Scaling Down: The Subnational Comparative Method

Abstract

Subnational units of analysis play an increasingly important role in comparative politics. Although many recent studies of
topics such as ethnic conflict, economic policy reform, and democratization rely on comparisons across subnational political
units, insufficient attention has been devoted to the methodological issues that arise in the comparative analysis of these
units. To help fill this gap, this article explores how subnational comparisons can expand and strengthen the methodological
repertoire available to social science researchers. First, because a focus on subnational units is an important tool for increasing
the number of observations and for making controlled comparisons, it helps mitigate some of the characteristic limitations
of a small-N research design. Second, a focus on subnational units strengthens the capacity of comparativists to accurately
code cases and thus make valid causal inferences. Finally, subnational comparisons better equip researchers to handle the
spatially uneven nature of major processes of political and economic transformation.

Full text

Snyder 93
Scaling Down:
The Subnational Comparative Method*
Richard Snyder
Subnationalunitsofanalysisplayanincreasingly importantroleincomparat ive
politics.Althoughmanyrecentstudiesoftopicssuchasethnicconflict,economic
policyreform,anddemocratizationrelyoncomparisonsacrosssubnationalpoliti-
calunits,insufficientattentionhasbeendevotedtothemethodologicalissuesthat
ariseinthecomparativeanalysisoftheseunits.Tohelp fillthisgap, thisarticle
explores how subn ational com parisons ca ne xpand and streng then the method -
ologicalrepertoireavailabletosocialscienceresearchers.First,becauseafocuson
subnationalunitsisanimportanttoolfor increasingthenumber of observations
andformakingcontrolledcomparisons,ithelpsmitigatesomeofthecharacteristic
limi tations of asm all-N res earch desi gn. Second, a fo cus on subnat ional unit s
strengthensthecapacityofcomparat iviststoaccuratelycodecasesandthusmake
validcausalinferences.Finally,subnationalcomparisonsbetterequipresearchers
tohandlethespatiallyunevennatureofmajorprocessesofpoliticalandeconomic
transformation.
Weliveinanageofdecentralization.Theeconomicandpoliticalmega-
trendsthatarethehallmarksofthelasttwodecades—free-marketre-
forms and democrati zation—have unleashed powerful centr ifugal forces. In
theadvancedindustrialcountries,forexample,free-marketpolicyreformswere
implementedinconjunctionwitha“devolutionrevolution”thattransferred
authorityandresourc esfromcentraltolocalgovernments.Guidedbythe
doctrine  of “fiscal fe deralism ,” the World Bank and othe r development
NGOsexportedthe  devolution revolutiontopoor countries byfervently
promotingthe  decentralizationofpolicy making andgovernment services.
Combinedwiththeworldw idemovetowardderegulated,openeconomies,these
decentralizingreformsundercutthecapacityofmanycentralgovernmentsto
coordinateeconomicactivity.
TheThirdWaveofdemocratizationthatswepttheglobeduringthelastquar-
tercenturyalsogeneratedpow erfuldecentralizingpressures.Thedemocrati-
Richar dS nyderis assi stant professor of poli tical science at theU niversity ofIl linois at Urbana-Champaign .
HeistheauthorofPoliticsafterNeoliberalis m(2001).Hi sarticlesonregimechangeandthepoliticaleconomy
of development have appeared in World Politics, Comparative  Politi cs, Journal of Democr acy, and Brit ish
Journal of Political Science.
Studi esin Comparative Intern ational Developmen t,Spr ing 2001,Vol.36, No. 1, pp.93–110.

94 Studies in Comparative International Development / Spring 2001
zationofwhatwereonceonlynominallyfederalsystemsincasessuchasRus-
siaand Mexico shifted  the distribution of pow er  between national and
subn ationa l govern ments  inwaysthatgreatly strengthened thelatter.More-
over, democratization wasassociate d with bo th the“federalization ” offor-
merlyunitarysystems(e.g.,Spain)and thebreakup offederalsystems (e.g.,
Czechoslovakia,Yugoslavia,SovietUnion).Anextremeformofdecentraliza-
tionisseenincaseswhereachangeofpoliticalregimeresultedinthecollapse
ofthecentralstateandtheemergenceofanarchicalwarlordpolitics(e.g.,Congo/
Zaire,Liberia,SierraLeone,andSomalia).
Inresponseto majorcentrifugal  trends such as  these,  co mparativists
havef o c u sedincreasingattentiononsubnational unitsofa n a lysis.Thus
manyrecentstudies oftopicssucha s e c onomicpolicyreform,democra-
tization,a n dethnicconflictrelyo nco m pari sonsacrosssubnationalpo-
litica l units.1Despitetheproliferationofworksthatemploysubnati onal
comparisons,insufficientat tentionhasbe e n devoted totheme t hodologi-
calissuesthatariseint heco m p a rativeanalysisofsubn ationalunits.2To
help fill this g ap , this articleexplores how  the subnatio nalcomparative
methodcanexpand andstrengthenthemethodologicalrepertoire available
tocomparativeresearchers.
Iarguethatthesubnationalcomparativemethodofferthreekeyadvantages.
First,itcanhelpbettermanagesomeofthecharacteristiclimitationsofasmall-
Nresearchdesign.Theuseofsubnationalunitsisanespeciallyeffectivestrat-
egyforincreasingthenumberofobservationsandthusmitigatingtheproblem
of“manyvariables,small N.”3Moreover, afocusonsubnational unitscan
makeiteasiertoconstructcontrolled comparisonsthatincreasetheprobabil-
ityofobtainingvalidcausalinferencesinsmall-Nresearch.
Second,subnationalcomparisonsstrengthenthecapacityofcomparativists
toaccuratelycodetheircases.Comparativiststoooftenrelyonnational-level
meansandaggregatedatawhenstudyingcountrieswithhighdegreesofinter-
nalheterogeneity. Thistendency tounr efle ctively g ravita te tow ard national-
leveldataandnationalunitsofanalysi s—atendencythatSteinRokkan(1970)
sometimeagocalled“whole-nationbias”—hascontributedtoamiscodingof
casesthatcandistortcausalinferencesandskeweffortsattheorybuilding.4A
greater  se nsitivityto w ithin-nat ion  variation  and c om plexity c an  he lp
comparativistsavoidthesepitfalls.
Third,afocusoncomparingsubnationalunitsbetterequipsustohandlethe
spatially unevennatureofmajorprocessesofpoliticalandeconomictransfor-
mation.Processes suchasde mocr atiz ation  andeconomic reformoftenhave
varied effectsacross the territoria lly-defined subunits ofapolitical system.5
Becausethesubnationalcomparativemethodmakesiteasiertoseethiswithin-
nationvariation,itcontributestoamoreadequatedescription ofcomplexpro-
cessesofchange.Inadditiontoprovidingasalutaryincreaseinourabilityto
accurately describe complex processes,afocusonsubnational unitshas im-
portantimplicationsforhowwetheorizesuchprocesses.Disaggregatingcoun-
tries alongterritoriallinesm akes it possi ble to expl ore the dynamic linkages
am ong the distinct regions andlevelsofapolitical system.Analyzing these

