types of both, and classify theobserved regimes into these categories.We treat as democracies regimes whichhold elections in which the oppositionhas some chance to win and to assumeoffice. Our treatment of dictatorships isbasically residual: all regimes that fail tobe qualified as democracies are consid-ered to be dictatorships or, a term we useinterchangeably, authoritarian regimes.Note that if different dictatorships suc-ceed one another, we treat them as a con-tinuous spell of authoritarianism.

Whatever are the peculiarities of ourrules, the resulting classification differs

Democracy and Development:Political Regimes and Economic Performance, 1950-1990

Adam Przeworski, New York [email protected]

Mike Alvarez, De Paul University, ChicagoJosé Antonio Cheibub, University of Pennsylvania

[email protected] Limongi, University of Sao Paulo

[email protected]

Forthcoming, Cambridge University Press


little from alternative approaches. Hence,there is no reason to think that our re-sults are idiosyncratic to the particularclassification of regimes.

We also distinguish three types of de-mocracies: parliamentary, mixed (semi-presidential), and presidential – as wellas different types of dictatorships: mainlyinstitutionalized ones, which we call “bu-reaucracies,” and the non-institutional-ized “autocracies.”

Once the regimes have been classified,we need to understand how they comeabout and die. Using a dynamic probitmodel, we discover that democracy ismore likely to survive in wealthier coun-tries, while dictatorships endure in poorcountries, are less stable in countries withmiddle income levels, and somewhatmore stable again in wealthier countries.Perhaps the most startling discovery isthat no democracy ever fell, during theperiod under our scrutiny or ever before,in a country with a per capita incomehigher than that of Argentina in 1976. Inturn, the level of development alone doesnot predict a single transition to democ-racy. Hence, while Lipset was correct toargue that democracies are more stablein the more affluent countries, the cen-tral hypothesis of modernization theory– that development under dictatorshipbreeds democracy – is false. The ob-served cross-sectional pattern – the factthat democracies are much more frequentin wealthy countries – is thus the resultof the process in which dictatorships diealmost at random, but if they happen todie in a wealthier country, democracy isalmost certain to survive forever. Thesefindings offer evidence that regime se-lection depends on the level of economicdevelopment.

This analysis is then replicated withregard to the impact of the rates of eco-nomic growth. We find that democraciesare more vulnerable to economic crisesthan dictatorships. Moreover, all this in-stability is limited to poor democracies,which are extremely brittle when theyface economic crises. Hence, we find evi-dence for endogenous selection of re-gimes, conditional on economic growth.

The same analysis is then conductedwith regard to different types of democ-

racies and dictatorships. The central find-ing is that parliamentary democracies aremuch more stable than presidential ones.We wonder why so many new democra-cies choose presidential regimes, and findsome evidence that they constitute alegacy of military dictatorships. We ar-gue that the choice of presidentialism isdue to the threat posed by the military.

We conclude this part by summariz-ing the observed patterns of regime se-lection and then speculate about mecha-nisms of selection which we are unableto observe: in particular, the quality ofthe political leadership. The models ofselection summarized above are used togenerate a statistical instrument for theprobability of particular regimes, and thisinstrument is then used in the analysis ofthe impact of regimes on development.

Part Two: Democracy and DevelopmentThis part has a similar structure. We

begin with a conceptual discussion of thedependent variable – development – andrelate it to our measures of economic per-formance: the growth of income and ofconsumption, the share of physical in-vestment in gross income, the growth ofthe labor force and of human capital.Then we present a general outline of thehistorical patterns of growth and in theirlight examine the economic models ofgrowth. Our purpose at this stage is toconstruct a robust specification of thegrowth model, a specification to be usedin assessing the impact of regimes. Sincewe are concerned that the variables usedto explain growth should not be them-selves endogenous to regimes, we endwith a specification that is minimalist butwhich does include the standard pano-ply of production-function variables.

Using a selection-augmented model,we then proceed to examine the impactof democracy and dictatorship on eco-nomic performance. We learn that, con-trary to the most cherished beliefs, theseregimes invest at an almost identical rate.Indeed, in very poor countries the shareof investment is higher under democracythan under dictatorship. In turn, to oursurprise, we discover that the rate ofgrowth of labor force (and of population)is much higher under dictatorship. Fi-

nally, and most importantly, we find thatthe selection-corrected average rates ofgrowth of per capita income and of percapita consumption are not different un-der the two regimes. This finding is ro-bust under various specifications and dif-ferent estimators. It survives controls forthe quality of the economic data and ex-periments with a different classificationof regimes.

Yet to say that regimes do not differ intheir average performance is not to saythey do not matter. While we find thatthe allocative efficiency of investment isabout the same under the two regimes,the elasticity of output with regard to la-bor, which under competitive conditionsequals the labor share, is much higher indemocracies. Hence, dictatorships usemore labor, get less out of it, and pay itless: their growth is more labor-intensivethan that of democracies.

