Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

Transcription

1 Democracy with Adjectives: Conceptual Innovation in Comparative Research Author(s): David Collier and Steven Levitsky Source: World Politics, Vol. 49, No. 3 (Apr., 1997), pp Published by: Cambridge University Press Stable URL: Accessed: 26/08/ :37 Your use JSTOR archive indicates your acceptance JSTOR's Terms and Conditions Use, available at JSTOR's Terms and Conditions Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue a journal or multiple copies articles, and you may use content in JSTOR archive only for your personal, non-commercial use. Please contact publisher regarding any furr use this work. Publisher contact information may be obtained at Each copy any part a JSTOR transmission must contain same copyright notice that appears on screen or printed page such transmission. JSTOR is a not-for-prit service that helps scholars, researchers, and students discover, use, and build upon a wide range content in a trusted digital archive. We use information technology and ols increase productivity and facilitate new forms scholarship. For more information about JSTOR, please contact support@jsr.org. Cambridge University Press is collaborating with JSTOR digitize, preserve and extend access World Politics.

2 Research Note DEMOCRACY WITH ADJECTIVES Conceptual Innovation in Comparative Research By DAVID COLLIER and STEVEN LEVITSKY* recent global wave democratization has presented scholars THE with challenge dealing ly with a great diversity postauthoritarian regimes. Although new national political regimes in Latin America, Africa, Asia, and former communist world share important attributes, many m differ proundly both from each or and from democracies in advanced industrial countries. Indeed, many are not considered fiilly democratic. This article argues that scholars respond this challenge by pursu two ing potentially contradicry goals. On one hand, researchers attempt increase in analytic order differentiation capture di verse forms that have emerged. On or hand, scholars are concerned with validity. Specifically, y seek avoid problem stretching that arises when con cept is cases applied for which, by relevant scholarly standards, it is not appropriate.1 The result has been a proliferation alternative forms, a including surprising number * We acknowledge valuable suggestions Ruth Berins Collier, Larry Diamond, Andrew Gould, Peter Houtzager, Marcus Kurtz, Terry Karl, David Laitin, George Lakf, Arend Lijphart, James Mahoney, Scott Mainwaring, Carol Medlin, Gerardo Munck, Guillermo O'Donnell, Michael Pr?tes, Philippe Schmitter, Laura Sker, Mark Turner, Samuel Valenzuela, and participants in on Berkeley Working Group Comparative Method. Steve Levitsky's participation in this research was supported by a National Science Foundation Graduate Fellowship, and David Collier's work on this at project Center for Advanced Study in Behavioral Sciences was supported by National Sci ence Foundation Grant no. SBR Giovanni Sarri, "Concept Misformation in Comparative Politics," American Political Science Re view 64 (December 1970); and David Collier and James E. Mahon, Jr., "Conceptual 'Stretching' Re visited: Adapting Categories in Comparative Analysis," American Political Science Review 87 (December 1993). World Politics 49 (April 1997),

3 involving hundreds DEMOCRACY WITH ADJECTIVES 431 "with adjectives."2 that have appeared Examples from among include "authoritarian democ racy," "neopatrimonial," "military-dominated," and "pro." This proliferation has occurred despite efforts by leading analysts standardize usage term on basis procedural definitions in tradition Joseph Schumpeter and Robert A. Dahl.3 In important respects this standardization has been successful. Yet as democratization has continued and attention has focused on an in creasingly diverse set cases, proliferation and or innovations has continued. Hence, given risk growing confusion, earlier effort standardize usage must now be supplemented by assessing structure meaning that underlies se diverse forms concept. This article initiates this assessment, on cate focusing qualitative gories4 employed in study recent cases democratization at level national political regimes, with particular attention work on Latin America.5 Our goal is twold: make more comprehensible structure complex alternative innovation that have emerged and examine trade-fs among se strate gies. We begin with Sarri's well-known moving up and down a ladder generality? aimed at avoiding stretching and increasing differentiation, respectively. Because this ap cannot proach be used pursue both at goals once, we find that scholars have ten turned or : creating "diminished", "precising" definition by adding defining attributes, and shifting overarching concept with which is associated (for example, from democratic regime democratic state). 2 A parallel expression, " without adjectives," appeared in debates in Latin America among observers concerned with persistence incomplete and qualified forms. See, for instance, Enrique Krauze, Por una democracia sin adjetivos (Mexico City: Joaqu?n Mortiz/Planeta, 1986). 3 Schumpeter, Capitalism, Socialism and Democracy (New York: Harper, 1947); and Dahl, Polyarchy: Participation and Opposition (New Haven: Yale University Press, 1971). 4 Along with qualitative categories that are focus this discussion, valuable quantitative in dicars have been developed for recent cases comparing democratization. Ultimately, it will be pro ductive bring ger insights about innovation employed in se alternative approaches. However, an essential prior step, which is our present concern, is learn more about innovations introduced by scholars who employ qualitative categories. 5 We are thus not primarily concerned with literature on advanced industrial democracies, al though this literature is an important point reference in studies we are examining. In a few places, we have included recent studies countries that are not actually part current episode democratization, but whose new relatively democracies are a point comparison in studies under review, for example, Colombia. We also include a few references or hisrical cases that have been used in recent as scholarship important points analytic contrast.

