Вы находитесь на странице: 1из 24

International Organization Foundation

Diffusion and the International Context of Democratization Author(s): Kristian Skrede Gleditsch and Michael D. Ward Source: International Organization, Vol. 60, No. 4 (Autumn, 2006), pp. 911-933 Published by: Cambridge University Press on behalf of the International Organization Foundation Stable URL: http://www.jstor.org/stable/3877851 . Accessed: 31/05/2011 20:42
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=cup. . 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. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Cambridge University Press and International Organization Foundation are collaborating with JSTOR to digitize, preserve and extend access to International Organization.

http://www.jstor.org

Diffusion and the International Contextof Democratization


Kristian SkredeGleditsch and MichaelD. Ward

Abstract Democracy does notevolvesui generis. spatial The in clustering democthat factors role racyand transitions suggests international playa prominent in forgas We that their often ingdemocracies well as influencing durability. argue democracy comes aboutas a resultof changesin the relative actorsand powerof important evaluations particular of both groupsas well as their institutions, of whichare often influenced forcesoutsidethecountry question. in The scope and extent conof by nections with other democratic countries a region strengthen in can fordemsupport ocratic reform helpsustain and in institutions transitional Resultsfrom democracies. a transition modeldemonstrate international that factors exerta strong can influence on theprospects transitions democracy, the spatialclustering democfor to and in cannotadequately explainedby the hypothesized be domestic racy and transitions social requisites individual of countries.

The many transitions democratic rule in the so-called "thirdwave" of democto ratizationhave renewed scholarlyinterest what affectsthe prospectsfordemocin ratization.So far,however, an understanding the causes for the emergence of of democratic political institutions has remained elusive. In retrospect, is easy to it look back on particulartransitions democracy as ineluctable. However, providto ing generalizationson circumstancesthathave been favorable fordemocratictransitions requires one to see beyond the idiosyncrasies of individual changes. Is democracy "caused" by economic or social factors,or by political culture,or do transitions come about by just plain luck? The idea thatdemocracy has certain requisites can be traced to Lipset's thesis thateconomic developmentis a key precondition for democratic rule.' Other perspectives give prominence to norms or We aregrateful comments Brian for from Nathaniel A'Hearn, Beck,Scott Gates, KyleBeardsley, Havard DavidLektzian, Pevehouse, Reiter, Jon Dan Kenneth Heather Hegre, Schultz, Smith, Haivard and the and as atthe Strand, Kaare Strom, editors, twoanonymous reviewers,wellas the participants on Conference theInternational of Diffusion Democracy Markets, of and Los UniversityCalifornia, LiberalizationtheWeatherhead at Center International for Harvard Affairs, University, Cambridge, October 2003. Mass., 1. Lipset 1960.
International 60, Organization Fall 2006, pp. 911-933 ? 2006byTheIO Foundation. on of Diffusion Politicaland Economic Angeles,March2003, and theConference theInternational

DOI: 10.1017/S0020818306060309

912 International Organization valuesheldto favor development durability democratic the of and rule,2 inequality in or therelative of strength classes,3or eventsat "critical junctures" a country's and and and Przeworski Limongi Przeworski More political development.4 recently, to abouttransitions democrevealnothing colleagueshaveclaimedthat requisites and demobetweensocial and economicfactors racy:the apparent relationship under to cratic is institutions stems from howdemocracy more likely survive merely events.5 are certain and themselves random to conditions, transitions democracy These variousexplanations and predictions clearlydiffer entailquitedifferent a in aboutprospects democracy. for Still,theyare all "similar" relating country's that and for to internal societies presuming to factors prospects democracy various of or eventsin other countries not affect do politicalinstitutions the likelihood the influence prosIn factors transitions. thisarticle, arguethatinternational we random aremorelikely but and transitions notsimply are pectsfordemocracy, that in thewakeofchanges theexternal in The and environment. temporal spatialclusin and or diffusion, enduring, cross-boundary tering democracy transitions suggests of thatinfluence development persistence politicalinstituand the dependencies of theories democrations.We reconsider roleof diffusion light current the in of between on the and can factors change balanceofpower tization, focus howexternal hold that and forces well as theevaluations different as groups regimes opposition can overparticular of forms governance. democratization comeaboutin Although coninternational actors, multiple waysandcan involvea wide rangeof different for of indicators theprospects textand external shocksgenerally providebetter transition do theattributes individual than of states.

Democratization:StylizedFacts
is We see democracy a form governance as the of where powerofexecutives limor itedby otherinstitutions wheregovernments selectedeither and are directly withopen or unrestricted forcanelections, entry indirectly through competitive the to didates.6 Amongmanyefforts measuredemocracy, Politydata providean additive scale Figure1 shows twenty-one-point of a state'sdegreeof democracy.7
2. See Almondand Verba1963; and Mullerand Seligson 1994. 1990. 3. See Muller1988; Rueschemeyer, and 1992; and Vanhanen Stephens, Stephens and 4. See Bollen 1979; Casper and Taylor1996; Moore 1993; O'Donnell, Schmitter, Whitehead 1988. 1986; and Przeworski 5. See Przeworski Limongi1997; and Przeworski al. 2000. et and of suchas protection politi6. Alternative definitions democracy of other features, mayemphasize cal rights theinclusiveness participation; overviews, Beetham1994; Doorenspleet of or for see 2000; thatproceand Vanhanen also includeoutcomes of 1990. Some definitions "substantive" democracy duraldemocracy assumedto lead to; see Shapiro2003. is measuresof democracyAccessed 30 June2006. Otherempirical essex.ac.uk/-ksg/Polity.html). in Alvarezet al. 1996; Bollen 1990; theFreedom House indexdescribed Gastil 1985; and including

7. We use a modifiedand expanded version of the Polity IV data, available from(http://privatewww.

of Context Democratization913 International

- 10

.5-

.4o3

ofdemocracies Proportion score -Mean Polity

- 5

.2 ,-

-5
.1

-10

1850

1900 Year

1950

2000

1816-1998 FIGURE Theglobal distribution democracy, 1. of of theglobal averageof thePolityscale and theproportion theworld'sindepenWhereas dentstatesthatare considered democratic over thelast two centuries.8 in in about5 percent thestates theworldweredemocracies 1816,democof only raciesoutnumbered autocracies theendofthe 1990s.The shareof democracies by over and in thesystem notincreased but has expanded contracted gradually, rather timein whatHuntington three "waves of democracy."9 calls The shareof countries thatare democracies, however, dependsnot only on of statesbut is also affected changesin the number by changeswithin existing artiare ofdemocracy merely that "waves" states. Some have argued independent than facts the of overtimerather changesin institutions.10 reflecting growth states existence in states continuous in nineteen the of However, distribution democracy
than time or Vanhanen set available a smaller ofcountries a shorter interval the for 1990-areonly can such other data. definitions emphasize criteria, as broad that participation, leadto Polity Although centuries twentieth and in somewhat ofdemocracies thelatenineteenth early different classifications for criteria 2000-alternative Paxton when women many were denied and voting rights-see groups with tend of such and rights, togo together competitive democracy, as participation protectionhuman elections the in contemporary era. the or of 8. Thethreshold "democracy" is here toa score seven above, set for following suggested we 479.Forthe threshold "coherent and for see means, treat 1995, democracies"; Jaggers Gurr global scaleas valuesonthePolity without caseswith "in institutions transition""interrupted" or regular value nondemocracies assign numerical of- 10. and a 9. Huntington 1991. 2000. and 10. See Doorenspleet andAlesina, 2000; Spolaore, Wacziarg