Snyder 95
linkagesisanindispensablestepforunderstandingandexplainingfundamen-
talprocessesofpoliticalandeconomicchange.
The Subnational Comparative Method and Research Design
Thesubnational comparativemethodhastwokeystrengths withregard to
resea rchdesign:(1)itcanserveasapowerfultoolforincreasingthenumber
ofobservations,and(2)itmakesiteasiertoconstructcontrolledcomparisons.
Increasing the Number of Observations
Focusingonsubnatio nalunitscanprovideanefficientwaytoincreasethe
numberofobserva tionsinthecontextofasmall-Nresearchdesign.Asem-
phasizedbyGaryKing,RobertKeohane,andSidneyVerba(1994:208,see
also30–31),ourtheoriesusuallyhaveobservableimplicationsatmanylev-
elsofanalysisand“whatmayappeartobeasingle-casestudy,orastudyof
onlyafewcases,mayindeedcontainmanypotentialobservations,a tdiffer-
entlevelsofanalysis,thatarerelevanttothetheorybeingevalua ted.”6Thus
atheoryoriginallyfocusedonanationalcasecanpotentially betestedinsub-
unitsofthatcase:suchasstates,counties,cities,regions,oreconomicsectors
(K inge tal. 1994: 21 9).7Moreover,moving downtoasubnatio nal levelof ob-
servation hasthe additional advantage ofnot requirin g trave l to learn about
anothercountry.
Althoughthestrategyofincreasingthenumberofobservationsbyshifting
toalowerlevelofaggregationhasconsiderablemerits,itshouldbeusedwith
caution.First,whenmakingcross-levelinferences,theresearchermustbecare-
fultochooselower-levelunits thatareappropriateforreplicatingthehypoth-
esisunderconsideration.Thatis,thelower-levelunitsshouldbeonesinwhich
“theprocessentailedbythehypothesiscantakeplace”(Kingetal.1994:221).8
Thus,  it would be inappropriate to study subnational political unitsto test a
theory about macroeconomic policy making.  Second,observations atthe
subna tiona l level do not necessarilyprovideindependen ttests of the theory,
becausestates,counties,andothersubunitswithinapoliticalsystemareoften
interconnected,notfreestanding.Suchpotentialinterconnectionsamongcases
can lead to “Galton’sproblem” andthus require caution against overstatin g
thecertaintyofconclusionsbasedontestsusingsubunitsdrawnfromthesame
political system (Kingetal.1994: 222) .9Moreover,it isag ood pra ctice to
analyzethesourcesofinterconnectionsamongsubunits(e.g.,policydiffusion,
borrowing,migration)inanefforttodiscoverpotentiallyconfoundingomitted
variables.10
Making Controlled Comparisons: Strategies of Subnational Analysis
A  fo cus o n s ubn at ional u ni ts c an g re atly s tr engthe n th e a bi lity  of
comparativiststoestablishcontroloverpotentialexplanatoryvariables(Linz
anddeMiguel1966;Lijphart1971:689–90).Inthisregard,twostrategiesof
subnational analysiscanbedistinguished:(1)within-nationcomparisonsthat

96 Studies in Comparative International Development / Spring 2001
focus on subnational  caseswithin asingle cou ntry; a nd (2)between-nation
comparisonsthatfocusonsubnationalcasesacrosscountries.Moreover,these
twostrategiescanbefruitfullycombinedinasinglestudy.
Within-NationComparisons.Subnationalunitswithinasinglecountrycan
oftenbemoreeasilymatchedoncultural,historic al ,ecological,andsocioeco-
nomicdimensions thancannationalunits.Forexample,myownworkonthe
politicsofeconomicpolicyreforminMexicoemployswhatIcalla“onesec-
tor, many places”strategythatfocuses on four of Mexico’s m ost imp orta nt
coffee-producing state s (Snyder1999a, 2001).11 Because thefour sta te s—
Oaxaca,Guerrero,Chiapas,andPuebla—arealllocatedinsouthern Mexico,
havelargeindigenouspopulations,andareamongthepoorestinthecountry,I
amabletocontrolforecological,cultural,andsocioecono micconditionstoa
fargreaterextentthanisusuallypossibleinstudiesthatcomparenationalunits.
Bymakingiteasiertoestablishcontroloversuchnon-politicalfactors,afocus
onsubnationalunitshelpspinpointhowvariationinpoliticalinstitutionsshape s
economicperformanceandpolicychoice.
Itshouldbeemphasizedthatwithin-nationcomparisonsdonotnecessarily
improve ourability toholdconstant cultural, histor ical,  ecological,and  so-
cioeconomicconditions:theremaybeasmuch,ifnotmore,variationwithin
countriesasbetweenthem.Forexample , statesinsouthern  Mexico  may
havemore in common withsubnational units inneighboring Guatemala
than with  statesinnorthern Mexico.Thus,asB . G uyPeters (1998: 35)
cautions,comparativistsshouldbecarefultoavoidbeinglulledintoa“false
senseofsecurity” sim ply becausethey areusing subnational unitswithina
singlecountry.12
Between-NationComparisons.Comparingsimilarsubnationalunitsacross
distinctnationalunitsmaybeamorepowerfulstrategyformakingvalidcausal
inferences thancomparingnational units. Forexample, Linz and deMiguel
(1966:269)argue thatacomparisonbetween“advanc edandbackwardsec-
tions”ofSpainandItalywithsimilarculturalandsocioeconomicfeatures
isanespeciallyeffective waytoexplorehowthedifferentpoliticalinsti-
tutionsof the twocountries influencemembershipinvoluntaryassocia-
tions.  Similarly,O’Donnell(1973: 21) proposesa“cross-modernareas”
comparisonthatjuxtaposes themostdevelopedregionsandsectorsofBrazil
andArgentina.13
Byanalyzingcontiguoussubnationalunitsacrossnations,scholarsmaybe
abletodesign between-nationcomparisonsthatachieveanespeciallystrong
degree ofcontrolover c ultural,historical,andecologicalconditions. Forex-
ample,studiesoftheimpactofcolonialis minAfricahavefocusedonadjacent
villagesorregionslocatedonoppositesidesofaninternationalbordersepa-
ratingaformerFrenchfromaformerBritishcolony(Miles1994;Firmin-Sell-
ers 2000) . Inhis c lassic  article on the c ompa rative method,Arend Lijphart
(1971)endorses RaoulNaroll’s(1966) suggestion that inorder to study the
effectsofpresidentialandparliamentarysystems,ratherthancomparingthe
UnitedStatesandBritain,  itmaybe a m orefruitfulstra tegytocompare
NorthDakotaandManitoba.Suchacomparisonwouldminimizevariation
infactorssuch aslevelsofeconomicdevelopmentandeducation.  I nhis