4 432 WORLD POLITICS More broadly, analysis seeks encourage scholars be more careful in ir definition and use concepts. The and or forms examined here are, after all, generally critical compo nents main substantive arguments presented by se researchers, ten advancing author's overall characterization case or cases in question. These are "data containers" that convey most salient facts about regimes under discussion.6 If one is describe new regimes adequately, se data containers must be in a employed clear and manner. appropriate Improved description, in turn, is essential for assessing causes and consequences, which is a central goal this literature. Many studies have treated as an outcome be explained, including major works comparative-hisrical analysis and old and new studies "social requisites."7 Or analyses have looked at impact and specific types on economic growth, income distribution, economic liberalization and adjustment, and international conflict.8 In se studies, results causal assess ment can be strongly influenced by meaning em ployed by author.9 We hope that present discussion can serve as a step ward a greater consistency and clarity meaning that will pro vide a more adequate basis for assessing causal relationships. 6 Sarri (fn. 1), Barringn Moore, Jr., Social Origins Dictarship and Democracy: Lord and Peasant in Making Modern World (Bosn: Beacon Press, 1966); Gregory M. Luebbert, Liberalism, Fascism, or Social Democracy: Social Classes and Political Origins in Interwar Regimes Europe (New York: Oxford Uni versity Press, 1991); Dietrich Rueschemeyer, Evelyne Huber Stephens, and John D. Stephens, Capi talist and Development Democracy (Chicago: University Chicago Press, 1992); Seymour Martin Lipset, "Some Social Requisites Democracy: Economic Development and Political Legitimacy," American Political Science Review 53 (March 1959); and idem, "The Social Requisites Democracy Revisited," American Sociological Review 59 (February 1994); John B. Londregan and Keith T. Poole, "Does High Income Promote Democracy?" World Politics 49 (Ocber 1996); and Adam Przeworski and Fernando Limongi, "Modernization: Theories and Facts," World Politics 49 (January 1997). 8 Adam Przeworski and Fernando Limongi, "Political Regimes and Economic Growth," Journal Economic 7 Perspectives (Summer 1993); Kenneth A. Bollen and Robert W. Jackman, "Political Democracy and Size Distribution Income," American Sociological Review 50 (August 1985); Larry Sirowy and Alex Inkeles, "The Effects on Democracy Economic Growth and Inequality: A Review," Studies in Comparative International Development 25 (Spring 1990); Karen L. Remmer,"The Politics Economic Stabilization: IMF Standby Programs in Latin America, ," Compara tive Politics 19 (Ocber 1986); Barbara Stailings and Robert Kaufman, eds., Debt and Democracy in Latin America (Boulder, Colo.: Westview Press, 1989); Bruce Russett, Grasping Democratic Peace: a Principles for Post-Cold War World (Princen: Princen University Press, 1993); Michael E. Brown, Sean M. Lynn-Jones, and Steven E. Miller, eds., Debating Democratic Peace: An International Secu rity Reader (Cambridge: MIT Press, 1996); Alfred Stepan and Cindy Skach, "Constitutional Frame works and Democratic Consolidation: Parliamentarianism versus Presidentialism," World Politics 46 (Ocber 1993); Juan J. Linz and Arturo Valenzuela, eds., The Failure Presidential Democracy (Balti more: Johns Hopkins University Press, 1994); and Guillermo O'Donnell, "Delegative Democracy," Journal Democracy 5 (January 1994). 9 See, for example, Kenneth A. Bollen and Robert W. Jackman, "Democracy, Stability, and Di chomies," American Sociological Review 54 (August 1989), ; and Russett (fn. 8),

5 DEMOCRACY WITH ADJECTIVES 433 It merits emphasis that se innovation are no means by unique qualitative research on recent democratization. are They found in many domains, both in social sciences and beyond.10 A furr goal this article is refore advance broader understanding how qualitative researchers deal with se basic issues analytic differentiation and validity. I. Definitions Democracy in Research on Recent Democratization In his famous analysis "essentially contested concepts," philoso pher W. B. Gallie argues that is " con appraisive political cept par excellence.'m one finds endless over Correspondingly, disputes appropriate meaning and definition. However, goal Gallie's analysis is not simply underscore importance such disputes, but show that a recognition contested status a con given cept opens possibility understanding each meaning within its own framework. With reference, he argues that "politics being art possible, democratic targets will be raised or low ered as circumstances alter," and he insists that se alternative stan dards should be taken on seriously ir own terms.12 In this spirit, we focus on procedural definitions that have been most widely in employed research on recent democratization at level national political regimes. These definitions refer democratic rar than substantive or procedures^ policies or outcomes that might be viewed as democratic. These definitions are also "minimal," in that y deliberately focus on smallest possible number attrib utes that are still seen as a producing viable standard for not ; re is surprisingly, disagreement about which attributes are needed for definition be viable. For most example, se scholars dif ferentiate what y view as more specifically political features regime from characteristics society and economy, on 10 For an analysis that focuses on some se same with reference anor social sci ence concept, see David Collier, "Trajecry a Concept: in 'Corporatism' Study Latin Amer ican Politics," in Peter H. Smith, ed., Latin America in Comparative Perspective: New Approaches Method and Analysis (Boulder, Colo.: Westview Press, 1995). For discussions by linguists and cognitive scientists intuitive structure that underlies se, see D. A. Cruse, Lexical Semantics (Cambridge: Cambridge University Press, 1986), chap. 6; George Lakf, Women, Fire, and Dangerous Things: What Categories Reveal about Mind (Chicago: University Chicago Press, 1987), chaps. 2, 6; and John R. Taylor, in Linguistic Categorization: Protypes Linguistic Theory, 2d ed. (Oxford: Oxford University Press, 1989), chaps W. B. Gallie, "Essentially Contested Concepts," Proceedings Aristelian Society 56 (London: Harrison and Sons, 1956), 184; emphasis in original. 12 Ibid., quote at 186; see also pp. 178,189,190,193.