914 International Organization .8ofdemocracies Proportion Mean score Polity 10

.6-

0
2

I -10 -5

1850

1900
Year

1950

2000

Year

FIGURE Measuresofdemocracy a sampleofnineteen 2. continuously for states,1816-1998 independent

from1816 (shownin Figure2) suggests thatthere have been manyinstitutional within evident periods in in withlargeincreases theshareofcounstates, changes tries that democratic well as periods which are as in states becameless demmany thetwoworldwarsand theperiodof decolonialization. ocratic, notably during and Assessingtheshareof democracy changesovertimebased on independent statesalone also excludesfrom denominator of theworld'spopulation in the all colonizedterritories. 3 displaysa cartogram thedistribution democof of Figure where relative ofa state scaledaccording population assigned the size to and is racy, a shade accordingto its Polityscore." Populations nonsovereign nonin and democratic entities displayed a residual are as blockfor eachgeographical region.12 the has how democracy become Comparing maps for 1945 and 2002 illustrates morewidespread bothindustrialized developingsocieties. in and considerably Whereasmoststates Europeand LatinAmerica in in have democratic institutions wereautocracies 1945. Even though shareofpopulation in the 2002, many living under democratic remains rule lowerbycomparison Africa Asia,many in and areas
11. We use Jaggers Gurr'ssuggested and cut-off to between "coherent democrapoints distinguish and and cies," "coherent autocracies," "anocracies";see Jaggers Gurr1995,479. 12. Thesecartograms based on a "density are see and 2004; equalizing" approach; Gastner Newman and historical estimates from Gleditsch 2005. population

International Context Democratization915 of

1945

Autocracy

Democracy
0 1 2 3 4 5 6 7 8 9 10

2002
.?

-10-9

-8 -7 -6 -S -4 -3 -2 -1

\P

?"1t

P..

"

FIGURE Cartogram democracies 3. and autocracies, to of proportional population thatwerecoloniesin 1945 have becomedemocratic statesby 2002. Indeed,the Middle East (including ArabNorth remains domiAfrica)is theonlyregionthat natedby autocracies. The growth the shareof the world'spopulation of living under democratic institution 1945 to 2002 is all themoreremarkable popfrom as ulationgrowth been higher thelow-income has in countries assumedto be less to receptive democracy. The notion global"waves" ofdemocracy autocracy alerted of and has researchers to thepossiblerole of international to how thesecinfluences.'3 Manypoint ond and third waves of democratization coincidedwithtwo majorwatersheds in world the WarII andtheendoftheCold War. history, namely endofWorld Merely democratization autocratization some"international or to howattributing context,"
13. See Huntington 1991; and Ray 1995.

916 International Organization context how and international without the ever,explainslittle clarifying relevant this influences trendshighand manyof the global prospectsfordemocracy, For do of with lighted notvaryconsistently thedistribution democracy. example, warsseemto havepreceded and is and bothdemocratization autocratization, there of the interno obvious relationship at betweenwar and democracy the level in to national in variation democracy shifts politiLikewise,attributing system.14 " inturn thequestion cal ideology thechanging or States of begs position theUnited ofwhyideologies foreign sometimes favor or changein waysthat policydoctrines and favor democracy sometimes autocracy. are In ourview,international democratization notparinfluence processesthat global influticularly likelyto be foundat a global level. Lookingforuniversal encesthat as identical affect countries all alikeis probably ill-conceived assuming as and independent The each country. global level is an aggregate processeswithin thatmaskslargeregionaldifferences variation. democracies and Although presin dominate someregions, in has ently autocracy beenwidespread thesameregions in at other timeperiods. demonstrates clustering thedisregional Figure3 clearly of tribution democracy both1945 and 2002, and similar of in patterns geographithat cal clustering forother hold timeperiodsas well. Since 1815,theprobability a randomly of chosencountry be a democracy about0.75 ifthemajority its will is are of are but neighbors democracies, only 0.14 if the majority its neighbors nondemocracies. have and Transitions democracy to havealso clustered geographically, countries in transitions been farmorelikelyto undergo to transitions democracy following of estimates local states. neighboring Figure4 displaysnonparametric regression in thelikelihood transitions and of between democratic autocratic regimes a given a within 500 km of year,giventheproportion otherstatesthatare democracies radiusof a country.16 that will be The unconditional probability an autocracy in verysmall,in fact replacedby a democratic regime any one yearis obviously will less than.015. However, estimated that the probability an autocracy become as a democracy, sharply an S-curve givenby thesolid line in Figure4, increases withhigher the More precisely, estimated of proportions democratic neighbors. in demof to exceeds.1 fora country a relatively probability transition democracy ocraticregiontoward right thehorizontal axis. The riskthata democracy the of will be replacedby an autocracy, indicated thedashedline,displaysa similar by the As withtheregionalcontext. wouldbe expectedfrom S-shapedrelationship the seculartrend toward higher a of probabilproportion democracies, maximum twice as highas the is will become a democracy roughly itythatan autocracy likelihood democracies of highest goingauthoritarian.

14. See Gleditsch 2002a; and Mitchell, Gates, and Hegre 1999. 15. For example, Robinson 1996. 16. The geographical informationis taken from the Gleditsch and Ward 2001 minimum distance data.

International Context Democratization917 of .12.10 -o .08,=

Transitiondemocracy to Transitionautocracy to

.06 .04-

.2

.8 .6 ofdemocratic Proportion neighbors .4

FIGURE Transition 4. neighbors probabilities proportion democratic of by

The Diffusion Democracy of


in of to couldbe attributed increases wealth Changesin thedistribution democracy and other social requisites sincethe heldto be conducive democracy. to However, domestic social andeconomic tend conditions deemedimportant to changeslowly overtime, is difficult see how thegreat it could arise to in variability democracy from stablerelationships alone.The geobetween social requisites democracy and in also graphicalpatterns suggestthatthe likelihoodof a transition a country on the international other context and eventsin otherstates.Although depends studies have shownsimilar in of evidenceof "diffusion" democracy the empirical sense of spatialclustering,7 it is less clear whatthisstemsfrom, whatit is or aboutdemocracy one state influences prospects democracy another. in in for that the In thisarticle, extend we to theories democratization therole of interof existing national influences. The study democratizationcomplicated themany of is possiblewaysin which by one regimemaydisappear in and be replacedby another, by thedifficulties and
17. See O'Loughlinet al. 1998; and Starr1991.