Snyder 97
analysis ofagrarianradicalismintheNorthAmericanw h eatbelt,Lipset
(1950)comparesthecasesofSaskatchewanand  NorthDakota,whichhe
(1950:215)describesasnearreplicasineconomicandgeographicterms.14
Thisbetween-nation com parisonhighlights how theNew Dealundermined
agrariansocialistmovementsinthewheatbeltoftheUnitedStates,whereas
theabsenceofareformistfederalgovernmentprograminCanadaduringthe
depressionprovidedamorefavorablecontextforagrarianradicalism(Lipset
1950: 17, 119–20).In h is study of po litica l c onflic t in the Basque country,
Linz (1986: 372–98) also usesbe tween-nationcompariso nsbyanalyzing the
Spanish and FrenchBasque regions. Thisresearch design allowshim to explore
howindividuals belonging toacommon ethnicgroup (i.e., Basque)yet living in
differentcountriescanhavesharplycontrastingunderstandingsoftheirethnic
andnationalidentities.15
Thewithin-nationandbetween-nationstrategiesofsubnationalanalysiscan
be c om bined in a  single st udy.For  exam ple , i na d ditiontocomparing
SaskatchewanandNorthDakota,Lipset(1950:216–19,121–4)alsocompares
Saskatchewanwithits neighbo ring provinces ofA lber ta andManitoba.In
Albe rta , w her e the  eco n om y an d so cia l  s tru ctur e cl o sel y re sem b le d
Saskatchew a n’s, Lipse t (1950:  216)  found t hat “far m ers, l ike thoseof
SaskatchewanandNorthDakota , arequic k toreactpoliticallywhenathreat
ar ises.” By contrast, in Manitoba, w hichdiffere d fromSaskatchewanalong
manyofthekeydimensionsthatLipsethypothesizedwouldinfluencelevels
ofpoliticalactivism,theagrarianpoliticalmovementwasweak.16 Thusthese
subnationalcomparisonsservetostrengthen Lipset’soverall argumentabout
thecausesofagrarianradicalism.
Althoughthesevariousstrategiesofsubnationalanalysisserveaspowerful
toolsformakingcontrolledcomparisons,theysharealimitation:thepotentia l
fordiffusionandborrowingamongsubnational units inasingle country and
amongcontiguoussubnationalunitsinneighboringcountriescanmakeitdif-
ficulttoachieveindependentobservationsandtests.17 Onetechniqueformiti-
gatingthetradeoffbetween(1)theabilitytoestablish controloverpotentia l
explanatoryfactorsand(2)theabilitytoachieveindependenceamongthecases
istocombinewithin-nationcomparisonsandbetween-nationcomparisonsof
non-contiguoussubnational units in orde r to reduce the potential ef fects  of
diffusion(alongthelinesofcomparing“advancedandbackward”sectionsof
SpainandItaly,asproposedbyLinzanddeMiguel(1966:269),andthe“cross-
modernareas”comparisonproposedbyO’Donnell(1973:21).18 Suchadual
strategycouldhelp ma ximize both controlove rp oten tial exp lana tor yfactors
andindependenceamongthecases.
The Subnational Comparative Method and Measurement: Coding Cases
Al l compara tive researchfacesthechallenge of coding cas es: the analyst
mustemployimplicitorexplicitindicatorsforcategorizingthecasesincluded
inthestudy.Subnationalcomparativeanalysiscanhelpmitigatetwoformsof
biasthataffecthownationalcasesarecoded.Thefirstinvolvestheinappropri-
ateuseofnational-levelaveragesinstudyingcountrieswithahighdegreeof

98 Studies in Comparative International Development / Spring 2001
internalheterogeneity.Thesecondkindofbiasresultswhentheattributesofa
well-studiedorespeciallysalientsubunitareimproperlyusedtocharacterize
thenationalcaseasawhole.
Mean-Spirited Analysis: Inappropriate Uses of National Averages
Cross-na tionalresearchoftenreliesonnational- l evelmeanstocodecases.
Theuseofnationalmeanshasplayedanespeciallyprominentroleinquantita-
tiveresearch.19 However,nationalmeansarealsocommonlyemployedinquali-
tative research—forexample,ac ountr y withalargedeveloped sectoranda
largeunderdevelopedsectormaybecodedas“semi-developed.”Consequently,
althoughtheymaynotexpressnationalmeansinnumericalterms,manyquali-
tativestudiesrelyonwha tcouldbecalled“mean-spiri ted”analysis.
Theproblemsassociatedwithmean-spiritedanalysisaresuggestedbyLinz
anddeMiguel’s (1 966: 271)ob serva tion th at “m any socie ties w ecall semi-
developedonthebasisofanumberofnationalindicesarereallyamixtureof
developedandunderdevelopedsectors.”20 Moreover,intermsofsuchnational
indices,countriesthatareinfact“semi-developed”wouldpotentiallyhavethe
samemeanascontrastingcasesthatarecompositesofdevelopedandunderde-
velopedsectors.Consequent ly,thetwotypesofcasesareindistinguishableif
onereliessolelyonnationalmeans.
O’Donnell(1973)madethispointforcefullyinhiscompar isonofArge ntina
andBrazil,highlightinghowthecodingofcasesbasedonaggregate,national-
level averages masksimportant diff eren ces in the form of a distribution  of
traits(e.g.,normal,bimodal,etc.)aswellasinthedegreeofdispersionofsuch
traits.O’Donnellshowsthatafocusonnational-levelmeansledmanystudies
inthe1960stocodeBrazilasanunderdevelopedcountry.Thiscategorization
obscuredthefactthatBrazil’smodernsectorwasinfactquitecomparablein
siz e toArgentina’s, acasethat lac ked al arge ruralperiphery and w as thus
frequentlycodedassemi-developed(ratherthanunderdeveloped).
Linz andde Miguel’s (1966)  pioneering stu dy of regional differences in
Spain illustrates howcarefully-selectedwithin-nation  comparisons canhelp
guard aga inst the miscoding of anational case. LinzanddeM iguel  (1966:
314–15) used subnational comparisonstoshow that causalrelationshipsbe-
tweenindustrializationandreligiosity thatwereestablishe dusingaggregate,
nationaldatadidnotholdatthesubnationallevel.Byanalyzingregional-level
dataonculture andeconomicstructureinSpain,LinzanddeMiguelshowed
that,contrarytostandardversionsofmodernizationtheory,theleast industri-
alizedregionsofSpainwereverysecular.Moreove r,amongSpain’smostin-
dustrialize dregionstheyfoundawiderangeofvariation:Cataloniawa squite
secular,whereastheBasqueCountrywashighlyreligious.
MyownresearchonthepoliticsofeconomicreformincontemporaryMexico
provides  arecentexam ple ofhow disaggregat ing countries along territorial
linescanimprovethecodingofnationalcases(Snyder1999a,2001).Basedon
highlya ggregatednationalindicatorsofmacro-economicpolicy,suchastariff
levelsandtheoverallsizeofthepublicsector,mostscholarshavecodedMexico
asaninstanceofsuc cessf ul free-market,orneoliberal,policyreform.21 This