6 434 WORLD POLITICS grounds that latter are more as appropriately analyzed potential causes or consequences, rar than as features democ racy itself.13 Within this framework, we focus on a "procedural minimum" defini tion that presumes fully contested elections with fiill suffrage and absence massive fraud, combined with effective guarantees civil liberties, including freedom speech, assembly, and association.14 However, re is no means consensus on a by single definition. Some scholars, for example, have created an uexpanded procedural minimum" definition by adding criterion that elected governments must have effective in some power as we govern?which, will see below, is a crucial issue countries. II. Sarri's Strategies We first consider Sarri's for achieving differentiation and avoiding stretching. Sarri builds on a basic insight about organization a concepts: significant aspect relationship be tween meaning concepts and range cases which y can apply be undersod in terms a "ladder generality."15 This lad der is based on a pattern inverse variation between number defining attributes and number cases. Thus, concepts With, fewer defining attributes more cases commonly apply and are refore on higher ladder generality, whereas concepts with more defining attributes cases apply fewer and hence are lower on ladder. 13 For discussions procedural definitions, see Guillermo O'Donnell and Philippe C. Schmitter, Transitions from Authoritarian Rule: Tentative Conclusions about Uncertain Democracies (Baltimore: Johns Hopkins University Press, 1986), chap. 2; Samuel P. Huntingn, "The Modest Meaning Democ racy," in Robert A. Pasr, ed., Democracy in Americas: Spping Pendulum (New York: Holmes and Meier, 1989); Schumpeter (fn.3); and Dahl (fn. 3). On minimal definitions, see Giuseppe Di Palma, To Craft Democracies: An on Essay Democratic Transitions (Berkeley: University California Press, 1990), 28; and Samuel P. Huntingn, The Third Wave: Democratization in Late Twentieth Century (Norman: University Oklahoma Press, 1991), 9. On treating characteristics society and economy as a cause or consequence see, Juan J. Linz, "Totalitarian and Authoritar ian Regimes," in Fred I. Greenstein and Nelson W. Polsby, eds., Handbook Political Science, vol. 3 (Reading, Mass.: Addison-Wesley, 1975), 182; and Terry Lynn Karl, "Dilemmas Democratization in Latin America," Politics Comparative 23 (Ocber 1990), O'Donnell and Schmitter (fn. 13), 8 (but see note 33 below); Larry Diamond, Juan J. Linz, and Seymour Martin Lipset, "Preface," in Diamond, Linz, and Lipset, eds., Democracy in Developing Coun tries: Latin America (Boulder, Colo.: Lynne Rienner, 1989), xvi; Di Palma (fn. 13), 16. See also Juan J. Linz, The Breakdown Democratic Regimes: Crisis, Breakdown, and Reequilibration (Baltimore: Johns Hopkins University Press, 1978), Sarri (fn. 1), 1040, actually refers a ladder "abstraction." However, because term abstract is ten undersod in contrast concrete, this label can be confusing. We refore find that "ladder generality" expresses intended meaning more clearly.

7 DEMOCRACY with adjectives 435 Differentiation One Sarri's goals is show how differentiation can be increased by moving down ladder generality concepts that have more defining attributes and fit a narrower range cases. These concepts more provide fine-grained distinctions that for some pur poses are invaluable researcher.16 This move down ladder is ten accomplished through creation what we will call "classical".17 Classical are undersod as ^///in stances root definition18 in relation which y are formed, at same time that are y differentiated vis-?-vis or classical this concept. Thus, "parliamentary," "multiparty," and "federal are " all considered def democratic initely (by whatever standard author is at using), same time that each is considered a particular type (see Figure 1). In research on recent cases democratization, use classical achieve differentiation is found, for example, in important debate on consequences as parliamentary, opposed presidential,.19 Moving down ladder generality provides useful differentiation, and just noted an play important role in recent litera ture. Yet formed in this manner may leave analyst vulner able stretching, because y presume cases under discussion are definitely democracies. If case particular being stud ied is less than fully democratic, n use se as a ol differentiation may not be appropriate. Analysts re fore seek concepts that distinguish among different degrees democ racy, in addition distinguishing among different types. Because classical only contribute second se two goals, y have not been most common means con differentiation in studies recent democratization. ceptual 16 Sarri (fn. 1), We refer se as classical because y fit within "classical" understanding cat egorization discussed by such authors as Lakf (fn. 10), 9 and passim; and Taylor (fn. 10), chap In referring root definition, we do not imply that it is "correct" definition relevant concept (in this case, ). It is simply definition that, for a particular author, is point departure in forming subtype. We will use occasionally expression "root concept" refer concept (again, in present context, ) that is point departure for various innovations analyzed here. 19 Linz and Valenzuela (fn. 8); Stepan and Skach (fn. 8); and Giovanni Sarri, Comparative Consti tutional An in Engineering: Inquiry Structures, Incentives, and Outcomes (New York: New York Uni versity Press, 1994).

8 436 WORLD POLITICS REGIME Up Ladder: Avoiding Conceptual Stretching Civilian regime* Competitive Elecral regimeb regime0 I Root Concept I DEMOCRACY i Down Ladder: Increasing Differentiation? Parliamentary *1 Two-party 0 Federal f Figure 1 The Ladder Generality: increasing differentiation versus avoiding Conceptual Stretching "John A. Booth, "Framework for Analysis," in John A. Booth and Mitchell A. Seligson, eds., Elec tions and Democracy in Central America (Chapel Hill: University North Carolina Press, 1989), 26. bruth Berins Collier and David Collier, Shaping Political Arena: Critical Junctures, Labor Movement, and Regime Dynamics in Latin America (Princen: Princen University Press, 1991), 354. cjames Petras and Fernando Ignacio Leiva, Democracy Politics (Boulder, Colo.: Westview Press, 1994), 89. and Poverty in Chile: The Limits Elecral djuan J. Linz. "Presidential or Parliamentary Democracy: Does It Make a Difference?" in Juan J. Linz and Arturo Valenzuela, eds., The Failure Presidential Democracy (Baltimore: Johns Hopkins University Press, 1994), 3. emarkj. Gasiorowski, "The Political Regimes Project," Studies in Comparative International Devel opment 25 (Spring 1990), 113. Raymond Duncan Gastil, "The Comparative Survey Freedom: Experiences and Suggestions," Studies in Comparative International Development 25 (Spring 1990), 35. Avoiding Conceptual Stretching Sarri's proposal for avoiding stretching is move up ladder generality concepts that have fewer defining attributes and correspondingly fit a broader range cases.20 In present context, 20 Sarri (fn. 1), 1041.