918 International Organization movers as assessingex antewho shouldbe regarded themainagentsor potential in transitions. somecases, themostinteresting In in transitions feature explaining is retrospectively to accountforhow a new politicalcoalitionseizes powerand as developsnew institutions, whenSlobodanMilosevi6was deposedin Yugoslavia. However, other unrelated in for cases,autocratic maywithdraw reasons regimes to theforces as and actorsinfluencing in institutions their aftermath, was thecase with fallofthemilitary in the of restoration democracy Argenjuntaandsubsequent tinaafter Falklands the war.Sometransitions, as thefallofMilo'evi6,involve such or themselves carwhileother transitions initiated rulers are popularuprising, by riedoutbyactors In for with previous the closelyassociated leadership. Paraguay, a withlitmadetheinitial democracy example, military government stepstoward tlepopular of forces in and whereas Uruguay, pressure, representatives thearmed at to to at politicians agreed return openelections a closedmeeting theNaval Club. of it a to Nonetheless, although is perhaps stretch talkabouta canonicaltheory can a widerange possibilities which transitions occur of in democratization, regime and undera framework on maybe subsumed focusing power,mobilization, the evaluations important of on actors. Muchof theliterature democratization argues thatdemocracy as whenno singleactoror of emerges an outcome social conflict Bueno de Mesquitaand colleaguessuggest groupcan imposeitsruleon others.'8 thatpoliticalcoalitions in survive proportion theratioof the size of thewinto to whohave a rolein determinningcoalition thesize of thegroupofindividuals methods the coalition, so-calledselectorate.19 ing thatwinning Institutionalizing forsharing becomerational powerand establishing optionswhen politicalrights theselectorate or dominate control are actors unableto fully politexpandsso that ical powerby repressive meansor by distributing goods. Fromthisperprivate thatinfluence structural of theories democracy spective, existing pointto factors institherelative for of as powerandresources groups, well as support democratic tutions. Powertendsto be generally moredispersed amonggroupsin economiwitha moreadvanceddivision laborthanin agricultural of callydevelopedstates societieswherelandis theprimary values favoring sourceof wealth.20Similarly, democratic are morelikelyto be widespread rule whenno groupcan achieveits unrestricted are and preferences, whendemocratic politicalsystems seen as well to relative autocratic to and alternatives as less of a threat powerful functioning interests.2 However, there no inherent is reasonwhystruggles overinfluence and Accordresources states. shouldbe confined the of within boundaries individual one to actorsand of in of ingly, can think diffusion terms how linkages external eventsinfluence relative the and powerand thelikelystrategies choicesof relevantgroupsin struggles overpoliticalinstitutions outcomes. and

20. See Boix 2003; and Vanhanen 1990. 21. For example, Almondand Verba1963.

19. Bueno de Mesquita et al. 2003.

18. See Olson 1993; Przeworski 1990. 1988; and Vanhanen

International Context Democratization919 of and Power ofActorsand Groups Diffusion theRelative Bothdomestic external the and eventsand processescan influence powerof specific or demgroups, seeking thereby groups undermining existing regimes assisting ocratic reforms. idea offoreign-imposed an The changerepresents extreme regime case of coerciveexternal calls forregime influence. changein Despitetherecent of as there fewclearcases where are has roguestates, democracy emerged a result and The are intervention, mostimposedregimes autocratic. SovietUnion foreign anditsWarsaw Pactallies,for in in intervened Hungary 1956andCzechoexample, slovakiain 1968 partly offearthat out couldlead to democracy local reforms and theabandonment socialism. of intervention does through Accordingly, imposition notseeman important sourceof democratization. coercionmay also take moresubtleforms thandirect intervention. However, The fearofSovietintervention itself to reforms sufficed deter political by generally in Eastern of the until adoption theso-calledSinatra the Europeduring Cold War, where Mikhail Gorbachev.22 countries could"do ittheir under doctrine, More way," other or actors promote can democratization actions generally, states transnational by thatstrengthen domestic actorsseeking and democratic reform weakenthepower of autocratic We stateswill tendto supthat regimes.23 hypothesize democratic aboutmore movements government and that reforms wouldbring portopposition similar to that in Likewise,opposition regimes. groups autocracies are connected or interact open,democratic with to from societies morelikely receivesupport are transnational actors. External can on dramatic support have a particularly power impact therelative of groupswhenwe see shifts the coalitionsthathold powerin neighboring in entities. conmodel"suggests thatsmallchangesin external Schelling's "tipping textmay suffice yield cascades thatcan generate critical to a mass in political contestation.24 processes often Such are heldtohaveplayedoutin thefallofsocialism in Eastern Europe,wheretheinitial politicalchangesin Polandand Hungary effects in and spurred subsequent changes Czechoslovakia East Germany.25 Tipping shouldlead to a clustering transitions, one transition of the with increasing likelihood of subsequent transitions connected in states. The military for of coercioncan also have implications component interstate theprospects democracy. for constrains the Many researchers arguethatconflict for is rule,and thatdemocracy likelyto breakdownunder prospects democratic thethreat conflict.26 of institutions been have that initial Thompson argues political

22. Ash 1999. 23. See Deutsch 1954; Keck and Sikkink1999; Randle 1991; Smith, Pagnucco,and Lopez 1998; and Solingen1998. 24. Schelling1971. 25. See Kuran1989; and Lohmann1994. 26. For example, and Gates,Knutsen, Moses 1996.

920 International Organization Sustained shaped by rulers'need to obtainresourcesand mobilizemilitarily.27 and fostered as authoritarianism powerbecame morecentralized. rivalry threats relative as By contrast, propolitical peace facilitated pluralism, internal political cesses could unfold withmoreinsulation threats. Barzel and Kiser from external as external of institutions insethreats hindered development voting the arguethat cure rulerswere unable to make crediblecommitments contracts and withthe withfragruled.28Mansfield and Snyder hold thatleadersin transitional regimes to ile institutions likely relyon nationalism diversionary conflict remain are to and in power, The reversals.29 geographical the thereby increasing riskof democratic isolationand protection from threat external mayin partexplainwhyearlysteps "zones toward weremoredurable England in than France.Similarly, in democracy of peace" first to whenpowerful stateswereforced abandonambitions emerged of regional to we and hegemony domination.30 Accordingly, expectdemocracies be morelikelyto emerge and thrive regions withstablepeace. in and Diffusion Evaluations Institutional of Arrangements of Theoriesof democratization also be cast in terms evaluations particular of can institutional Even in situations whereno singlegroupcan monoparrangements. olize politicalpower, as need notlead to democratic institutions, powersharing ruleandresist actors often feartheconsequences unmitigated of powerful popular such as Mill and Marx,expected theorists, democracy. Manynineteenth-century theexpansion suffrage thelaborclass to inevitably to massiveredistriof lead to bution private of of and wave,many property income.31Untiltheadvent thethird of elites in Southern Europe and LatinAmericatendedto be skeptical whether The warin Bosdemocratic institutions couldmaintain order property and rights.32 nia was in partdriven an exaggerated of view of theeffectiveness democracy, by minorwhere ethnic Serbsbelieved they wouldbe a perpetually that repressed many the undermajority radicalchangesfollowing about rule.33 Most predictions ity in that introduction majority of materialize countries havefailedto rule,however, have undergone leadersin autocracies reluctant transitions democracy. to Initially of states be morewilling initiate to if difficult reforms theexperiences other may had be as bad as they that costsand consequences reforms not the of suggest may hold on to and thatnumerous have been able to former rulers autocratic feared, fears influence under democratic rule.Accordingly, of democracy poweror retain In are likelyto weakenas morereference countries become democratic. many
27. Thompson 1996. 28. Barzel Kiser and 1997.
29. 30. 31. 32. 33.

Mansfield Snyder and 2002. 1996. Thompson Muller1999. See Alexander 2002 on Spain; and Pevehouse2002a and 2002b on LatinAmerica. See Muller2000.