A comparison of the performance ofparliamentary and presidential democra-cies generates overwhelming evidence infavor of the former. We attribute this dif-ference to the mechanisms of political ac-countability characteristic of these twotypes of democracy. We discover that thegreat majority of presidents who peace-fully left office did so because of impend-ing term limits, while among those whoran for re-election few were defeated. Weargue that presidential regimes give anexcessive advantage to incumbents –who are simultaneously heads of govern-ment, of state, and typically of the armedforces – and then counteract this advan-tage by constitutional rules limiting re-election, thus depriving the voters of theelectoral mechanism to control chief ex-ecutives.

Since we are still analyzing the impactof the types of dictatorships, it is too earlyto report these results. Our main ques-tion is whether the fact that a dictator-ship is institutionalized – namely, that isformally organized and rules by laws –affects economic performance. Thus farour main surprise is that the survival ofdictators in power is sensitive to eco-nomic performance. We argue that elec-tions do matter under dictatorships; eventhough they are not a mechanism for se-lecting rulers, once elections are held,


dictators are concerned that they demon-strate a show of strength for the regime.They care whether the turnout is 90 or95 percent, since they fear to be over-thrown if they do not show themselvesto be fully in control. Whether this mecha-nism of accountability has consequencesfor economic performance, we do not yetknow.

We close the analysis of the impact ofregimes on development by building andtesting a model in which this impact ismediated by the size of the government,specifically of the public productive ex-penditures. Drawing on economic mod-els of the optimal size of government,we argue that their actual size should betoo small under autocracy, too large un-der bureaucracy, and closer to optimalunder democracy, and that the size ofgovernment should have an impact on therates of growth. The statistical analysisis still in the process.

ConclusionThe hypothesis that emerged in the late

1950’s and dominated U.S. foreignpolicy over two decades was that we facea trade-off between democracy and de-velopment. Dictatorships were needed togenerate development. Yet the future wasnot so bleak for democracy, since dicta-torships would self-destruct as the resultof their own success. According to thedominant canon of the time, democracywould naturally emerge after a societyhad undergone necessary economic andsocial transformations.

Since in this view dictatorships gener-ate development while developmentleads to democracy, the best way to de-mocracy was seen as a circuitous one.The policy prescriptions that resultedfrom this mode of thinking rationalizedsupporting dictatorships, at least thosethat were “capable of change,” that is,anti-communist ones.

Communism is now dead, and the ideathat it ever represented the future appearsludicrous, albeit in the omniscient retro-spect. Yet doubts remain. For many,Pinochet’s Chile is the paradigm of suc-cessful economic reforms; the economicsuccess of authoritarian China is themodel for Russia. Even if democratic ide-

als nourish political forces from Argen-tina to Mongolia, the allure of a “stronggovernment,” “insulated from pres-sures,” guided by technical rationality,capable of imposing order and discipline,continues to seduce. Whether in the caseof the Tienanmen Square massacre or theautogolpe of the Peruvian PresidentAlberto Fujimori, international financialinstitutions, as well as governments ofdeveloped countries, are still willing toclose their eyes at violations of demo-cratic, and even human, rights on behalfof the purported economic effectivenessof dictatorships. Thus, the question of therelative economic merits of political re-gimes continues to evoke political as wellas intellectual passions.

Our results should dispel such doubts.True, the “tigers” tend to be dictatorships:the fastest growing country in the worldin the 1950’s, at least if we are to believeits own statistics, was Romania; the eco-nomic miracle of the early 1970’s wasthe military-ruled Brazil; the heroes ofthe 1980’s were the dictatorships ofSingapore, South Korea, and Taiwan; inthe 1990’s, it is China. But are dictator-ships the tigers? The list of economicdisasters generated by authoritarianismis long and tragic. Even the economiccollapse of communism pales in com-parison with the destruction caused bydictatorships in many African and LatinAmerican countries and the squanderingof resources in the Middle East. Hence,we must compare the average, not thebest, practice.

There is just no evidence that, on theaverage, dictatorships are better at gen-erating growth than democracies. More-over, when dictatorships generate devel-opment, they tend to do it by exploitingcheap labor. Once different types of de-mocracy are distinguished, it becomesclear that parliamentary democracies arebetter at generating development thanother political regimes, including presi-dential democracies but also differenttypes of dictatorships.

Neither is there evidence that economicdevelopment under dictatorships breedsdemocracy. In those few countries thatdid develop under dictatorship, the re-gime survived well into levels of devel-

opment under which other countries havelong enjoyed stable democracies. Thelevel of economic development just failsto predict transitions to democracy, whileex post explanations of development-driven transitions entail a logical fallacy.