9 DEMOCRACY WITH ADJECTIVES 437 this involves concepts located above on ladder gener ality. Scholars commonly view as a specific type in relation overarching concept regime. Hence, if y have as misgivings wher a case particular is a really democratic can move regime, y up ladder and simply call it a regime. However, because a as as shifting concept general regime entails a great loss differentiation, scholars have typically moved an intermediate level (Figure 1)?adding adjectives term regime and reby generating classical differentiate spe cific types regime. The resulting remain more general than concept, in that y encompass not only democra cies but also some /w^-democracies. Examples include "civilian re gime," "competitive regime," and "elecral regime." Although scholars thus achieve some differentiation in relation regime, y do not specifically commit mselves idea that case under discussion is a. A similar pattern is followed when scholars use a synonym for as regime, in "civilian rule" and "competitive polity."21 Although climbing ladder generality helps avoid stretching, it has an important drawback. Because se re main more than general concept, this approach leads a loss differentiation. Thus, taken ger, Sarri's two can advance one or or se goals, but not both at once. As a consequence, many scholars have turned or. III. Diminished Subtypes An alternative strategy innovation, that creating "di minished",22 can contribute both achieving differentiation and avoiding stretching. It is a strategy widely used in literature on recent democratization. Two are points crucial for under standing diminished. First, in contrast classical sub types discussed above, diminished are not fiill instances root definition "" employed by author who presents subtype. For example, "limited-suffrage " and "tutelary are " undersod as less than complete instances democ 21 See, respectively, Richard Wilson, "Continued Counterinsurgency: Civilian Rule in Guatemala," in Barry Gills, Joel Rocamora, and Richard Wilson, eds., Low Intensity Democracy: Political Power in New World Order (London: Plu Press, 1993); and Terry Lynn Karl, "Democracy by Design: The Christian Democratic Party in El Salvador," in Giuseppe Di Palma and Laurence Whitehead, eds., The Central American Impasse (London: Croom Helm, 1986). 22 The idea diminished builds on discussion radial concepts in Collier and Mahon (fn. 1), See also Lakf (fn. 10), chap. 6.

10 438 WORLD POLITICS racy because lack one or more its y defining attributes.23 Conse quendy, in using se analyst makes a more modest claim about extent democratization and is refore less vulnerable stretching. The second concerns point differentiation. Because diminished sub an types represent incomplete form, y might be seen as hwingfewer defining attributes, with consequence that y would be higher on ladder generality and would refore provide less, rar than more, differentiation. However, distinctive feature diminished is that y generally identify specific attributes that are missings reby establishing diminished character at subtype, same time that y identify or attributes that are present. Because y specify missing attributes, y also increase differentiation, and diminished subtype in fact refers a different set cases than does root definition. The inclusion and exclusion cases that occurs with a diminished as subtype, opposed moving up or down ladder can generality, be illustrated with examples contemporary Britain, United States, and Guatemala (Figure 2). Britain and United States, but probably not Guatemala (at least up through mid-1990s), would be seen as democratic in terms procedural minimum definition. If we climb ladder generality, we find that broader concept "elecral regime"24 encompasses all three cases. Lower down on ladder classical subtype "parliamentary " would include one two democracies, that is, Britain. By contrast, dimin ished subtype "illiberal " would include only Guatemala, case that specifically did not fit root definition.25 Figure 3 some presents examples many diminished that have been generated in relation procedural minimum and minimum expanded procedural definitions noted above. In many instances, scholars created diminished in which more than one component attribute is missing, but for pur 23 Because y are less than complete instances, it might be objected that y are not really "sub types" at all. on a Drawing term from one can cognitive linguistics, refer m as con ceptual "blends" that are derived in part from concept. However, avoid referring repeatedly " and blends," it seems simpler in discussion below call m. See Gilles Fauconnier and Mark Turner, "Conceptual Projection and Middle no. Spaces," Report 9401, Department Cognitive Science (San Diego: University California, San Diego, 1994). 24 This subtype is undersod have meaning explained above in discussion Figure Regarding illiberal see, Figure 3. Two furr points about diminished should be underscored. First, if scholars fail identify root definition in relation which y form, it is difficult determine wher a given subtype is classical or diminished.

11 DEMOCRACY WITH ADJECTIVES 439 Up Ladder Elecral Regime Cases: Britain, U.S., Guatemala.? Root Concept Democracy Cases: Britain, U.S. ~~T~ Down Ladder Parliamentary Democracy Cases: Britain Diminished Subtype Illiberal Democracy Cases: Guatemala Figure 2 Inclusion and Exclusion Cases: Ladder Generality versus Diminished Subtypes pose illustration we focus on examples in which author has been reasonably careful in a isolating single missing attribute. The in first group (la) refer cases where missing attribute is full suffrage. Here we find terms such as "male" or "oligarchical'', which are used in pointing contrast between cases contemporary and hisrical cases prior advent universal suffrage. Where attribute full contestation is missing (lb), as when important parties Second, fact that a subtype refers what might be undersod as a "problematic" feature does not mean necessarily that it is a diminished subtype. For example, O'Donnell's con cept "delegative," which refers cases with weak horizontal accountability among branches government, in fact meets his minimum definition, given that he does not include horizontal accountability in definition. See O'Donnell (fh. 8), 56. Hence, in his usage, del egative is a classical subtype. For a discussion that refer "problematic" democ racies, see a longer version present analysis in David Collier and Steven Levitsky, "Democracy 'with Adjectives': Conceptual Innovation in Comparative Research," no. Working Paper 230 (Notre Dame, Ind.: The Kellogg Institute, University Notre Dame, 1996), The above characteriza tion as a delegative classical subtype should be undersod as correcting assessment this subtype presented in Collier (fn. 10),