Snyder 99
relianceonna tiona l-l evel da ta obscured the  fact tha t theimplementation of
neolibe ralreforms inMexico, rathe r thanunleashing marketforces,actually
re sulte d intheconstructionofnewregulatory institutions atthesubna tiona l
level.22 Looking belowthenational level made iteasiertoseethat theimple-
mentationofneoliberalreform striggered“reregulation ”processesthatresulte d
indifferentkindsofnewinstitutionsformarketregulation,notinthe triumph
offreemarkets.Thusafocusonsubnationa lunitsprovidedabasisforamore
accuratedescriptionandcodingoftheMexicancaseandchallengedthecon-
ventionalviewthattheimplement ationoffree-marketreformsatthenational
levelresultsinderegulated,competitivemarketsacrossanentire country.23
Misusing the Best Known Place: Invalid Part-to-Whole Mappings24
Invalidpart-to-wholemappingsresultwhentraitsorprocessesspecifictoa
well-studied  regionorother subnationalunitareim prope rly e levated to the
sta tus ofanational paradigm. Consequently, national cases are inappropri-
atelycodedasifthewholenationpossessedthesetofattributescharacterizing
aspecificregionorsetoflocalities .25 Suchmiscodingcandistortcausalinfer-
encesandskeweffortsattheorybuilding.
TheliteratureonGermanindustriali zationprovidesanespecially vividex-
ampleoftheconsequen cesofmisusingthebestknownplace.Buildingonthe
pioneeringworkofAlexanderGerschenkron(1962),mostscholarshaveseen
the  Germancase asa paradigmatic instance ofcentralizedindustriali zation
dominatedbylargefirmsandbig,national-le velbanks.Aprovocative recent
bookbyGaryHerrigel (1996) challengesthisconventionalviewby showing
that the industriali zation processthat Gerschenkron a nd othe rs at tribut ed to
Germa ny as awholewas,infact,aregionally-specificformofindustrializa-
tion.Herrigelshowsthatother,less-studiedregionsofGermanyexperienceda
differentkindofindustrialization rooted insmall-scalefirms andbanks.By
situatingtheprocessofindustrializationinaregionally-differentiatednational
framework,Herrigelisableto“recode”Germanindustrializationasabimodal
phenomenon.Whatemergesisareconstructednational compositecharacter-
izedbytwodistinctmodesofindustrialization:centralized,“organizedcapi-
talism”insomeregionsanda“decentralizedindustrialorder”inothers.26
Inaddition to offeringa more valid descriptionofthe Germancase,this
reconstructionofthenationalcompositehasimportantimplicationsforcausal
inference  and theorybuilding. Inlight ofHerrigel ’s analysis, it wouldbe a
mistake to conclude basedontheGermancase tha t anentire c ountr y could
followthepathofcentralized,organizedcapitalism.Rather,Herrigel’srecoding
ofGermanysuggests  thatorganizedcapitalismissuitedonly forspecificre-
gionsandkindsofenterprise s, not entirecountries.Moreover, tothe degree
thatGermany’scentralizedindustria lizationwasnotanindependent, separately
det ermi ned  proc ess,  but w as in fac t cau sally  connected tothe decentralized
industrialorderexaminedbyH errigel,afailuretoanalyzethelatterphenom-
enoncouldleadtoaseriousmisunderstandingoftheformer.Thusashifttoa
territorially-differentiatedframeworkinanalyzingaparadigmatic caseserves
bothtocallintoquestionalongstandingmodelofindustrializationandtoopen