International Context Democratization921 of been seen as a prereqand have increasingly circles, democracy good governance democuisiteforeconomicgrowth development. Such beliefscan facilitate and a relationship eventhough does notunambiguously research racy, support empirical between and democracy economicperformance.34 The likelihood democratic of reform dependsnotonlyon theperexperiments ceivedbenefits from but costsofnotbeinga democdemocracy, also theexpected the which haveincreased over racy, considerably time.During Cold War, probably a in did fashion notnecessarily imposeparticular ruling country an authoritarian ties for or to problems a country's standing a leader'sability maintain withother as in world. werenotdemocracies, states, mostcountries especially thedeveloping was aboutprotecting freeworld, the DespiteU.S. rhetoric democracy clearlynot a requirement theselection allies andaid recipients.35 increase democin in The of that raciesin thedeveloping world, however, changestheset of countries a state is likely be compared With declining of to to. the strategic importance allies in the that worldafter Cold War, the rulers had enjoyed autocratic longdeveloping many isolated. international found themselves Hence,many increasingly standing support leadersmayseek to initiate to in democratic reforms efforts keep on good terms states. withtherestof theworldor notto look bad relative other to comparable

Empirical Analysis
influence the In thissection, reexamine factors we and how international regional mechanisms likelihood thata country the will be democratic. Although specific transithatlead to changesare notdirectly we observable, can observewhether distionsare moreor less common the that factors reflect mechanisms following in cussed in theprevioussection. examinethesepropositions an analysisof We and Gleditsch Wardsugtransitions between and two-way democracy autocracy. could be analyzedas a Markovchain gestedthatchangesin politicalstructures we between different statesovertime.36 simplicity, here For processof transition limitourselves twopossiblestates, whichwe define to and democracy autocracy, an has operationally whether observation a value of 7 or above on theinstituby of tionalized distribution a scale. In a transition model,theprobability democracy variable forobservation at timet is modeledas a function i's prior of i yi, history are or stateat previoustimeperiodst - 1, t - 2,..., t - T. If theobservations Markov conditional on theprevious we chain.37 observations, havea first-order only

34. For example, Przeworski Limongi1993. and

36. Gleditsch Ward1997. and 37. Harary, and Norman, Cartwright 1965,371-77.

35. See Meernik, Krueger, and Poe 1998; and Reiter 2001a.

922 International Organization The transition outcome is matrix a first-order for Markovchainwitha binary

Poo Poil
Pio
Pll

where indicates probability changefrom to 1 (that Yit 1,yit-1= 0), the of 0 is, = poi and t indicates probability remaining 1 from - 1 to t (that Yit, 1, the at of is, = Pll 1). Yit-1= We can estimate conditional the transition givensomesetofcovariprobabilities ates of interest by xit
= + Yi, xta] Pr(yit = 1 yi, t-, t-1,xit) F[[xip

of indicate effects the where is either logitor a probit F a The link.38 / parameters t covariates theprobability a 1 at timet givena 0 at time - 1,that Pr(yit = on of is, = 0). The effects theprobability a 1 at timet givena 1 at timet - 1, 1 on of Yi,-1 To comy Pr(yit= l yi-1 = 1), are givenby theparameters = a + /3. facilitate withPrzeworski Limongi'sresults,39 letyit= 1 if a statei is an we and parison In at autocracy timet andyit= 0 if it is a democracy. thiscase, theestimated / on coefficients be interpreted indicating effects a covariate thelikecan of the as a effect lihoodthat democracy becomean autocracy; indicates covariate's a will y on the likelihood Since theprobability autocracies. thatautocracies will remain the forall the possible outcomesat timet given = 1 mustsum to unity, yi,-1 at likelihood thatan autocracy timet - 1 will becomea democracy timet is at will or 1 minus probability an autocracy endure. the that 1I1, four We examinediffusion, the influence international of or factors, through If of covariates. The first three to local and globaldiffusion regime types. pertain to local diffusion we shouldexpectautocracies be morelikely processesoperate, to experience of the transitions democracy greater proportion democratic to the < 0). We identify local context theprothe countries (thatis, y by neighboring are based of stateswithin 500 kmradiusthat democracies, a portion neighboring to on minimum drivestransitions if distancedata.40 Similarly, global diffusion autocracies theglobal as to autocracies should becomeless likely remain democracy, of increases. We assumethattransitions be contaproportion democracies may We and increasethelikelihood transitions neighboring in states. further of gious assumethatdemocratic the in statesdo notinfluence likelihood transitions other the thatdemocratic but stateswill becomeautocracies, mayinfluence likelihood the that autocracies becomedemocracies. light this, constrain paramwe will In of
38. See Beck et al. 2001; and Yamaguchi1991,chap. 3. 39. Przeworski Limongi1997. and do 40. Gleditsch the and Ward2001. Although 500 km threshold somewhat a is arbitrary, results notlook dramatically different other for see distance 2002a, 94-95. thresholds; also Gleditsch

International Context Democratization923 of eter forthisvariable be 0 and lety = a. We use a binary of indicator whether to /3 a transition democracy to a state takesplace in a neighboring within 500 kmradius. Ourfinal We covariate diffusion to reflecting pertains theroleofconflict. believe that conflicts to for threats a state'sterritory decreasetheprospects may reflecting from conflict shouldbe distinguished These,however, democracy. participation, which may include decisions to intervene conflicts in elsewherein the internationalsystem thatare unlikely influence prospects democracy. for the We to consider of territorial threats a simplecountof thenumber yearsthat enduring by a country remained peace on its territory a proxyforthe stability has of at as of data are based on the Correlates War (COW) data, with peace. Our conflict somemodifications updates.41 and Most previous for factors transtudies diffusion therole of international on or sitions democracy to of attributes have disregarded potential the impact domestic bothinternational domestic and processes;42 we consider and factors. aggreAn in between country's and a institutions thecountries itsregional gaterelationship context alonedoes notprovide because evidence diffusion of processes, convincing the principal social and economicconditions democto hypothesized influence suchas differences grossdomestic in (GDP) percapita,can also be racy, product shownto cluster to countries and maybe responding geographically43 individual common trends As rather than transitions. such,we face displaying interdependent an inverse form Galton'sproblem distinguishing funcof between of independent tionalrelationships interdependent whatexisting studand diffusion processes:44 ies that notconsider do to domestic attribute diffusion actually characteristics may stemfrom that in attributes influence geographical prospects clustering domestic fordemocracy. needto demonstrate theobserved conWe of effects regional that textand diffusion notmerely do domestic factors. stemfrom plausibleomitted The primary which of measure "social requisites" a country's is GDP percapita, we measure thenatural of thelaggedlevel of real GDP percapita.45 as Many log researchers have arguedthatnegative or economicperformance crisescan affect theprospects democracy. for economic declinearemore Countries experience that to heldtohave and declineis often transitions, economic likely experience regime undermined transidemocracies the wake of decolonialization promoted in and tionsfrom of in the to We wave.46 consider effect autocracy democracy thethird in and of growth real GDP percapitaas a measure economicperformance crises, in the case of negativegrowth. may also stemfroma Negativeperformance in to suchas changes commodity economic shocks, country's exposure exogenous
41. Gleditsch 2004. 42. Forexample, O'Loughlin al. 1998;andStarr see et 1991. 43. Gleditsch 2002a. 44. Galton 1889.