12 1. Diminished from Procedural Minimum Definition (la) Missing Attribute: Full Suffrage Limited * Male 1* Oligarchical 0 (lb) Missing Attribute: Full Contestation Controlled *1 De fac one-party * Restrictive f (lc) Missing Attribute: Civil Liberties Elecral g Hard 11 Illiberal 1 2. Diminished from Expanded Procedural Minimum Definition Missing Attribute: Elected Government Has Effective Power Govern Guarded Protected Tutelary 1 1 Figure 3 Partial Democracies: Examples Diminished Subtypes "Ronald P. Archer, "Party Strength and Weakness in Colombia's Besieged Democracy," in Scott Mainwaring and Timothy R. Scully, eds., Building Democratic Institutions: Party Systems in Latin Amer ica (Stanford: Stanford University Press, 1995), 166. bgeorg Sorensen, Democracy and Democratization: Process and in a Prospects Changing World (Boul der, Colo.: Westview Press, 1993), 20. c Jonathan Hartlyn and Arturo Valenzuela, "Democracy in Latin America since 1930," in Leslie Bell, ed., The Cambridge Hisry Latin America, vol. 6 (Cambridge: Cambridge University Press, 1994), 99. dbruce Michael Bagley, "Colombia: National Front and Economic Development," in Robert Wes son, ed., Politics, Policies, and Economic Development in Latin America (Stanford: Hoover Institution Press, 1984), 125. cadrian Leftwich, "Governance, Democracy, and Development in Third World," Third World 14 Quarterly (1993), 613. fcarlos H. Waisman, "Argentina: Autarkic Industrialization and Illegitimacy," in Larry Diamond, Juan J. Linz, and Seymour Martin Lipset, eds., Democracy in Developing Countries: Latin America (Boulder, Colo.: Lynne Rienner, 1989), 69. gaxel Hadenius, "The Duration Democracy: Institutional vs. Socio-economic Facrs," in David Beetham, ed. Defining and Measuring Democracy (London: Sage Publications, 1994), 69. hguillermo ODonnell and Philippe C. Schmitter, Transitions from Authoritarian Rule: Tentative Conclusions about Uncertain Democracies (Baltimore: Johns Hopkins University Press, 1986), 9. This is our translation ir democradura. In English y refer this as "limited," same term used in la above, but y make it clear that ir meaning corresponds lc. Donald Emmerson, "Region and Recalcitrance: Questioning Democracy in Souast Asia" (Paper at presented World Congress International Political Science Association, Berlin, 1994), 14. JEdelber Torres Rivas, "La gobemabilidad centroamericana en los noventa," America Latina, Hoy 2 (June 1994), 27. This is our translation his democracia " vigilada. kbrian Loveman, 'Protected Democracies' and Military Guardianship: Political Transitions in Latin America, ,"JournalInteramerican Studies and World 36 Affairs (Summer 1994), 'Adam Przeworski, as "Democracy a Contingent Outcome Conflicts," in Jon Elster and Rune Slagstad, eds., Constitutionalism and Democracy (Cambridge: Cambridge University Press, 1988),

13 DEMOCRACY WITH ADJECTIVES 441 are banned from elecral we competition, find terms such as "controlled" and "restrictive". Where civil liberties are in complete (lc), scholars have used terms such as "elecral" and "illib eral". The in final group (2), introduced by scholars who created expanded procedural minimum definition, a provide usefiil reminder that meaning on depends root defini tion in relation which are y formed. From point departure that definition, se scholars introduced diminished in which missing attribute is effective power elected government govern. These refore do not meet expanded procedural minimum standard for, although y do meet minimum procedural standard. Examples that refer cases where military is seen as an having inordinate degree polit ical power include "protected" and "tutelary". Diminished, n, are a useful means avoid stretching in cases that are less than fully democratic. They also provide differentiation new by creating analytic categories. Various scholars have pointed need move a beyond dichomous ization authoritarianism and and recognize "hybrid" or "mixed" character many postauthoritarian regimes.26 Figure 3 suggests scale. that this recognition has indeed occurred, and on a rar large For countries that are less than fully democratic, however, ques tion arises as wher it would be better avoid identifying m as, for example, in cases gross violations civil liberties and/or severe restrictions on elecral competition. An example such questioning is Bruce Bagley's rejection numerous dimin ished that have been applied National Front period in Colombia ( ); se include "restricted," "controlled," "limited," "oligarchical," "elitist," and "elitist-pluralist". Bagley instead characterizes Colombia as a au subtype 26 James M. Malloy, "The Politics Transition in Latin America," in James M. Malloy and Mitchell A. Seligson, t?s., Authoritarians and Democrats: Regime Transition in Latin America (Pitts burgh: University Pittsburgh Press, 1987), ; Carine M. Conaghan and Rosario Espinal, "Unlikely Transitions Uncertain Regimes? Democracy without Compromise in Dominican Re public and Ecuador," Journal Latin American Studies 22 (Ocber 1990), 555; Jonathan Hartlyn, "Crisis-Ridden Elections (Again) in Dominican Republic: Neopatrimonialism, Presidentialism, and Weak Elecral Oversight," Journal Interamerican Studies and World 36 Affairs (Winter 1994), 93-96; Terry Lynn Karl, "The Hybrid Regimes Central America," Journal Democracy 6 (Summer 1995); and Francisco Weffort, Qualdemocracia? (S?o Paulo: Companhia das Letras, 1992),