100 Studies in Comparative International Development / Spring 2001
anewtheoreticalagendatha tfocusesonthelinkagesbetweendistinctmodes
ofindustrializationwithinasinglenationalunit.
Insum,mean-spiritedanalysisandthemisuseofthebestknownplace pose
important problems of measurement anddescription:mean-spirited analysis
inappropriately employs highly aggregateddata to code cases,  whereas the
misuseofthebestknownplaceinappropriatelyemploysdisaggregateddatato
makeinferencesaboutnational units.Astheearlierexa mplessuggest,inhet-
erogeneous societies thedistortionsassociated  w ith these twoproblems can
leadtotheinvalid codingofnationalcases.
The Subnationa l Comparative Method and Theory Building:
Taking Territory Seriously
Afinaladvantage ofthesubnationalcomparative  methodconcernshowit
canhelpusbuild theories that explainthe dynamic interconnec tionsamong
the  levels andregions of apolitical sy stem . Di saggr egating countries a long
territorial linesmakesitea sie r toseehowtheconstituentpartsofapolitical
systeminteract.27  This invites us to explore the causa lc onnecti ons amo ngre-
gionsthatexperience diverg ent p atte rns ofchange.Analyzingtheseconnec-
tionscanhelpusgainastrongerunderstandingbothofnationalpoliticsandof
majorprocessesofpoliticalandeconomictransformation.
Within-Nation Variation and National Regimes
A r egiona lly differentiated perspect ive that highlights variation across
subnationalunitsinacountrycanbeessentialforunderstandinghownational
politicalregimeswork.Implementingtheirpolicyagendas—orsimplystaying
inoffice—oftenrequiresnational-levelpoliticianstobuildcoalitionswithdif-
ferenttypesofsubnationalactors.Thusa“center-centered”approachthattreats
thenationallevelasanautonomous,separatesphereandobscurestheconnec-
tionsbetweenactorsintheperipheryandthecentermaymischaracterizethe
strategiccontextin whichnational politicians labor.28 Avividexamplefrom
the lit eratu re on political regim es inLatin Ame rica show s how a foc us on
subnational units canhe lp avoid this  proble m and thus im prove our  under-
standingofpoliticsatthenationallevel.
Popu l i s t Coalitions:  The Missing Rural Dimension.  Populism in Latin
Americahas typically beenviewed as a“metropolitan”phenomenonrooted
inthemosturbanizedandmodernregionsofthecountry.Thestandard inter-
pretationregardspopulismasapoliticalstrategyfavoring“urban-industrial inter-
ests” at the expense of a tr aditional oligarc hy whose power is usua lly seen as
anchoredinrural,peripheralareasofacountry(DiTella1965,1968).
Apath-breakingarticlebyEdwardGibson(1997) challengesthisviewby
showingthattheprototypicalpopulistregimesofJuanDomingoPeróninAr-
gentina  andLázaroCárdenas in Mexicowerein fact based on tw odistinc t
subcoalitions:ametropolitancoalitioncenteredinurbanareasthatgavesupport
to the development strategies of the populist r egime, and aprovincialcoalition
centeredinruralareasthatwasresponsibleforgeneratingtheelectoralmajorities

Snyder 101
thatkeptthepopulistregimeinpower.29 Fromthisperspective,thestandardinter-
pretationofLatinAmericanpopulismasalabor-basedphenome no nwasinsuffi-
cientlyattuned tothelinkages between metropolitanandperipheralregions .
Consequently,populism’smetropolitanfacewasseenasitsonlyface.
Whatiswrongwiththisunderstandingofpopulism?First,suchaperspec-
tivepotentially distortstheforcesthatkeptpopulistgovernmentsinpow er:a
focus limited to themetropolitansideofpopulism canobscurethe fact that
laboralonewaselectorallyinsufficienttosustainpopulistpolicies.Moreover,
ignoringtheprovincialdimensionofpopulist governmentsmiscaststherela-
tionshipbetweenthesegovernmentsandtheold,oligarchicorderthatpreceded
them. Thelopsided viewofpo pulism  as anurban, labor-based phenomenon
eitherpitspopulistgovernments againstperipheral,traditional elitesor,more
simplistically,relegatessuchelite s toamarginalroleinnationalpolitics.By
contrast,abalanced understandi ngofpopulismasacompositeofmetropoli -
tan  andprovincial coalitions provides ave ry d iffer ent,  lessadversarialper-
spectiveonhowpopulistgovernmentsinteractedwithruralelites:Ithighlights
howpopulist governments accommodated and even de pended onprovincial
oligarc hsinorderto“linkpopulismtothetraditional order,giveitcoherence
asanationalelectoralforce,andextenditsreachthroughoutthenationalterri-
tory”(Gibson1997:341).30
Backwaters or Undercurrents? Illiberal Peripheries in Latin America
Focusingonthedynamiclinkagesamongthelevelsandregionsofapoliti-
calsystemprovidesanewwayoflookingattherelationshipbetweencontrast-
ingpoliticalphenomenaobservedatthe“center”(i.e.,thenationallevel)and
atthe “periphery”  (i.e ., the subnational  le vel).  Inthe study of political and
economicliberalizationincontemporaryLatinAmerica,aregionallydifferen-
tiatedframew orkhighlightshowilliberalpoliticalandpolicyregimesthatper-
sist, and evenproliferate,in the peripheralareas ofm any countries maybe
undercu rrent sthatarecausallyintegralcomponentsoftheprocessofliberal-
ization seen atthe center.31 Bycontrast, studiesthat focusonly onpolitics at
the national level e ither  do not “see” these illiberal peripheries, or,alterna-
tively,inappropriatelydismissthemasisolatedbackwaters ,disconnectedfrom
andlef t behind bythe proc esses ofpolitical transf ormation observed atthe
center.Suchacenter-centeredperspective,whichhastendedtodominatere-
centresearchonpoliticalandeconomicliberalizationindevelopingcountries,
canleadustomisunderstandthesefundamentalprocessesasfarmoreuniform
thantheyinfactare.
TheSpatial Unevenness of D emocracy.GuillermoO’Donnell’spioneering
workontheshortcomingsofcontemporarydemocraticregimesinLatinAmerica
highlightshowafocusontheinteraction amongtheterritoriallydefinedcom-
ponentsof a political system helps us bett er un derst and  theprocess ofde-
m o c ra t i z a t i on .  O ’ D on n e l l  a r g u e s  t h a t  p o l i t i c a l l i b e ra l i z a t io n  a n d
democratization areterritoriallyunevenprocesses.Consequently,within
aparticularcountry, therulesandnormsthatdefinethen ational regim e
andstatecannotbeassumedtoapplyequallythro ughoutthecountry.  For