Hence, to summarize these results in apositive tone, we end arguing that to getdemocracy we should support democ-racy, not dictatorship. And to the extentto which political regimes matter for eco-nomic development, it is clear that par-liamentary democracies dominate all thealternatives.

Preliminary Results Already Available

Alvarez, Mike, and Adam Przeworski.1995. “Parliamentarism andPresidentialism: Which Lasts? WhichWorks?” In Antoni Sulek and JozefStyk, eds., Ludzie i Instytucje. Lublin:Wydawnictwo Universytetu Marii Cu-rie-Sklodowskiej.

Alvarez, Mike, José Antonio Cheibub,Fernando Limongi, and AdamPrzeworski. Forthcoming in 1996.“Classifying Political Regimes.” Com-parative Studies in International De-velopment.

Cheibub, José Antonio, and AdamPrzeworski. Forthcoming in 1997. “AnEconometric Evaluation of the Impactof Government Expenditures on Eco-nomic Growth.” Co-author. In AlbertBreton, Pierre Salmon, and RonaldWintrobe, eds., Democracy. NewYork: Cambridge University Press.

Cheibub, José Antonio, and AdamPrzeworski. Forthcoming. “Democ-racy, Elections, and Accountability forEconomic Outcomes.” In BernardManin, Adam Przeworski, and SusanC. Stokes, eds., Democracy and Ac-countability. New York: CambridgeUniversity Press.

Przeworski, Adam, and FernandoLimongi. 1993. “Political Regimes andEconomic Growth.” Journal of Eco-nomic Perspectives 7: 51-69. Re-printed in Amiya Bagchi, ed., Democ-racy and Development. Oxford: Ox-ford University Press, 1995.


Data, Archiving and theReal World of Com-parative ScholarshipDavid D. Laitin


After six years of field research andwriting, my new book manuscript Iden-tities in Formation: The Russian-Speak-ing Population in the Near Abroad isbeing prepared for publication at CornellUniversity Press. This project has crys-tallized for me some of the issues con-cerning the replication norm debate asit affects the comparative politics com-munity. As an instigator of that debateon these pages, I sought to promote anorm that was both reflective of thepractical realities of our subdisciplineand the requirements of a scientific com-munity. In this discussion, I shall reviewsome of the practical realities, in orderbetter to specify a reasonable norm.

I had written, in my role as presidentof this section, a commentary in supportof the development of a strong norm onthe archiving of data amongcomparativists in order to promote rep-lication as a standard part of our scien-tific repertoire. Some of our most hon-ored members (for example, Ian Lustick,Sidney Tarrow and Robert Putnam), inreaction, raised important points againsta norm that called for the archiving ofvirtually all field data. They were con-cerned, for example, about the confiden-tiality of sources, and how this mightbe violated unwittingly by replicators.They were concerned as well for therights of the collector of the data to theinitial interpretation of it. Since, in com-parative politics, the initial interpreta-tion, if done right, may take a long time,the collector of the data can easily loseout to “quick and dirty” interpretations,whose authors preemptively exploit thedata set. In response, at the general meet-ing of the section, I sought a compro-

mise that would inter alia allow for theexception of any data set that could com-promise any individual, and for the post-ponement of the dissemination of thedata until the collector of the data had areasonable time to publish his/her re-sults.

At that very time, I began to face therealities of the data gold mine that I (incollaboration with Jerry Hough) col-lected in the former Soviet Union. Thebook manuscript, Identities in Forma-tion, examines the political and culturalreactions of the Russian and Russian-speaking settlers that were “beached”outside of Russia proper when the So-viet Union receded. The empirical workcovers four republics: Estonia, Latvia,Ukraine, and Kazakhstan. The book isdata rich, and the data come from a widevariety of sources. To be specific, thebook reports on:

1. Large-n surveys administered toRussians and Titulars (titularmeaning the nationality group af-ter which the republic was named)in all four republics. About 150questions were asked and therewere from 1,100 to 2,300 respon-dents, depending on the populationsize of the republic (and the costsof the survey).

2. The “matched-guise” experiment,well-known in socio-linguistics asan unobtrusive measure of lan-guage status. In this experiment, re-spondents listen to a tape record-ing of several voices reading thesame text. Half the voices speakRussian; the other half speak thetitular language. Respondents arenot aware that each person speak-ing has read the passage in bothlanguages, and thereby has two“guises”. Respondents rank eachvoice on a number of attributes,and the researcher can comparescores on both guises of a singleindividual speaker. The matched-guise experiment was performed inseveral schools in each of the four

republics.3. A content analysis of newspaper

articles dealing with the national-ity issue in all four republics, writ-ten both in Russia and in the re-publics. In this analysis, the vari-ety of terms used to identify thebeached diaspora were coded on avariety of dimensions. There werefrom 73 to 88 articles coded fromeach republic, with nearly 2,000identity terms coded in the totalsample.