45. Gleditsch 2002b.We prefer simpler the over natural specification usingGDP percapitaand log itssquareas suggested Przeworski Limongi1997,becausewe see no clearreasonwhydemocand by racyshouldbecomeless likelybeyondsome level of income. 46. See Gasiorowski 1991. 1995; and Remmer

924 International Organization whichin turn undermine that regime prices, mayinducesocial conflict can in turn or Based on Rodrik's of indicator external stability. exposure sensitivisuggested of terms trade of economicshocksby thevolatility a country's we ty,47 measure of overa five-year totaltradeas a proportion its period, multiplied a country's by and conflict strife, and GDP.48Finally, to democracy mayalso be related domestic we consider whether state a in was involved a civilwar.49 Manycivilwarsrevolve such around control overthegovernment. forms violent of Other conflict, as secesThe sionistmovements, shouldalso be expectedto undermine existing regimes. sincecivil wars of are however, implications civilwarfordemocracy ambiguous, of can toppleautocracies, undermine and democracies, lead to theemergence new autocratic regimes. the 1875to thepresent, and dataareavailablefrom Although geographic political theavailability dataon laggedGDP percapitaandeconomic constrains of growth for oursampleto theyears1951 to 1998. Economicdataareoften missing develin socialisteconomies, and statesinvolvedin conflict the stanopingcountries, darddatasourcesthat work.50 havebeenused in mostexisting Here,we use more biasesdue GDP datato protect comprehensive, expanded sampleselection against to nonrandom data.51 missing

Results
The results our baselinemodel are shownas Model 1 in Table 1. Each row of and a namedisplays estimated coefficient theimplied = the Y listing covariate /3 a + p coefficient that The standard for column fields. covariate thesubsequent in errors thecoefficients shownin parentheses therowbelow.52As can be for in are of seen at thebottom Table 1, thelikelihood of ratiotestof thenull hypothesis states(thatis, that = y or thattheparamequal slopes acrossprevious P3 regime etersa arejointlyinsignificant) clearlyrejected. is Hence,thecovariates appear will becomeautocrato have different on effects thelikelihood thatdemocracies cies and thelikelihood that will remain autocracies. autocracies are Most of our hypotheses diffusion democratization strongly on and supthe results Model 1 in Table 1. As can be seen from negative for portedby the^ estimates for and / in thefourth of themainbodyof thetable,a higher row
47. Rodrik 1999. 48. These dataareunfortunately available after 1965. only 49. Gleditsch 2004. out World 50. Przeworskial. 2000, example, include for in Penn et countriesthe Data, leaving only in in analysis terms the points our socialist developing and we states. many Although canreplicate key
of signand significance estimating of coefficients measure, usingtheAlvarezet al. 1996 democracy thissourcedramatically of reducessamplesize and introduces sample possibleproblems nonrandom attrition. 51. Gleditsch 2002b. + 52. The variance j is givenby Var(&) + Var(/3) 2Cov(&,/3). for

International Context Democratization925 of TABLE Results estimation transition 1. model for of


Model 1 (1951-98) Covariates Constant
LOGGED GDP PER CAPITA LOGGED ENERGY CONSUMPTION PER CAPITA

Model 2 (1951-98) /3 1.863 (0.868) -0.401 (0.099) -0.483 (0.268) 0.423 (0.233) 0.003 (0.002) -0.021 (0.012) -1.025 (1.091) -0.023 y 3.682 (0.551) -0.067 (0.062) -0.712 (0.209) -0.016 (0.157) -0.005 (0.002) 0.003 (0.008) -2.590 (0.672) -0.176

Model 3 (1875-1998)

/3 2.276 (0.802)

y 3.675 (0.552) -0.064 (0.062) -0.717 (0.21) -0.013 (0.157) -0.004 (0.002) 0.003 (0.008) -2.592 (0.672) -0.172 (0.066)

/P
-1.297 (0.234) -0.526 (0.126) -0.591 (0.197) 0.381 (0.192) 0.001 (0.002) -0.575 (0.723) -0.008

y 2.871 (0.196) -0.073 (0.079) -1.04 (0.179) 0.07 (0.153) -0.002 (0.002) -1.856 (0.559) -0.251 (0.055) 0.002

-0.501

(0.088)

PROPORTION NEIGHBORING -0.525 OF DEMOCRACIES (0.258) CIVILWAR 0.379 (0.225)


YEARS OF PEACE AT TERRITORY

0.002

ECONOMIC GROWTH GLOBAL PROPORTION OF NEIGHBORING TRANSITION


TO DEMOCRACY
TIME AS DEMOCRACY TIME AS AUTOCRACY DEMOCRACIES

(0.002) -0.025 (0.013) -0.512 (1.047)

(0.066)
0.001

(0.008)

(0.004)

(0.002)
6,159 6909.6 (df= 15) 4412.87 (df= 5) 6,159 6920.8 (df = 17) 2417.2 (df = 8)

(0.002)
8,788 9835.1 (df = 15) 3293.1 (df = 7)

N Model LR-X2 TestofHo: Constant slopes X2

of decreases likelihood autocthe that proportion democratic neighbors significantly ~ racieswill endure and increases likelihood the that democ(that is, = -0.717) racieswillbreak down(that a in = is,/3 -0.525). Moreover, transition a neighboring decreases likelihood an autocracy endure. more the A that will country significantly environment decreases likelihood an autocracy endure the that will peaceful regional butdoes nothave a statistically on that significant impact thelikelihood democracies willbreakdown.53 we evidence that transitions democracy to are Finally, find morelikelythehigher globalproportion democracies; the of the however, global of does notexerta significant effect thesurvival on rates proportion democracies
53. Additional with DiehlandGoertz's tests the 2000measure enduring (not of rivals shown) likewisesuggest an enduring that makes autocracies likely endure hasno consistent more to but rivalry effect the on survival democracy. of

926 International Organization and of democracies.54 attributes globalcharacterfor Controlling country-specific of isticsdoes notremovetheeffects theregional and of proportion democracies conclusion that transitions. contrast Przeworski Limongi's In and to neighboring out transitions as emerge exogenously a "deus ex machina" of thewhimsof hisare the for tory,55 results Model 1 in Table 1 showthatautocracies significantly whena neighboring less likelyto endure a region in withmoredemocratic states, is stateexperiences transition democracy, whenthere less conflict.56 a to and withexpectations The effects thedomestic of consistent covariates largely are and previous The research. results Model 1 in Table 1 can be readas supportfor GDP per capita a and thatalthough higher ing Przeworski Limongi'sconclusion does decreasetheriskthat will breakdown,it does notsignificantly democracies to increasethelikelihood a transition of from (thatis, the autocracy democracy ~ the estimated is effectively zero). Likewise,economicgrowth strengthens surin vival of democracies does notpromote but transitions autocracies. Indeed,the ^ that democthe estimated coefficient positive. is Civil warsincrease likelihood that racieswillbreakdownbuthaveno substantive on effect thelikelihood autocthat racieswill endure. Additional (not regressions shown)indicated ourmeasure after of exogenous economicshockswas notrelated regime to typeor transitions this data available.57 1965; however, maybe due to thelimited Markov It maybe questioned followa first-order whether dataactually our procan cess or all theinformation abouttransition survival and probabilities be sumWe marized theprevious statesof regimes by plus thecovariates. wereunableto did the that Markovspecification notfit rejectthenullhypothesis a second-order data significantly better thanthefirst-order MarkovprocessforModel 1. This is notparticularly Markovprocessesinvarias surprising, modelsforhigher-order