102 Studies in Comparative International Development / Spring 2001
example,inma nyLa tinAme ricancountries“blueareas” c haracterizedby
robu struleoflawcoexistwith“greenareas”wheretheruleoflawispartially
attenuatedand“brown areas” wheretherule of lawisextremely attenuated
(O’Donnell1999:139).
O’Donnell’sproposalthatwedisaggregatecountries intoblue, green,and
browna re as prom ises  asa luta ry inc rea se ind escri ptive accuracy.Moreover,
thisproposalhasprofoundimplicationsforhowwetheorizeaboutdemoc rati-
zation.Byrejectingthenotionthatpoliticalliberalizationisaterritoriallyuni-
formprocess,O’Donnellurgesustoexplorethepotentialcausalrelationships
betweenplacesandunitsthatexperiencedivergentpatternsofchangeinthe
contextofaunified(butnotuniform)politicalsystem.
The“bluing”ofanationalregimemayhaveadirectcausalrelationshipwith
“brow ning” in subnational units.  For example,  incountries with highly-
malapportioned electoral sys tems tha t over represent rural ar eas—a pattern
commonacrossLatinAmerica(SamuelsandSnyder2001;SnyderandSamuels
2001)—theabilityofnational-levelpoliticians to imple ment andconsolidate
democraticreformsmayironic allydependonwinningtheovervaluedsupport
ofsubnational authoritarian  elites.Atthe same tim e, overrepresentationof
subnationalauthoritarianenclavesatthenationallevelcanstrengthentheabil-
ityofthesesubnationalelitestofendoffeffortsbyexternalgroupswhoseek
to reform local politics.A goodexample of this scenar io isthe roleof so-
called“Stalinist”votere servesinsouthern Mexicanstatestha thelpedtheIn-
stitutional Revolutionary Party(PRI)securethepoliticalmajoritiesthatkept
theliberalizingreformersoftheSalinasandZedilloadministrationsinpower
during the1980sand1990s.32 Hence,adialecticalprocesscanresultwhereby
de mocra cy issimultaneously strengthen edat the center andundermined on
theperiphery.33
PatchworkLiberalization:E conomicReforminContemporaryLatinAmerica.
Economic  liberalization,  like democratization,hasalso bee na territorially
unevenprocessincontemporaryLatinAmerica.Recentresearch(e.g.,Snyder
1999a,2001;GibsonandCalvo2000;RemmerandWibbels2000),highlights
howneoliberaleconomicreformshavecontributedtothemaintenanceandeven
proliferationofilliberalpolicyregimesatthesubnational(especiallystateand
provincial) level. U nderstanding this patc hwork pattern r equires af ocus onthe
interconnectionsamongthelevelsandregionsthatcomprisethepoliticalsystem.
Forexample, toexplainhowthedismantling ofamassive federalgovern -
mententerprise inMexicore sultedinthefor mationofnewregulatoryinstitu-
tionsatthestatelevel,Ianalyze the linkages betweenpoliticiansatthestate
andfederallevels(Snyder2001).Myresearchshowsthatstategovernorswere
ableto take controlofpolicyareasabandonedbytheoldfederalgovernment
enterpriseandreregulatewhatfederallawhadderegulatedinpartbecausethe
rulingPRIfacedagrowingthreatfromoppositionpartiesatthenationallevel.
Thuspresident CarlosSalinasdeGortarihadstrongincentivestoallow state
governorstochoosepoliciesthathelpedmaintainpoliticalstabilityandgener-
ateelectoralmajorities,whetherornotthesepoliciesconformedtotheortho-
dox neoliberal prescriptions that he endorsed.This permissive environment
madeitpossible forg overno rs topursue regulatory po licie s thatcut against

Snyder 103
theagendaofneoliberalreformsetatthenational level by the president.In
fact,Salinasdidno t merelytoleratesuch subnational  deviations fr om his
neolibe ralline,heselectivelypromotedreregulationbyvigorouslysupporting
populistgovernorsincertainstates.Thusliberalizingeconomicreformsatthe
na tionallevel andthe proliferation ofilliber alpolicy enclaves atthesubnational
levelwerecomplementaryandcloselyarticulatedcomponentsofasingleprocess:
thelatterhelpedsecurethepoliticalstabilitynecessaryfortheformertoproceed.
GibsonandCalvo(2000)alsoemployasubnationalcomparativemethodto
explaintheterritoriallyunevennatureofeconomicreformsinArgentina.They
showhow electoralrulesheavilyoverrepresentedsparselypopulatedruralar-
easinthenationallegislature.Thissitu ation  gave natio nal po liticia ns str on g
incentivestoshieldruralareasfromneoliberalpolicyreform sinordertose-
cure  the le gisla t ive m ajoriti es n ee de d t o im pl ement the se r eforms in
underrepresented,urbanareas.34 Thusthepersistent illiberalpolicyregimesin
theArgentineperipherywerenotstagnantbackwatersleftbehindbythepro-
cessofeconomic reforminurban areas.Rather,theywe recausally integra l
componentsoftheoverarchingeconomicreforms.
Astheseexamplessuggest,understandingeconomictransformationincon-
temporaryLatinAmericarequiresthatweexplorethelinkagesbetweenliberal
andilliberalareasinthecontextofasingle,unifiedsystem. Thesubnational
comparativemethodisakeytoolforachievingthisgoal.
Conclusion
Thesubnational comparativemethodoffersimportant advantagesinthree
coreareasoftheresearchprocess:researchdesign,measurement,andtheory -
building. Byincreasing th e number ofobservations and m aking  it ea sier to
co ns tru ct controlled c omparison s,  a f oc us on s ubnation al unitshelps
comparativistsmanagetheproblemof“manyvariables,smallN.”Subnational
comparativea nal ysis alsostrengthensthecapacity ofcomparativiststocode
casesaccuratelyandthus makevalidc ausa l inf erenc es.Finally,subnational
comparisonsprovideafirmfoundationforbuildingtheoriesthatexplainspa-
tiallyuneve n processesofpolitical andeconomictransformation.Inaddition
tothesemethodologicalbenefits, subnational comparativeanalysisoffersan
indispensabletoolforunderstandingthedecentralizingpoliticalandeconomic
trendsofthecontemporaryera.
Althoughthisarticlehashighlightedtheadvantagesofanalyzingsubnational
units,ithasalsoemphasizedthat,likeallmethodologies,thesubnationalcom-
parativemethodhaslimitations.Onelimitationconcernsgeneralizability.This
problememergesmostclear lyinthecontextofwhatIcallthe“within-nation”
strategyofsubnationalcomparativeanalysisthatfocusesoncasesfrom asingle
country.Althoughwithin-nationcomparisonscanprovideanespecially pow-
erfulwaytoe stabli sh control overhistorical, ecological ,andculturalcondi-
tions,thisstrategyofteninvolvesatrade-offbetweentheabilitytogaincontrol
andtheabilitytogeneralize.35 Onetechniqueformitigatingthistrade-offisto
combine within-nation co mparisonsand between-nationcomparisonsof
subnationalunitsfromdifferentcountries.Alternatively,subnationalcompari-