4. Ethnographic field notes collectedin all four republics. I put togethera research team in which eachmember spent months in the field,living with and interviewing fami-lies of both Russian and Titular na-tionality. The team of four (VelloPettai and I shared Estonia; Pettaialso did Latvia; Dominique Areldid Ukraine; Bhavna Dave didKazakhstan) coordinated on a setof participant observation tech-niques, and extended interviews towrite family biographies. Thesedata gave me a sense of the microelements of the Russian-speakersas they adjusted to their new rolesas minorities in the new republics.

5. Official documents, newspapers,electronic news sources, and sec-ondary sources. These materialshelp provide political narrative tothe events that occurred in these re-publics, and in the world.

Of all these data, only the large-n sur-veys have been properly archived at ISR,housed at the University of Michigan,Ann Arbor. In preparing these data forarchiving, I was sobered by a factorraised not fully elaborated by Lustick.This factor is time and expense. The foursurveys were written originally in Rus-sian, and local teams translated the sur-vey into the titular language. But the sur-veys had organic existences in each ofthe republics. We found that preciselysimilar wording had different meaningsin different republics, and to keep the

Data Sets & Archives


meaning of the question the same, wehad to change the wording. Also, withslightly different political situations, wecouldn’t ask the same questions in eachrepublic. Allow me to give a simple ex-ample, partly simplified for purposes ofillustration. We wanted to know whetherrespondents were sympathetic or not tothe use of the titular language as a me-dium of instruction in the schools. Butwe wanted to make the question con-crete. If the titular language was usedas a medium of instruction in fewschools, we were able to ask “do youthink the titular language should be usedas a medium of instruction in mostschools?” But if it is already used as amedium of instruction in most schools,we changed the question to “Do youthink it is a good thing that the titularlanguage is the medium of instructionin most schools?” In sum, the surveyswere not precisely the same.

Getting the four surveys translatedproperly into English, and in a way thatall future researchers will immediatelysee the wording (or choice set) differ-ences on each survey was a vast enter-prise. Of course, I could have depositedthe surveys only in Russian; but thatwould have constrained replication tothose who read Russian or who couldpay for translation. I felt this to be anunnecessary restriction for data thatwould have general appeal. But morethan translation is involved. Since thesurvey teams were not always the same,in some cases different protocols wereused in sampling. All of these sensitiveissues were resolved as best I could andnoted in a wide variety of files and note-books. To get all this information clearlywritten, and for deposit into an archive,I employed two students for about fivemonths in an exercise that had no use tome, but was essential for archiving ifreplication is to mean anything.

There were other tasks that had to bedone to make the data set useful. I con-structed several indices that played acrucial role in the specification of thevariables for the book. As the projectdeveloped, and my research assistantsperformed tests to see if the elements of

the indices were cross-pressuring oneanother. In some, they were. Each timeI did re-codes for new specifications ofindices, I had to revise the package I hadprepared for the archive. Thus, thearchived data were held up until the lastrevision was done, or else replicatorswould not have access to the key vari-ables used in the book.

The NSF – which funded the surveys– demanded archiving these data. Theywere so important for future research-ers that I had no qualms about spendingthe time and money to do it right. Butthe difference between orderliness of thedata that is needed in order to write abook and orderliness that is needed forother scholars to understand the data setis enormous. NSF panels, Gary Kingtells me, have begun – rightly, in myjudgment – to demand a proper budgetfor the fulfillment of that data archivingdemands they set. If political scientiststhrough their contacts helped make it anorm for all public and private grantingagencies to make the same provisions,the quality of our scientific communitywould be enhanced.

The matched-guise and the contentanalysis data cry out for replication, butI have not (as yet) prepared the data forarchiving. (NSF did not require thearchiving of the matched-guise tests; thecontent analysis was supported by othersources). The likelihood, however, ofreplication for these extremely focusedtests is quite small. The expected returnsfor archiving the data are therefore quitelow. It follows, I think, that it would beproper to make known that any schol-ars who wish to replicate these data arewelcome to contact the author, who willcooperate in all ways to aid the replicatorunderstand the data set. This is compa-rable in biology and chemistry to invit-ing colleagues to one’s lab in order toshow how a new experimental techniqueis done. This kind of collegiality is farcheaper than would be fulfilling a bu-reaucratically-set norm requiring thepublic archiving of all data.

The ethnographic data are often quitesensitive. To “clean” them up wouldtake a great amount of effort. Again, I

would give access to any scholar whoestablished his or her bona fides to thosedata, if they promised to assure confi-dentiality of sources. (Also, the asidesby the field researchers telling their re-actions are a great source of material;but it is questionable whether we shouldpermit replicators to quote our semipri-vate comments on the margins of ourfield notes into their publications. Thismakes me queasy).