14 442 WORLD POLITICS thoritarianism: as an "inclusionary authoritarian regime."27 Or scholars have addressed this issue by climbing ladder generality labels such as "civilian," or "competitive," "elecral" regime, which are found in upper part Figure 1. A third option is use dismis sive like those noted above, such as "facade," in which adjective essentially cancels democratic character subtype. Scholars should be self-conscious about analytic and nor mative implications choosing form in relation democ racy, as some opposed or concept. IV. Precising Definition Democracy Anor strategy innovation focuses on definition itself and is concerned with "precising" definition by adding defining attributes.28 As concept is extended new set tings, researchers may confront a case particular that is classified as a on basis a commonly accepted definition yet is not seen as fully democratic in light a larger shared understanding concept. This mismatch between case and formal definition may lead analysts make one or more explicit criteria that are implic itly undersod be part overall meaning, but that are not in cluded in definition. The result is a new definition intended change way a case particular is classified. Although this procedure precising definition could be seen as raising standard for, it can also be undersod as adapting definition a new context. This innovation increases differentiation, by a adding furr criterion for establishing cutf between democ racy and non. The strategy may reby also avoid concep tual stretching because it does not apply label cases "" that, in light this new criterion, sees as analyst incompletely de mocratic. Although use this strategy may arise from a concern with adapting concept fit a particular context, modified definition should not be undersod as being relevant only that context. Indeed, modified definition can also new provide sight in or cases for which significance new defining tributes had not previously been fully appreciated. 27 Bagley, "Colombia: National Front and Economic Development," in Robert Wesson, ed., Politics, Poli cies, and Economic in Latin America Development (Stanford, Calif: Hoover Institution Press, 1984), See Giovanni Sarri, "Guidelines for Concept Analysis," in Sarri, ed., Social Science Concepts: A Systematic Analysis (Beverly Hills, Calif.: Sage Publications, 1984), 81; and Irving M. Copi and Carl Cohen, Introduction Logic, 9th ed. (New York: Macmillan, 1994), In Social Science Concepts (p. 42), Sarri also uses this as a verb, as in " precise" a definition. in at

15 DEMOCRACY WITH ADJECTIVES 443 One example precising definition is emergence standard an expanded procedural minimum, noted above. In several Central American countries, as well as in South American cases such as Chile and one Paraguay, legacy authoritarian rule has been persistence "reserved domains" military power over which elected governments have little or no authority.29 Hence, despite free or rela tively free elections, civilian governments in se countries are seen by some as analysts lacking effective power govern. In light se au thoritarian legacies, and ten in response claims that because se countries have held free elections are y "democratic," some scholars have modified procedural minimum definition by as an specifying criterion explicit that elected government must a reasonable degree have effective power rule. With this revised defin ition, countries such as Chile, El Salvador, and Paraguay have been ex cluded some by scholars from set cases classified as democracies, even though y held relatively free elections.30 These scholars have thus adapted definition explicitly include an attribute that is ten taken for granted in studies advanced industrial democracies but that is absent in se Latin American cases. This revised definition has received substantial acceptance, although re certainly has not been full agreement on treatment specific cases. For in example, analyzing Chile in post-1990 period, Rhoda Rabkin takes exception usage adopted by scholars who intro duced expanded procedural minimum definition. She argues that problem civilian control military does not a represent suf ficient challenge democratically elected government qualify Chile as a "borderline".31 Two or initiatives precise definition have not received simi lar acceptance, but serve y usefiilly illustrate issues that arise with this strategy. The first is found in discussions what might be called a Tocquevillean definition that includes a focus on selected aspects social relations. In analyzing postauthoritarian Brazil, scholars such as Francisco Weffort and Guillermo O'Donnell 29 J. Samuel Valenzuela, "Democratic Consolidation in Post-Transitional Settings: Notion, Process, and Facilitating Conditions," in Scott Mainwaring, Guillermo O'Donnell, and J. Samuel Valenzuela, eds., Issues in Democratic Consolidation: The New South American Democracies in Comparative Perspective (Notre Dame, Ind.: University Notre Dame Press, 1992), Karl (fn. 13), 2; Valenzuela (fn. 29); and Brian Loveman, "'Protected Democracies' and Military Guardianship: Political Transitions in Latin America, ," Journal Interamerican Studies and World Affairs 36 (Summer 1994). See also Humber Rubin, "One Step Away from Democracy" Jour nal Democracy 1 (Fall 1990). 31 Rhoda Rabkin, "The Aylwin Government and 'Tutelary' Democracy: A Concept in Search a Case?"Journal Interamerican Studies and World Affairs 34 (Winter ), 165.

16 444 WORLD POLITICS have been struck by degree which rights citizenship are un dermined by pervasive semifeudal and authoritarian social relations that persist in some regions country. In light this concern, y have precised definition so as exclude Brazil. Thus, Weffort adds definitional requirement "some level social equality" for a country be considered a, and O'Donnell introduces a similar stipulation.32 In adopting this usage, se authors view mselves as remaining within procedural framework. Yet in troducing issues social relations noneless an represents important departure from earlier procedural definitions. We will see in next section that O'Donnell has subsequently arrived at an alternative means incorporating this set concerns in his ization. Anor effort precise definition has arisen from a concern that in many new democracies in Latin America and in former com munist countries, elected at presidents times make extensive use de cree power, circumvent democratic institutions such as legislature and political parties, and govern in a manner plebiscitar?an that is seen as having strong authoritarian undercurrents. In Latin American context prominent examples include Carlos Menem in Argentina, Fer nando Collor de Mello in Brazil, and, in most extreme case, Al ber Fujimori in Peru. The concern with se authoritarian tendencies has led some authors include checks on executive power in ir procedural criteria for and thus exclude cases unconstrained presidentialism.33 However, this innovation has likewise not been widely adopted. Precising definition can thus serve usefully both introduce finer differentiation and avoid stretching, and associ ated debates have raised essential issues about meaning that schol ars wish attach term "." Yet caution is in order. Among alternative innovation examined in this article, precising in a sense introduces most drastic change: it modifies definition itself. If an innovation based on 32 Francisco Weffort, "New Democracies, Which Democracies?" no. Working Paper 198, Latin American Program (Washingn, D.C.: Woodrow Wilson International Center for Scholars, 1992), 18; Weffort (fn. 26), ; Guillermo O'Donnell, "Challenges Democratization in Brazil," World Policy Journal 5 (1988), ; and idem, "Transitions, Continuities, and Paradoxes," in Main waring, O'Donnell, and Valenzuela (fn. 29), Authors who have employed horizontal accountability in ir definitions include Philippe C. Schmitter and Terry Lynn Karl, "What Democracy Is... and Is Not," Journal Democracy 2 (Sum mer 1991), 76, 87; and Alan R. Ball, Modern Politics and Government, 5th ed. (Chatham, N.J.: Chatham House, 1994), O'Donnell and Schmitter (fn. 13), 8, actually include it in ir formal definition, but it appears play no role in ir subsequent analysis.