104 Studies in Comparative International Development / Spring 2001
sonscanbeusedinconjunctionwithcross-nationalcomparisons.Howdistinc t
strategies ofcom parison canbe combined to strengthen  both th e ab ility to
establishcontroloverconditions andtheabilitytogenerali zeisanimportan t
questionforfutureresearch.
Thepossibilitie sforgeneralizi ng from subnationalcasescanalso be en-
hancedbyimproving ourtoolsformakingvalidinferencesfromsubnationa l
tonational levelsofanalysis.Recentworkonquantitativemethodsprovides
newwaystomanagetheproblemsthatariseinmakingcross-level inferences
(e.g.,Grofman1995;King1997;KatzandKing1999).Adaptingsuchinsights
toqualitative,small-Nresearchshouldprovidebettertechniquesforgeneral-
izingfromsubnationa lunits.
A se cond  limitation of thesubnational comparative m ethod concerns the
problemofinterdependence.Becauseofthepossibilityofdiffusionand bor-
rowing amongsubnational  units ina single country,within-nation compari-
sonscaninvolveatrade-offbetweentheabilitytoestablishcontroloverpotential
explanatoryfactorsandtheabilitytomakeindependentobservationsandtests
of hypothe s es. Combiningwithin-nation comparison s andcomparisons of
subn ationa l unitsfr omdifferent countriescanreducetheeffectsofdiffusion
andcouldthushelpmitigatethistrade-off.
Itbearsemphasisthattheproblemofachievingindependenceamongcases
isbynomeanslimitedtosubnationalcomparativeanalysis:cross-nationalre-
searchondiffusion,dependency,andinternationaldemonstra tioneffectshas
longrecognizedtheimportanceofinterconnectionsamongnationalunits.The
processofglobalizationandtherecentgrowthofinformationtechnologyhave
undoubtedlymagnifiedthescopeanddensityofinterconnectionsamongboth
national andsubnational units. Thusfreestandingpolitical units that canbe
legitimatelytreatedasisolated,experimentalobservationsareprobablyharder
tofindatanylevelofanalysis. Studiesthataddressthemethodolo gicalchal-
lengesposed bythetrendtowardmoreinterconnectednesswillprovideafar
strongergroundingforfuturecomparativeresearch.
Notes
*IappreciatehelpfulcommentsonthismaterialfromNancyBermeo,DexterBoniface,David
Collier,John Gerring,Edwa rd Gibson,Robert Kaufman,JuanLinz,JamesMahoney,Kell y
McM ann,G erardo Munc k, Pete rNa rdulli, David Sa muels, Judit hTendle r,a nd two anony-
mousreviewers.Ialsobenefitedgreatlyfromtheinsightfulcommentsonanearlierdraftpro-
videdbytheparticipantsintheconferenceon“RegimesandPoliticalChangeinLatinAmerica,”
heldattheUniversityofIllinoisatUrbana-ChampaigninAugust1999.
1. See, for example ,Kohli (1987, 1990) ; Fishman (1993) ;Fox ( 1993, 1996); Putnam (1993) ;
Locke(1995);Hagopian(1996);LockeandJacoby(1997);Rubin(1997);Stoner-Weiss(1997);
Tendler (1997); Sam uels (1998); Snyder (1999a , 2001); Firm in-Sell ers(2000); G ibson and
Calvo(2000);Heller(2000);Montero(2000);RemmerandWibbels(2000);Varshney(2001);
andJonesLuong(2002).Althoughsubnationalunitscommandincreasingattention,subnational
politicalanalysisisnot an ent irely new phenome non in c ompara tivepoli tics: a number of
earlierworksfocusedonsubnationalcases(forexample,Tilly1964;LinzanddeMiguel1966;
KesselmanandRosenthal1974;andTarrow1967,1976) .
2. TheseissuesdoreceivelimitedattentioninLijphart(1971);King,Keohane,andVerba(1994);

Snyder 105
andPeters(1998).LinzanddeMiguel(1966)isoneofthefewworksthatfocusesprimarilyon
methodologicalissuesinvolvedincomparingsubnationalunits.
3. TheclassicdiscussionofthisproblemisLijphart(1971).SeealsoCollier(1993).
4. Theproblemofwhole-nationbiasisalsodiscus sedinLijphart(1975:166-9) .
5. See,forexample,O’Donnell(1993);Fox(1994);Kurtz(1999);Snyder(1999a,2001);Gibson
andCalvo(2000);andRemmerandWibbels(2000).
6. Collier(1993:112)alsodiscussesthevalueofwithin-case,“internalcomparisons”forincreas-
ingthenumberofobservations.
7. Inthisconte xt,Kingetal.discussAtulKohli’s(1987)researchonsocialpolicyinIndia.To
testhypothesesabouttheeffectsofstate-levelpoliticalregimesonpovertypolicy,Kohliuses
additionalobservationsfromthelocal,panchayat level(Kohlialsolooksatothercountriesto
evaluatetheobservableimplicationsofhishypotheses).Recentworkondivided government
atthestatelevelintheUnitedStatesprovides anothergood illustrationofhowmoving toa
lowerlevelofaggregation canhelpsolvethesmall-Nproblembyinc reasingthenumberof
observations.See ,forexample,AltandLowry(1998);andLowry,Alt,andFerree(1998).
8. Lieberson(1985:108)adoptsaquitedifferentviewofthisissue,concludingthatevidenceata
lower levelof aggregation “isirrelevantfor determiningthevalidityofaproposition about
processesoperatingonthehigherlevel”andtha t“nousefulunderstandingofthehigher-level
str ucturecanbeobtainedfromlower-levelanalyses.”
9. “Galton’sproblem”issonamedbecauseoftheobjectionraisedbyFrancisGaltonatthemeet-
ingoftheRoyalAnthropologicalInstitutein1889toapioneeringpaperbyEdwa rdB. Tylor
introducingthecross-culturalsurveymethod.Galtonpointedoutthatbecausetraitsoftenspread
bydiffusion(forexample ,byborrowingormigration),observationsofsuchtraitsacrosscul-
tureswerenotnecessarilyindependentinstances(Naroll1961:15).
10.Asdiscussedbelow,dynamicinteractionsamongthesubunitsofapoliticalsystemshouldnot
necessarilybeseenasatroubling hindrancetomakingvalidcausalinferences.Rather,such
interactionscanbeafertilesourceoftheoreticalinnovationandprogress.
11. SeeLocke(1995)foranintriguingresearchdesignthatcombinesthestudyofvariedsubnational
unitswithananalysisofmultiplesectors.LockeoffersasubnationalstudyoftheItalianauto-
mobileandtextileindustries.
12. Rustow(1968:45)madearelatedpointwhenhearguedthat“meregeographicproximitydoes
notnecessarilyfurnishthebestbasisofcomparison.”
13. AsO’Donnellhimselfobserves(1973:21),suchastrategyoverlooksthepossibilitythatthe
presenceora bsenceofalarge,underdevelopedperipherymay haveacrucialimpactonna-
tionalpolitics.Hence,eventhoughthemodernsectorsinArgentinaandBrazilaresimilarin
size,theireffectonnationalpoliticsmaydifferbecauseofthediffer entsizesoftheagrarian
sectorsinthetwocases.
14. Thisbetween-nationcomparisonthusenablesLipsettoemploya“mostsimilar”systemsde-
sign.On“mostsimilar”and“mostdifferent”systemsdesignsseePrzeworskiandTeune(1970).
15. Anothercommonly-employedstrategyofbetween-nationsubnationalcomparisoninvolvesfo-
cusingonthesameeconomicsectoracrossdistinctcountries(see,forexample,Evans1995;
Bates1997; Ka rl1997; a ndPa ige 1997).  Because of cross -country differ enceson a wide
rangeofothervariables,suchanapproachma ybelesseffectivethanwithin-countrycompari-
sonsforexploringhowvariationinpoliticalfactorsshapeseconomicoutcomes.
16. InthecontextoftheSaskatchewan-Alberta-Manitobajuxtaposition,Lipset(1950:217)makes
anintriguingcross-temporalcomparison,notingthat“during theperiodbetweenthe1890’s
and the earlytwentiethcen tury,when economic andecologicalconditions in Manitoba ap-
proximatedthoseofSaskatchewanandAlbertatoday[1950],theManitobaagrarianpolitical
movementwasstrong.”
17. Thisdifficultyofachievingindependentobservationsamongspatiallyproximateunitsisdis-
cussedintheeconometricliteratureon“spatialautocorrelation”(e.g.,Anselin1988).Thespa-
tialautocorrelation problem isi nm any waysanalogoustothe“temporalautocorrelation”
problemintime-seriesanalysis(Eagles1995:7;King1997:166).
18.Thevalueofbetween-nationcomparisonsofnon-contiguousunitsforincreasingtheindepen-
denceamongcasesissuggestedbyWaldoTobler’s“firstlawofgeography”whichpositsthat