that 54. This contrasts somewhat with as 2001b andPevehouse2002a find research, Reiter previous whileKadera, theregional of of the does notinfluence duration democracy, proportion democracies the increases surand Shannon2003 findthata higher of Crescenzi, global proportion democracies worldregionsas vival ratesof democracies. within The first use two studies, however, proportions defined the COW projectrather and none of the studies thancountry-specific reference by groups, consider bothglobal and regional influences. 55. Przeworski Limongi1997. and after 1945 yields 56. Reestimating Model 3 without under Sovietinfluence Eastern states European a larger significant and coefficient estimate thelog of energy for per consumption capita,but negative does notchangeourresults factors. withrespect theimpact international to of 57. We have also considered noneof which international factors researchers, emphasized other by for Pevehouse2002a and 2002b argues factors. changedour mainresults theimpactof international thatleadersin new democracies to policiesand assuage relyon international organizations "lock-in" of eliteswhofearunmitigated of density interpopulism. AddingPevehouse'smeasure thedemocratic with national inconsistent his expectations, somewhat (not suggestorganizations shown)yieldsresults to ing thattransitions democracyare more likelyin autocraciesconnectedto highlydemocratic democinternational but do organizations, international organizations notappearto helpto consolidate often the role of the CatholicChurch,whichhistorically racy.Othershave highlighted changing of after denounced and autocratic butbecamean activepromoter democracy rule democracy supported the Second VaticanCouncil,1962-65; see Huntington seems more thatdemocracy 1991. We found is in the Council(although difference not the likelyto endure Catholicsocietiesafter Second Vatican but seemedless frequent. statistically significant), transitions actually

International Context Democratization927 of of model As alternative ablyinvolvea veryhighnumber parameters."8 a simpler of timedependence, introduced covariates we thetimethatcountwo counting trieshave remained to democracies and autocracies, respectively, the first-order Markovmodel.The results thisestimation shownas Model 2 in Table 1. of are Consistent withtheories democratic of the in consolidation,59 estimates therow

less likelythelongercountries have remained the democratic. However, survival ratesof nondemocracies not appearto dependon time,as thecoefficient do for TIME AS AUTOCRACY is notdifferent not as from This is perhaps surprising, the 0. thatsharelittle in nondemocracy categories lumptogether regimes quitedifferent common and beyondnotbeingdemocracies maydisplaya greatdeal of instabilof transfers powerto identify autocraity.Usingirregular changeswithin regime and find autocratic evidence particular that become cies,Gleditsch Choung regimes morelikelyto survive longer the The of have heldpower.60 effects theother they do for covariates, however, notqualitatively changewhencontrolling timedepenfactors stillappearsto have dence,and theimpactof regionaland international effects transitions democracy.61 on to important Boix and Stokeshave recently and that Przeworski Limongi'sdismissal argued of modernization increasesthe prospects that is, theory-that thatdevelopment autocracies becomedemocratic-isan artifact limiting will of their datato a post1951 sample.62 Since manycountries thatdevelopedpriorto WorldWar II had turned democratic before1951,a samplebased on thisperiodmayunderalready statethe role of development the evolution democracy the first in in of wave. one a ask the Accordingly, might whether resultsshownherereflect shorttime can estimates energy of period.The COW project's consumption serveas a proxy foreconomicwealthprior 1945.63 Model 3 in Table 1 displaystheresults to for our model whenestimated the fulltimeperiod1875 to 1998, usinglogged for ^ thanGDP percapita.As can be seen,the energy consumption capitarather per estimate the log of energy for but consumption capita is negative, stillnot per More importantly, coefficient for the estimates the other statistically significant. covariates remain the of and consistent, ourconclusions regarding importance the do context not changewhenwe look at theextended regionaland international timeperiod. ensure To that Eastern alone do notdriveour the countries European we our under all states results, reestimated Model 3 excluding theEastern European Sovietinfluence after estimate 1945. This yieldsa larger coefficient for negative
58. 59. 60. 61. Berchtold Raftery and 2002. For example, Gasiorowski Power 1997. and Gleditsch Choung2004. and We also tried nonparametric a modelspecification, for timedependence, but allowing nonlinear

forTIME AS DEMOCRACY to suggest thattransitions autocracybecome increasingly

this did not notably improve on the linear specification.

62. Boix and Stokes2003.

63. We use energyconsumptionfiguresfromthe COW National MilitaryCapabilities data and extend the currentdata (Version 2.1) beyond 1992 with estimates predicted from a linear regression of a

country's loggedGDP percapitaand a timetrend.

928 International Organization the log of energy with per consumption capitabut does notchangeour results to of factors. respect theimpact international We have shownthattheeffect differences theregional on in context transiof covaritionprobabilities even whentaking intoaccountcountry-specific persists atesandcommon will trends. probability an autocracy becomea democracy The that increasesmarkedly moreof its neighboring or as statesare democracies experiof ence transitions democracy. to estimate the Figure5 displaysa nonparametric effects over differences theproportion neighboring of in democracies, marginal medibased on Model 3 in Table 1, keeping value of other variablesat their the ans. Substantively, translates a middle-income this that to country does notexpefor riencea civil war and has been an autocracy conflict without experiencing overthree of decades.The solid lineindicates predicted the probabilities a transiThe tionto democracy whenthere a transition a neighboring in is country. dashed line indicates transition the transia for probabilities a case without neighboring in tion.As can be seen,thetransition for autocracy a given probabilities a typical states of low,well below .015, whena smallproportion neighboring yearremain are democracies-toward leftside of the horizontal axis-and thereare no the in states exceeds transitions neighboring of states. Whentheproportion neighboring The the increase one-half, however, transition probabilities quitedramatically. like-

.30-

2 .20 c

.25.20.15.10-

>11Neighboringtransition ----- No transition neighboring

0,

.05--------- -----------------

.2

.8 .4 .6 ofneighboring democracies Proportion

FIGURE5. Transition probabilities by regional context,holding other covariates at the median

Context Democratization929 of International status TABLE Predicted 2. versusobserved regime


Predicted regime Model 1 Model 2 Model 3

Observedregime Autocracy Democracy Autocracy Democracy Autocracy Democracy Autocracy Democracy 2016 41 72 4029 2016 41 72 4029 2865 64 96 5762

of lihoodof a transition democracy to exceeds.10 whenmorethan75 percent the in countries the whenother are states democracies is evenhigher and neighboring to transitions democracy. regionexperience to from model refer the likelihood the Recall thatthe predicted probabilities of transitions a givenyear,and thatthe likelihoodof a transition in occurring over a longertime period will be higher.For an autocracythathas a modin of to likelihood transition democracy anyone givenyear erately highpredicted its domesticand regionalattributes example,.1), the impliedlikeli(for given hood thatit will remainan autocracy fiveyearsis (1 - .1)5, whichis less for than.6. of Table 2 comparesthe observedregimesto the predictions the model.The 98.1 to themodelsrangesfrom of observations classified correctly by percentage varies classified 98.2 percent. in The shareof democracies the samplecorrectly transition the from More importantly, predicted 96.55 to 96.75 percent. probabilthan itiesare muchhigher thecases wherewe observetransitions forthesets for of autocracies democracies general. in and to theseresults lend strong Overall, support our claims aboutthe salienceof locationand the the regionalcontext regimechanges.Knowinga country's on There characteristics surrounding of considerable entities power. predictive yields context to is a marked forcases to changein wayssimilar their regional tendency states. connected overtime, transitions one country and often in spilloverto other the betweencountries, claim Given such evidenceof dependence and diffusion transithatregimechangeis entirely randomshouldnotbe accepted.Although under some morecommon tionsto democracy relatively are are clearly rare, they to the conditions thanothers. This undermines claim thattransitions democracy are random events.64
and 64. See Przeworski al. 2000; and Przeworski Limongi1997. et