17 DEMOCRACY WITH ADJECTIVES 445 precising is widely accepted, it has important effect changing definitional point departure with reference which all or are in pursued, effect unsettling "semantic field" in which se scholars are working.34 By contrast, introduction a new subtype does not affect semantic field in same way. In a litera ture in which confusion is a recurring problem, analytic gains from precising definition must be weighted against cost unsettling semantic field. Hence, it is important that scholars avoid "definitional gerrymander in sense a new ing,"35 introducing definition every time en y counter a somewhat anomalous case. However, contrast between first example (adding criterion effective power govern) and third example (adding horizontal accountability) shows that schol ars may in fact impose constructive limits on precising. In first ex ample, inability elected governments exercise effective power was seen as invalidating ir democratic character. By contrast, in third example, involving heavy-handed assertions power by pres ident, a crucial point is that se are presidents elected leaders. Hence, it might be argued that it is treat appropriate se as meet regimes a ing minimal standard for and avoid precising?as long as (1) y maintain presidential elections and a general respect for civil liberties and legislature and (2) opposition parties are not banned or dissolved (as occurred in Peru in 1992). Finally, initiative precising can raise issue bringing back in definition attributes that scholars previously had explicidy decided exclude. An example is concern with social re lationships in Tocquevillean approach. These authors could be seen as remaining within a procedural framework, in sense that y argue that political participation becomes less meaningfijl in context ex treme social inequality. However, this innovation reintroduces features social relations in a way that noneless a represents major shift from earlier recommendations about which attributes should be included in definitions. V. Shifting Overarching Concept Yet anor strategy innovation is shift overarching concept, in relation which is seen as a specific instance? that is, as a classical subtype. Thus, although scholars most commonly 34 On problem unsettling semantic field, see Sarri (fn. 28), Jennifer Whiting, personal communication, suggested this term.

18 446 WORLD POLITICS view as a subtype overarching concept "regime" (and procedural criteria for discussed above would routinely be undersod as applying some recent regime), literature has undersod as a in subtype relation or overarching as concepts, in "democratic government" and "democratic state." Hence, when a given country is labeled "democratic," can meaning vary according overarching concept which term is at tached. A shift in can an overarching concept yield alternative standard for a case declaring particular be a, yet without eir or modifying stretching concept "democratic regime." As can be seen in Figure 4, scholars have used this strategy create a standard that can be eir less or more demanding. For a example, scholar who finds Brazilian in immediate post-1985 period be so poorly institutionalized that it appears use inappropriate overar ching label "regime" may refer a "democratic situation." This distinc tion follows example Juan Linz's analysis Brazil during earlier post-1964 authoritarian period: he introduced concept an "authoritarian situation" take account weak institutionalization national political structures.36 Or analysts concerned with im mediate post-1985 period in Brazil have referred "democratic gov ernment" in order suggest that a although particular government (that is, head state and immediate political leadership that surrounds head state) has been elected democratically, ongo ing functioning democratic procedures is not necessarily assured. By shifting overarching concept from regime government in this way, scholars lower standard for applying label "democratic." Alternatively, by shifting overarching concept from "regime" "state," O'Donnell establishes a more demanding standard for labeling particular countries a. Writing after Brazil's presidential election 1989, which led scholars reinterpret Brazil as a having de mocratic regime, O'Donnell raises questions about democratic char acter state in Brazil, as well as in some or South American countries. He suggests that, in context "neeudalized" and at times "sultanistic" political relationships found in many parts country, national state does not protect basic rights citizenship, 36 See Juan J. Linz, "The Future an Authoritarian Situation or Institutionalization an Au thoritarian Regime: The Case Brazil," in Alfred Stepan, ed., Authoritarian Brazil: Origins, Policies, Future (New Haven: Yale University Press, 1973). uses Malloy "democratic moment" convey a sim ilar meaning. See Malloy (fn. 26), 236.

19 DEMOCRACY WITH ADJECTIVES 447 Lowering Standard Point Departure Raising Standard Author Democratic Situation Democratic Government Democratic Regime Democratic State Duncan Baretta and Markf Yes No Hagopian and Mainwarin^ Yes No O'Donnell (1988)c Yes No O'Donnell (1993)d Yes No Figure 4 Shifting Overarching Concept: Characterizing Post-1985 Brazil "Silvio Duncan Baretta and John Markf, "Brazil's Abertura: Transition What?" in James M. Malloy and Mitchell A. Seligson, eds., Authoritarians and Democrats: Regime Transition ica (Pittsburgh: University Pittsburgh Press, 1987), 62. in Latin Amer bfrances Hagopian and Scott Mainwaring, "Democracy in Brazil: Problems and Prospects," World Policy Journal A (Summer 1987), 485. cguillermo O'Donnell, "Challenges Democratization in Brazil," World Policy Journal S (Spring 1988), 281. dguillermo O'Donnell, "On State, Democratization, and Some Conceptual Problems," World Development 21 (August 1993), and specifically rights citizens fair and equal protection in ir social and economic relationships. This failure may not direcdy influence functioning regime, in sense direcdy affect ing elections and associated civil liberties that are core features procedural understanding a democratic regime. However, O'Donnell argues, this failure legal and bureaucratic institutions secr public protect and a promote broader set democratic rights citizens is a crucial feature Brazilian state. Hence, al though he recognizes that countries like Brazil have a democratic "regime," he excludes m from set countries he considers have democratic "states." This shift in overarching concept consti tutes anor way a more making differentiated assessment what is deemed be an case incomplete, specifically by estab