Perhaps it could be argued (as MiriamGolden did in her article on replicationin these pages, and more forcefully in aprivate communication in response to anearlier draft of this comment) that fieldresearchers who run off into the field,do some ad hoc open-ended interviews,and don’t enter their notes into computerfiles, are still living in a pre-scientificage. Once scholars have made the psy-chological leap into the replication age,she suggests, the whole research processwill be affected, as every step in it willbe open for scrutiny by those who fol-low the original researcher in the field.She recognizes that with this norm inplace there will be increasing pressureson comparativists to assemble relativelylarger, more standardized data sets. Shefurther recognizes that this sort of ratio-nalization of the data assembly processwill work against the solitary researcher,and give us incentives to rely on datathat are more easily quantified and thathave built-in economies of scale. Overtime, she concludes, this will change thenature of field research, and clearlydowngrade traditional area studies workin the comparative field.

I have sympathy for Golden’s analy-sis, but I am reluctant to accept it fully.If we systematically exclude the expe-riences many comparativists have hadin the field from the enterprise called“science” – background materialspicked up in places that we don’t wantthe world knowing that we inhabited,with people who may one day be in jail– our research intuitions throughout ourcareers about what is really going onwould be severely dulled. The advan-tages of keeping our two ears to theground to catch undercurrents of politi-


cal feeling – a point I emphasized in anessay I wrote in these pages on fieldwork – would be lessened if we felt thebreath of future researchers on ournecks. The plausible response, that wekeep two books – one cleansed, and oneprivate notes to ourselves – would goagainst the spirit of the replication norm.

For small, specialized data sets suchas in the matched guise and contentanalysis and for field notes as well, Icannot therefore recommend that re-searchers be required to archive theirdata. An alternative idea, brought to myattention by Gary King, is that insteadof depositing data in some centralarchive, comparativists would only reg-ister the existence of the data. The de-scription of the contents of data wouldbe publicized and included in a data baseavailable to all comparativists. If some-one wishes access to those data, theywould have to contact the researcher.That way, the comparativist who col-lected the data can clear confidentialityissues with people individually, and con-dition what they give to others on theresearch needs of those asking for ac-cess to the data. In recent years eco-nomic historians (as part of the Eco-nomic History Net Database Registry)have normalized this practice, and itmight be a model for us.

Finally, there are the documents andsecondary sources. As Lustick arguedin his response to the proposed replica-tion norm, comparativists use footnotesto aid replicators. For the secondarysources, documents, and electronicsources, I footnote sufficiently so thatfuture researchers can check the degreeto which my sources make claims that Iattribute to them. I would, however, ac-cept a rule that crucial documentationgarnered in the field, but not accessibleunder normal conditions within the sci-entific community, ought to be scanned(if legally permitted), and put onto thenet for use by the larger scholarly com-munity.

In sum, the experience of writing Iden-tity in Formation has given me a cleareridea of the type of differentiated repli-cation norm that is appropriate for our

sub-discipline. Certain kinds of data(such as large-n surveys) ought to bearchived no matter what the cost. Ofcourse, we must teach our fundingsources that they need to fund this cru-cial aspect of scientific practice. Otherkinds – more localized data sets – per-haps can be archived “in lab” open toother researchers, with an announce-ment of its nature in on-line disciplin-ary registries, but without needing thehigh up-front costs of explaining howto use the data for all potentialreplicators. Field notes may not be ap-propriate for archiving, even if theycould be cheaply scanned, and it is jus-tifiable to require restrictions to theiraccess by other scholars. The existenceof these data, however, should also beregistered. Secondary sources and the

privatization, a process in which the au-thors were all key participants1 . Their fo-cus is on the voucher privatization pro-gram in Russia, begun in late 1992 andcompleted in the summer of 1994. Theysuggest that privatization succeeded be-cause the primary objective of the pro-gram was to depoliticize economic activ-ity - to sever the links between managersand politicians2 .

The authors present a detailed explana-tion of how political control of economicactivity in Russia (and the USSR) led topoorly defined property rights, which inturn led to poor economic performance.Politicians used political control of pub-lic enterprises to overemploy people, payexcess wages, locate factories in areaswhere it is inefficient to produce, etc. Thefirst objective of privatization should beto free firms from politicians control; is-sues such as effective corporate gover-nance should be dealt with later.Privatization severs the links betweenpoliticians and managers, thereby robbingpoliticians of control over firms. In orderto accomplish this, control and cash flowrights must reside in the hands of enter-prise managers and outside investors. Fortactical reasons, other stakeholders infirms, namely the workers and local gov-

Privatizing RussiaMaxim Boycko, Andrei Shleiferand Robert VishnyMIT Press, Cambridge, 1995

Reviewed by Preston KeatUniversity of California, Los [email protected]

While numerous policy makers andscholars have explained the rather obvi-ous economic rationale for privatizingstate-owned enterprises (SOE’s) in for-merly communist countries, very fewhave successfully addressed the acutepolitical problems associated with thisprocess. Under what circumstances willmanagers, workers, and the general popu-lation embrace, or at least accept, a pro-gram of rapid privatization? In addition,what mechanisms are most effective inforcing politicians to relinquish controlover SOE’s? Privatizing Russia offersseveral important new insights into thesequestions, which have implications for thereform process not only in Russia andeastern Europe, but for a whole range ofcountries that are attempting to reform thestate sector of their economies.