930 International Organization

Conclusion
related Our results are attest how theprospects democracy notexclusively to for and events. to domesticattributes are also affected external conditions but by similar of Democratization cannot seenexclusively a result functionally be as proevents international cesses unfolding within each country. Rather, independently consistent and processesappearto exerta strong on influence democratization, with argument external our can that influences changetherelative powerof actors forparticular instituand groupsas well as the evaluations relative or payoffs tional Domestic aredeeplyaffected whatgoes by arrangements. political processes events on in neighboring societies,even if the specific ways in whichexternal Diffusion influence transitions from context context. to amongstates vary processes and there a is influence distribution democracy theinternational the of in system in of a association institutions theextent democracy and between country's strong but within thesurrounding similar Notonlyareregimes regions, region. generally A a there also a strong is for to convergence. tendency transitions impart regional will be thata country of priorregionalconflict decreasesthe likelihood history democratic. and involvesomeelement surprise their of timing maynotbe Manytransitions context suchas in external to predict other Our fully changes predictable. ability foreto conflict peace is also limited best,anditmaybe difficult accurately and at we can still cast how regions likelyto evolve overthenearfuture. are However, conditional on makeinferences of aboutan increasein thelikelihood transitions, influence features the that transitions other in and international events states shapthe to it for ing theprospects democratization. Although is difficult fullyspecify andshowhowinterfullrange possible of micro-level of processes democratization national thesein a modelat theaggregate factors influence level,it seemstheoretof or from independent treat domestic the arenaas isolated to icallyinappropriate to theinternational context morepermeable changes is context. Since theregional on influences democin theshort international term thansocioeconomic factors, We "social requisites." do not as racyare likelyto be as important thedomestic whichtranthink it willbe feasible sortthrough multiple to the that through paths in than others advance. sitions one maycomeaboutandselect avenueas morelikely institutional that our allow us to firmly However, results changeis rejecttheidea and driven unaffected regional international and by by entirely domestic processes and context assumethattranevents.It makelittle sense to excludetheregional whentheregional sitions democracy random determined to are and exogenously in transition context influence processes. appearsto exertan important, dynamic

References
Alesina, Alberto, Enrico Spolaore, and Romain Wacziarg. 2000. Economic Integration and Political Disintegration.American Economic Review 90 (5):1276-96.

International Context Democratization931 of


Gerard. 2002. The SourcesofDemocratic Consolidation. Alexander, Ithaca,N.Y.: CornellUniversity Press. Gabriel and SidneyVerba.1963. The CivicCulture: and in PoliticalAttitudes Democracy Almond, A., Five Nations.Boston:Little, Brown. MichaelR., AdamPrzeworski, Cheibub, Fernando PolitJose and Alvarez, Limongi.1996.Classifying ical Regimes.Studiesin Comparative 31 International Development (1):3-36. Garton.1999. TenYearsAfter. New York ReviewofBooks 46:16-19. Ash,Timothy Institutions: and and Barzel,Yoram, EdgarKiser.1997.The Development DeclineofMedievalVoting A Comparison Englandand France.EconomicInquiry (2):244-60. of 35 London:Sage. and Beetham, David, ed. 1994. Defining Measuring Democracy. Modelsof David Epstein, SimonJackman, Sharyn and O'Halloran.2001. Alternative Beck,Nathaniel, Time-Series Cross-Section Models: The Exampleof StateFailure.PaperpreDynamicsin Binary sented the 18th at AnnualMeeting theSocietyforPoliticalMethodology, Atlanta. of July, Distribution Model forHighTransition 2002. The Mixture Berchtold, Andre,and AdrianRaftery. OrderMarkovChainsand Non-Gaussian Science 17 (3):328-56. TimeSeries.Statistical Press. and Redistribution. Boix, Carles.2003. Democracy University Cambridge Cambridge: WorldPolitics 55 (4): Boix, Carles, and Susan C. Stokes. 2003. EndogenousDemocratization. 517-49. A. American of and theTiming Development. Bollen,Kenneth 1979. PoliticalDemocracy Sociological Review44 (4):572-87. in Studies Interand . 1990.Political Traps.Comparative Democracy: Conceptual Measurement nationalDevelopment (1):7-24. 25 Bueno de Mesquita,Bruce,Alastair 2003. The and D. M. Smith, Randolph Siverson, James Morrow. Mass.: MIT Press. Logic ofPoliticalSurvival. Cambridge, and AuthoriCasper,Gretchen, MichelleM. Taylor.1996. Negotiating Democracy:Transitions from tarianRule. Pittsburgh, Penn.:University Pittsburgh of Press. and Level: Problems Definition at Deutsch,Karl W. 1954. Political Community theInternational of Measurement. GardenCity, N.Y.: Doubleday. AnnArbor: Diehl,Paul F., andGaryGoertz.2000. Warand Peace inInternational University Rivalry. of Michigan Press. Politics 52 World Renske. 2000. Reassessingthe ThreeWaves of Democratization. Doorenspleet, (3):384-406. the of on Galton,Francis.1889. Comment E. B. Tyloron a Methodof Investigating Development Institutions: and InstiAppliedto Laws of Marriage Descent.Journal theRoyalAnthropological of tute18:268-9. MarkJ.1995. EconomicCrisisand PoliticalRegimeChange:An EventHistory Gasiorowski, Analysis. American PoliticalScienceReview89 (4):882-97. MarkJ.,and Timothy Consolidation Power.1997. Institutional Gasiorowski, Design and Democratic in theThird World.Comparative PoliticalStudies30 (2):123-55. D. at of Gastil,Raymond 1985. The Comparative Survey Freedom:1985. Freedom Issue 82:3-16. L. and Jonathon Moses. 1996. Democracyand Peace: A More W. Gates, Scott,Torbjorn Knutsen, View.Journal Peace Research33 (1):1-10. Skeptical of MichaelT., and MarkE. J.Newman.2004. Diffusion-Based MethodforProducing Gastner, DensityEqualizingMaps. Proceedings theNationalAcademy Sciences 101 (20):7499-504. of of InteKristian Skrede.2002a. All International PoliticsIs Local: TheDiffusion Conflict, Gleditsch, of and Ann Press. of gration, Democratization. Arbor: University Michigan . 2002b. ExpandedDyadic Tradeand GDP Data, 1946-92. Journal Conflict 46 Resolution of (5):712-24. . 2004. A Revised List of Wars Betweenand Within States,1816-2001. InterIndependent
national Interactions 30 (4):231-62. . 2005. Expanded Population Data, 1816-2002. Data set, Department of Government,University of Essex.