20 448 WORLD POLITICS lishing a higher and a lower standard for and declaring that se countries meet only lower standard.37 From standpoint maintaining a procedural definition, this innovation can be seen as a better solution prob lem that O'Donnell and ors initially tried address by creating Tocquevillean definition. Thus, in conjunction with shifting over arching concept, democratic "regime" continues have a procedural definition, and this concern with broader functioning citizenship in context authoritarian patterns social relations is addressed via concept state. To summarize, strategy shifting among alternative overarch can serve ing concepts introduce finer differentiation an by creating additional analytic category. When strategy is used lower standard for a case declaring be a, it can also help avoid stretching concept a democratic regime. When strategy is used raise standard it is not relevant problem stretching, because it is not concerned with avoiding what might be seen as mistake a case a calling given democratic regime. Rar, it provides additional information about cases that are as accepted hav ing democratic regimes. VI. Concluding Observations We have examined innovation used by analysts recent democratization as y seek meet a twold challenge: in creasing analytic differentiation in order adequately characterize diverse regimes that have emerged in recent years and con maintaining ceptual validity by avoiding stretching. Our goal has been both make more comprehensible structure complex se and evaluate strengths and weaknesses strate gies. Even when se scholars proceed intuitively, rar than self-con sciously, y tend operate within this structure, which, as noted above, is no means by unique research on recent democratization.38 Yet, in interest and analytic clarity, it is far more de sirable for m proceed self-consciously, with a full awareness trade-fs among different. Figure 5 an provides overview this structure. analytic Conceptual innovation has occurred at three levels root concept democ 37 Guillermo O'Donnell, "On State, Democratization and Some Conceptual Problems: A Latin American View with Glances at Some Postcommunist Countries," World (1993), 1359 and passim. 38 See again references in note 10. Development 21, no. 8

21 DEMOCRACY WITH ADJECTIVES Shifting Overarching Concept Over arching Concept 5a. Lowering Standard Increases differentiation, avoids stretching 2. Up Ladder Generality Does not increase differentiation, avoids stretching : : 5b. Raising Standard Increases differentiation, does not avoid stretching? Root Concept 4. Precising Definition Democracy Increases differentiation, avoids stretching Subtypes 1. Down Ladder Generality Increases differentiation, does not avoid stretching 3. Diminished Subtypes Democracy Increases differentiation, avoids stretching Figure 5 Evaluating Conceptual innovations: Contribution Increasing Differentiation and Avoiding Conceptual Stretching racy itself,, and overarching concept. We observed that Sarri's (1) moving down ladder generality clas sical and (2) moving up ladder classical can serve regime usefully eir increase differentiation or avoid stretching, but cannot y do both simultaneously. These two goals can be achieved simultaneously, however, by (3) creating diminished, (4) precising definition by adding defining attributes, and (5a) shifting overarching concept as a means lowering standard. contrast By (5b), shifting overar ching concept raise standard for does not serve

22 450 WORLD POLITICS avoid stretching vis-?-vis concept a democratic regime, but it does introduce new differentiation. We have also underscored issues that are distinctive particular. Diminished are useful for characterizing hybrid regimes, but y raise issue wher se regimes should in fact be treated as, rar than authoritar ianism or some or concept. The strategy precising definition is subject perennial problem scholarly disputes over definitions, as well as problem imposing limits on defini tional gerrymandering. Although strategy shifting overarch ing concept with goal raising standard is not relevant problem stretching, it does allow scholars introduce new analytic issues without a abandoning procedural definition and regime. Finally, se share two common problems. First, given structure complex se, potential for confusion and miscommunication is considerable. It is imperative that scholars clearly define and explicate conception are so as y using situate mselves unambiguously in relation this structure. Second, this literature faces a major dilemma in proliferation concepts and terms, many which mean same approximately thing. The consequence, once can again, be growing scholarly confusion. Al new terms are created in though part because scholars are pursuing se goals differentiation and avoiding stretching, y may also be introduced with goal developing compelling labels that vividly draw attention novel forms.39 In literature on national over political regimes past three decades, an important alytic innovations have periodically been introduced in conjunction with creation and/or systematization concepts and concept labels that vividly capture important constellations for exam phenomena: ple, "authoritarianism," "polyarchy," "bureaucratic authoritarianism," "corporatism," and "consociational."40 Correspondingly, invention additional concepts that play this same role is an impor 39 For a reminder how important vivid labels can be, one need only look at impressive evolu tion game ory, with its codification different patterns political interaction designated by such labels as "prisoners' dilemma," "chicken," "stag hunt," "slippery slope," and "battle sexes." 40 Juan J. Linz, "An Authoritarian Regime: Spain," in Erik Allardt and Yrj? Littunen, eds., Cleav ages, Ideologies and Party Systems: Contributions Comparative Political Sociology, Transactions Westermarck Society, vol. 10 (Helsinki: Academic Booksre, 1964); Dahl (fn. 3); Guillermo O'Donnell, Modernization and Bureaucratic-Authoritarianism: Studies in South American Politics, Institute Interna tional Studies, Politics Modernization Series no. 9 (Berkeley: University California, 1973); Philippe C Schmitter, "Still Century Corporatism?" Review Politics 36 (January 1974); and Arend Li jphart, Democracy in Plural Societies: A Comparative Exploration (New Haven: Yale University Press, 1977).

23 DEMOCRACY WITH ADJECTIVES 451 tant goal in ongoing study regimes. However, if research on de mocratization in a see degenerates competition who can come up with next famous concept, comparative study regimes will be in serious trouble. Hence, we propose anor major objective concept usage, one that introduces a furr trade-f vis-?-vis two goals achieving differentiation and avoiding stretching. In addition pur suing se goals, scholars should aim for parsimony and avoid ex cessive proliferation new terms and concepts. Orwise, that advantages derive from refinements discussed in this article will be overridden by resulting confusion.