The book tells the story of Russian

Book Reviews

like already archived in libraries requireonly footnotes; but it would be a usefulnorm requiring us to scan documentsthat are not accessible to other scholars.

Essential to the replication norm, andyet to be authorized by our section, isthat all books and articles in compara-tive politics should have a replicationfootnote, in which the author specifiesthe conditions under which the data canbe accessed by replicators. The justifi-cation of those choices should be as-sessed by peer reviewers just as theywould do for the data analysis itself.Community norms as they evolve, pre-sumably different in distinct subfields,rather than some predefined scientificstandard, would then define the criteriafor an acceptable replication footnote.


ernments, must also retain some controlrights.

They argue that a combination ofcorporatization (the allocation to firmmanagers of control rights that once be-longed to politicians) and privatization(allocation of cash flow rights to outsideinvestors) is the only politically viablestrategy for creating an efficient owner-ship structure. Effective privatizationstrategies thus require the building of po-litical coalitions that remove control rightsfrom traditional (non-reformist) politi-cians. In Russia this coalition includedinsiders such as firms’ managers andworkers, and outsiders such as the gen-eral public and local governments.

The authors describe how the Russianprivatization program fits this theoreticalsetup. Through a program of guaranteedinsider rights (managers and workers wereoffered a controlling stake in their indi-vidual firms) and public provisions(privatization vouchers were distributedto all citizens for a nominal fee and con-trol of much of the process was devolvedto the local government level), reformistpoliticians were able to develop a winningcoalition of managers, workers, local gov-ernments, and the general population. Theprivatization process is thus depicted as apolitical battle wherein reformist politi-cians try to establish a coalition that hasenough political power to effectively chal-lenge entrenched politicians.

Politicians receive tangible benefitsfrom public ownership - they can hirepolitical allies; elicit bribes; gain politicalsupport and votes from managers, work-ers and local community members; con-trol certain prices for political ends; andso forth. Their argument for why en-trenched politicians resist reform, and whyfirms should be de-politicized is thus clear.However, it appears that they are focus-ing primarily on national-level politicians.Would not non-reformist local politicianshave the same incentives to retain controlof SOE’s? This local dimension of thestory remains undeveloped, as local au-thorities are granted certain key controlrights in the privatization process with-out mention of the possible negative rami-fications.

Boycko, Shleifer, and Vishny are justlyproud of the program they helped to con-ceive and implement. Privatization is,

however, only one piece of a much largereconomic reform puzzle. Private owner-ship of firms in no way ensures that deci-sions regarding critical issues such as gov-ernment subsidies and investment andbankruptcy enforcement will be made inaccordance with reformist economiclogic. They do suggest that these and otherissues such as overall restructuring offirms, corporate governance, and the le-gal environment are important, but do notgo very far to integrate them into theirstory.

In the midst of a large and growing lit-erature on economic reform in Russia andEastern Europe, Privatizing Russia standsout as perhaps the best example of a meld-ing of solid theory with descriptive em-pirical work. In addition to telling a fasci-nating story, the authors have made animportant contribution to the relativelyrecent theoretical literature focusing onthe interests of politicians in the processof enterprise reform. They have also gonea long way toward answering a numberof key questions with important implica-tions for both the practice and compara-tive study of economic reform.

1 As advisors to the Russian government,they were active participants in both thedevelopment and implementation of theprivatization program.2 This suggestion is derived from previ-ous articles; in particular Schleifer andVishny. 1994. “Politicians and Firms.”Quarterly Journal of Economics, CIX:995-1025.

Challenging the State:Crisis & Innovation inLatin America & AfricaMerilee S. GrindleCambridge University Press,Cambridge, 1996

Reviewed by Tom LewisUniversity of [email protected]

Continuing the state-society analysisthat has marked her earlier work, MerileeGrindle undertakes an ambitious and re-vealing cross national comparison ofneoliberal market reforms and subsequentpolitical ramifications in sixteen develop-

ing nations from Africa and Latin America,with detailed case studies of Mexico andKenya. The literature on neoliberal reforms– economic prescriptions that drasticallydiminish the state’s presence in theeconomy – typically examines the inputside of the reforms, covering technicalissues about the shape and scope of re-form policy as well as the proper staterole in building a coalition to support re-form. Grindle, however, delves into theoutput side of the reform process to un-cover how reforms affect the state’s abil-ity to establish and maintain internal andexternal security, raise revenue, and main-tain dominance over alternative social or-ganizations. Her discovery is twofold:first, long-standing relationships amongthe state, society, and the economy havebeen destroyed by reform; and “the medi-cine applied to correct economic imbal-ances often contributed to weakening theability of the government to carry out andsustain needed reforms” (127).