932 International Organization


Kristian and Jinhee Choung.2004. Autocratic Transitions Democratization. and L. Gleditsch, Skrede, at StudiesAssociation, March,MonPaperpresented the45thAnnualMeetingof theInternational treal, Quebec. Kristian and Gleditsch, Skrede,and Michael D. Ward.1997. Double Take: Reexamining Democracy in 41 Resolution (3):361-83. Autocracy ModernPolities.Journal Conflict of . 2001. MeasuringSpace: A MinimumDistance Database. Journalof Peace Research 38 (6):739-68. Robert Norman, DorwinCartwright. Z. Models: An Introduction 1965. Structural and Frank, Harary, to theTheory DirectedGraphs.New York:Wiley. of Samuel P. 1991. The ThirdWave:Democratization theLate Twentieth in NorHuntington, Century. man:OklahomaUniversity Press. ThirdWave withthe PolityIII Data. Keith,and Ted R. Gurr.1995. Tracking Jaggers, Democracy's Journal Peace Research32 (4):469-82. of 2003. Democratic and Megan L. Shannon. Kadera,Kelly M., MarkJ.C. Crescenzi, Peace, Survival, and Warin theInternational American Journal PoliticalScience47 (2):234-47. System. of in Sikkink. 1999.Transnational and Keck,Margaret and Kathryn E., AdvocacyNetworks International Social ScienceJournal (1):89-101. 51 RegionalPolitics.International Revolution. Public Choice of Fires:A Theory Unanticipated Kuran,Timur.1989. Sparksand Prairie 61 (1):41-74. M. N.Y.: Anchor. Lipset,Seymour 1960. PoliticalMan. GardenCity, Susanne.1994. The Dynamics Informational of Cascades: The MondayDemonstrations in Lohmann, 1989-91. World Politics47 (1):42-101. Leipzig,East Germany, Institutional and 2002. Democratic Mansfield, Edward,and JackSnyder. Transitions, Strength, War. International 56 Organization (2):297-337. Eric L. Krueger, StevenC. Poe. 1998. Testing Models of U.S. ForeignPolicy: and Meernik, James, Aid Duringand After Cold War.Journal Politics60 (1):63-85. the Foreign of Sara McLaughlin, in ScottGates,andHavardHegre.1999.Evolution Democracy-War Mitchell, Dynamics. Journal Conflict Resolution (6):771-92. 43 of 1993. Social Originsof Dictatorship and Democracy:Lord and Peasant in the Moore,Barrington. Boston:Beacon Press. Makingof theModernWorld. EdwardN. 1988.Democracy, American and EconomicDevelopment, IncomeInequality. SociMuller, ological Review53 (1):50-68. . 1999. Capitalism, N.J.:Princeton and Ralph's Pretty Good Grocery. Princeton, Democracy, Press. University . 2000. The Banality 'EthnicWar.'International of 25 Security (1):42-70. EdwardN., and Mitchell Seligson. 1994. Civic Culture The Questionof and Democracy: A. Muller, Causal Relationships. American PoliticalScienceReview88 (3):635-52. C. and 1986. Transitions Authoritarian O'Donnell, Guillermo, from Philippe Schmitter, L. Whitehead. Rule. Baltimore, Md.: Johns Press. HopkinsUniversity MichaelD. Ward, David Reilly, Jordin Cohen,David S. Brown, S. John, O'Loughlin, CoreyL. Lofdahl, Kristian Gleditsch, MichaelShin. 1998. The Diffusion Democracy, S. and of 1946-1994.Annalsof theAssociation American 88 of Geographers (4):545-74. PoliticalScienceReview American and Olson,Mancur.1993.Dictatorship, Democracy, Development. 87 (3):567-76. Studiesin Comparative Pamela.2000. Women'sSuffrage theMeasurement Democracy. in of Paxton, International 35 Development (3):92-111. Jon International and from Outside-in? the Pevehouse, C. 2002a. Democracy Organizations Democratization. International 56 Organization (3):515-49. of -. 2002b.With LittleHelp from Friends? and a My RegionalOrganizations theConsolidation American Journal PoliticalScience46 (3):611-26. Democracy. of In Adam. 1988. Democracy a Contingent and Outcome Conflicts. Constitutionalism of as Przeworski, edited Jon Press. Elster RuneSlagstad, and 59-80. Cambridge: Democracy, Cambridge University by

International Context Democratization933 of

2000. DemocJoseAntonio and Przeworski, Adam,MichaelE. Alvarez, Cheibub, Fernando Limongi. PoliticalInstitutions Well-Being theWorld, 1950-1990. Cambridge: and in racyand Development: Press. Cambridge University Journal Przeworski, Adam,andFernando of Limongi.1993.PoliticalRegimesandEconomicGrowth. EconomicPerspectives (3):51-69. 7 . 1997. Modernization: Theories and Facts. World Politics49 (2):155-83. Randle,Michael. 1991. People Power: The Buildingof a New EuropeanHome. Stroud, England: Hawthorn. L. andRegimeTransitions, 1825-993.Journal Factors Ray,James 1995. GlobalTrends, State-Specific ofPeace Research32 (1):49-63. Dan. 2001a. WhyNATO Enlargement International Does Not SpreadDemocracy. 26 Reiter, Security (4):230-35. Journal Politics63 (3):935-48. . 2001b. Does Peace Nurture of Democracy? Karen L. 1991. The PoliticalImpactof EconomicCrisis in Latin Americain the 1980s. Remmer, American PoliticalScienceReview85 (3):777-800. WilliamI. 1996. Promoting and Hegemony. U.S. Intervention, Robinson, Globalization, Polyarchy: Press. Cambridge: Cambridge University Dani. 1999. WhereDid All the Growth and Go? External Rodrik, Shocks,Social Conflict, Growth 4 Collapses.Journal EconomicGrowth (4):385-412. of D. and Dietrich, Rueschemeyer, EvelyneHuberStephens, John Stephens.1992. Capitalist Development and Democracy. of Chicago:University ChicagoPress. ThomasC. 1971. DynamicModels of Segregation. Journal Mathematical Schelling, of Sociology1 (2):143-86. Ian. 2003. TheStateofDemocratic Press. N.J.:Princeton Shapiro, Princeton, Theory. University Ron Pagnucco,and GeorgeA. Lopez. 1998. Globalizing HumanRights: The Workof Smith, Jackie, Transnational HumanRights 20 NGOs in the 1990s.HumanRights Quarterly (2):379-412. Etel. 1998. RegionalOrdersat Century's Dawn: Global and DomesticInfluences Grand on Solingen, N.J.:Princeton Press. Princeton, Strategy. University in to Dominoes:Diffusion Starr, Harvey.1991. Democratic Approaches theSpreadof Democracy the International Journal Conflict Resolution (2):356-81. 35 System. of WilliamR. 1996. Democracy Peace: Putting CartBeforetheHorse?International and the Thompson, 50 Organization (1):141-74. Tatu.1990. TheProcessofDemocratization. New York:CraneRussak. Vanhanen, Kazuo. 1991. EventHistory Yamaguchi, Park,Calif.:Sage. Analysis. Newbury

Вам также может понравиться