During the 1980’s, massive budget defi-cits, the necessity of attracting foreigninvestment, the increasingly transnationalcharacter of capitalism, and requirementsof IMF and World Bank loan programsprecipitated a move in the developing na-tions of Latin America and Africa fromstate-led development strategies to poli-cies that placed greater emphasis on mar-ket forces to generate economic growth.Much like Samuel Huntington in Politi-cal Order in Changing Societies, Grindleassumes that strong institutions are nec-essary for determined policy implemen-tation: “Regimes attempt to negotiate andimpose formal and informal rules abouthow the state will relate to the economyand to society; durable and legitimate re-gimes have greater capacity to achievethese goals than do those that are less in-stitutionalized”(4). To gauge statestrength, Grindle employs a longitudinalstudy of four factors. The first, institu-tional capacity, deals with the state’spower to assert primacy in setting authori-tative rules governing economic and po-litical interactions. A state with great tech-nical capacity can manage effective mac-roeconomic policies. Administrative ca-pacity allows a state to deliver basic physi-cal and social infrastructure, a functionessential to economic development andsocial welfare. The final factor, political


capacity, involves the sufficiency and le-gitimacy of channels for societal demandmaking, representation, and conflict reso-lution. The case studies of Mexico andKenya in particular provide a fascinatingillustration of how these factors havewaxed and waned. Overall, Grindle con-cludes that as these states have imple-mented neoliberal reforms, their strengthhas fallen, jeopardizing the long-termchances for successful development.

Why have reforms had this paradoxi-cal effect? The answer involves the rela-tionships that neoliberal reforms have sev-ered. As Grindle recognizes, most LatinAmerican and African regimes dependedupon resource allocation to maintain amelange of relationships between the stateand society that underpinned social sta-bility and political support (155).Neoliberal reforms that diminish statecontrol over economic resources have dis-rupted this mélange of relationships; thus,the four dimensions of state strength de-cline. Once this happens, state leaders findthemselves in a situation similar to thatdescribed by Joel Migdal in Strong Soci-eties & Weak States and Barbara Geddesin Politician’s Dilemma: they must bal-ance long-term economic reforms with theneed for political stability. With statestrength diminished, leaders have gener-ally undermined their own reforms toplease societal groups whose support isnecessary for political stability, so thelong-term chances for successful devel-opment are sacrificed for political expe-diency.

Despite this pessimistic assessment ofthe reform process, Grindle declares thatall of this bodes well for more participa-tory politics in the largely authoritariannations in Africa and Latin America. Withfalling institutional, administrative, andpolitical capacity, states face a much morerestive society that contests the basic rulesof the game and demands economic re-dress. Such pressure from civil societyoften leads a state that now controls fewereconomic resources to allow a politicalopening, essentially trading more popu-lar control over the selection of local andnational leaders and more popular inputinto policy for political stability. Indeed,such an opening could be the only path tosuccessful development, for Grindle ar-gues that fully capable states may only

emerge where civil society is capable ofcontesting the state and its leaders, de-manding greater accountability and re-sponsiveness (194). So, though Grindleagrees with Huntington’s assessment thateconomic development can underminepolitical stability, she posits an oppositeremedy. More participatory political in-stitutions, rather than institutions insulatedfrom societal pressure as Huntington ar-gues, are the key.

Grindle offers an innovative account ofstate strength, and her comparison of six-teen nations from Africa and LatinAmerica is impressive and novel. How-ever, there is one inconsistency in the ar-gument. At times, Grindle asserts thatstrong political parties and well-institu-tionalized bureaucracies are needed topreserve state capacity and aid develop-ment (11). She also declares that power-ful resources resulting from the structureof political institutions, such aspresidentialism and centralized control inMexico, assist development (189-93).While strong political parties and bureau-cracies are consistent with the participa-tory politics that Grindle sees as an es-sential precondition for successful devel-opment, presidentialism and centralizedcontrol are not. This incongruence is puz-zling in an otherwise excellent analysis.

Internationalizationand Domestic PoliticsRobert O. Keohane and Helen V.Milner, editorsCambridge University PressCambridge, 1996

Reviewed by Erik WibbelsUniversity of New [email protected]

In few issue-areas has the division be-tween international relations and com-parative politics been more apparent thanin the study of political economy. WhileIPE specialists have examined trade, in-ternational capital flows and the like,comparativists have tended to focus oncomparative national case studies of eco-nomic crisis and development. The formerhave generally shunned systematic do-mestic political analysis, while the latter

frequently fail to address the internationaleconomic context.

Into this gap arrives the new edited vol-ume by Keohane and Milner entitled In-ternationalization and Domestic Politics.Grounded in the assumption that the in-creasing integration of domestic and in-ternational markets are raising the domes-tic political costs of ignoring the interna-tional economy, the authors focus on theeffects of internationalization on domes-tic politics. Generally speaking, they ar-gue that this can take place in three ways:by reshaping domestic political coalitionsand their policy preferences, by creatingeconomic and political crises, and by lim-iting governments’ capacity to manipulatemacroeconomic policy. Furthermore, do-mestic policy responses to internationaleconomic pressures, while viewed as re-active, are linked with the partisan make-up of governments, the organization oflabor and financial markets, and the na-ture of domestic political institutions.

The strongest feature of the volume arethe three theoretical chapters by Keohaneand Milner, Frieden and Rogowski, andGarrett and Lange respectively. The firstprovides a strong justification for analyz-ing the international economy and domes-tic politics jointly while forwarding anumber of testable hypotheses regardingthe international realm’s impact on do-mestic politics. The chapter by Friedenand Rogowski begins with the assump-tion of economic pluralism in which in-ternational prices determine domestic po-litical coalitions, but goes on to examinethe domestic political implications of threecompeting theories rooted in the factorand sectoral make-up of domestic econo-mies. Particularly successful, however, isthe theoretical chapter by Garrett andLange in which they explicitly account forthe mechanisms through which interna-tionalization impacts domestic policy out-comes by offering a model in which in-ternational economic impulses are filteredthrough domestic socio-economic andpolitical institutions. Their theoreticalmodel allows for domestic politics to in-fluence the process linking the internationaland domestic, and thereby avoids the deter-minism of economic pluralism while intro-ducing domestic agency as an importantvariable.

Somewhat less successful are the case

studies. With the exception of Garrett’s chap-ter examining the relationship between state-labor institutions and budget deficits, theremaining chapters generally stray from thetheoretical framework of the earlier chapters.Not systematically rooted in theories of ei-ther the international economy or domesticpolitics, the case studies tend to relay idio-syncratic accounts of the relationship be-tween internationalization and domestic poli-tics in the United States, Japan, the SovietUnion and China. Most interesting in thesestudies are the findings by Shirk andEvangelista that Soviet and Chinese institu-tions have been, and in the Chinese case arestill able to, severely limit the influence ofinternational economics on the shape ofdomestic political coalitions by blockingrelative changes in international prices.

More significantly, Shirk andEvangelista’s findings suggest a weaknessin the internationalization hypothesis thatinforms the volume: namely, that the rela-tionship between the international economyand domestic politics is unidirectional. Hag-gard and Maxfield note this in their chapteron LDC’s in which they suggest that “the

‘international’ variables that are at the cen-ter of this project...are themselves partly afunction of past government policy” (235).This suggests the need to further specify thedomestic institutional factors that providedefinition to the international/domesticnexus. Thus, while Keohane and Milnersuggest that institutions matter, and Garrettand Lange identify a number of domesticfactors such as regime type, the nature ofrepresentation, the number of veto points,and bureaucratic independence as relevantin differentiating government behavioracross nations, these factors are neveroperationalized and remain theoreticallymarginal to the project as a whole.

With their volume, Keohane and Milnerhave made impressive progress in unifyingthe study of political economy. It remainsfor future research to provide systematiccomparisons of the role of domestic politi-cal institutions in the relationship betweengovernmental policy and the internationaleconomy, and thereby to integrate morecomprehensive theories of politicaleconomy.

How to Subscribe

Subscriptions to the APSA-CP Newslet-ter are a benefit to members of the Or-ganized Section in Comparative Politicsof the American Political Science As-sociation. To join the Section, check theappropriate box when joining the APSAor renewing your Association member-ship. Sections dues currently run $7 an-nually, with a $2 surcharge for foreignaddresses. The printing and mailing ofthis Newsletter are paid for out of mem-bers’ dues. To join the APSA, contact:

American PoliticalScience Association

1527 New Hampshire Ave., NWWashington, DC 20036

Telephone: (202) 483-2512Facsimile: (202) 483-2657

Email: [email protected]

APSA-CP NewsletterMiriam Golden, EditorUniversity of California, Los AngelesDepartment of Political Science405 Hilgard AvenueLos Angeles, California [email protected]

Non-Profit Organization

U.S. Postage

PAID

UCLA


APSA-CP Winter 1997 · 2018. 12. 19. · APSA-CP Newsletter 2 Winter 1997 tradition when those questions, issues, and problems are closely related, be-cause they have been chosen

Documents