106 Studies in Comparative International Development / Spring 2001
“everythingisrelatedtoeverythingelse,butnearthingsaremorerelatedthandistantones”
(Eagles1995:7).
19. See,forexample,Lipset(1959)andthehugeliteratureinspiredbythisstudy.
20. Peters(1998:44)makesarelatedpointwhenheobservesthatwholecountriesareoftencoded
as outlie rs when only one par t of the countr y has extr eme values , where ast he res tof  the
countryactuallyhas“normal”valueswithrespecttothecross-nationaldistribution.
21. See,forexample,Nelson(1990)andCórdoba(1994) .
22. Asimilarrelianceoneasilyavailable,national-levelOECDdatainmuchrecentworkonthe
comparativepoliticaleconomyofadvancedindustrialcountrieshasalsoundoubtedlyobscured
importantprocessesoccurringbelowthenationallevel.
23. Moving tot hesubna tional leveld oesnot nec essar ilyguara nteeacorrectcodingofna tional
casesbecausescholarscanmiscodethesubnationalunitsthemselves.Forexample,Gaines
(1999) shows that schol ars misc oded province -level party s ystems in Cana da as two-pa rty
systemsandthusincorr ectlyconcludedthatCanadaconformedtoDuverger’slawatthepro-
vincial,ifnotatthenational,level(Duverger’slawstatesthatpluralityruleelectionsresultin
two-partycompetition).Bycontrast,Gainesshowsthatprovince-levelpartysystemsinCanada
actua llyha ve more t han two part ies. Thus Cana da stands as agenuinecounterexampleto
Duverger’slaw.
24. Ontheratherdifferentproblemof“themisuseofthelittleknowncase,”seeKenworthy(1973).
RuthBerinsCollierandDavidCollier’sinversionofKenworthy’sformulation—“themisuse
ofthebestknowncase”—hasobviouslyinfluencedmyformulation(CollierandCollier1991:
14).
25. Theproblemofmisusingthebestknowplacemightbecharacterizedasthewell-knowneco-
logicalfallacy“inreverse”inthatdisaggregateddatafromsubnunitsresultsinabiasedinfer-
enceabouttheaggregatelevel.Theecologicalfallacyreferstothebiasesthatcanresultwhen
usingaggregatedatatomakeinferencesaboutindividual-levelbehavior(Robinson 1950).A
methodformakingunbiasedinferencesfromaggregatedataispresentedinKing(1997).
26. Initsviewofnationalpoliticaleconomiesascomplexcompositesofdiversesubnationalsys-
tems,Herrigel’sworkexemplifieswhatmightbecalledthe“MITschool”ofpoliticaleconomy
associatedwiththeworkofscholarssuchasSuzanneBerger,MichaelPiore,Cha rlesSabel,
andJudithTendler.RichardLocke’sresearchonItalyalsoexemplifiestheMITschool.In
his  impo rta nt boo k on theItaliancase,Locke ( 1995)  arg ues for a“micro-politics” a p-
proacht opoliticalec onomy rootedin theassumptionthat“nati ona l politicaleconomies
arenotcoherent systemsbutratherincohe re ntcompositesofdiversesubna ti ona lpatterns
thatcoexist(oftenuneasily)withinthesamenationalterritory.”Locke(1995:16-20)criticiz es
tendenciesbysomescholarstoextrapolatetheeconomicsuccessesanddynamismoftheso-
called“ThirdItaly” basedonsmallenterprisesandflexibleproductiontothenationalcaseas
awhole.Thistendencyprovidesagood exampleofwhatImeanbymisusingawell-known
place.
27. Intheircritiqueof“dualist”perspectivesondevelopmentthatdividedcountriesintoamodern,
capitalisteconomyandabackward,pre-capitalistone,dependencytheoristsoftenexpresseda
strongconcernwithdescribingandexplainingtheconnectionsbetweencontrastingregionsof
acountry.See,forexample,Stavenhagen(1968)andPortes(1983).
28. Theterm“center-centered”isfromRubin(1997:12).
29. SeeTendler(1999:105-19)foranintriguingextensionofGibson’sargumentthatshowshow
“two-pronged ”coal ition-bui ldingstra tegiesthatmar rymodern andtradit ionalconst ituencies
havebeenimplementedatthesubnationallevelbycontemporaryBraziliangovernors.Roberts
(1995)makesarelatedargumentabout “neopopulism ”asacoalition-buildingtoolinthecon-
text of the impleme ntation of free- markete conomic ref ormsi nP eru during the 1990s. On
neopopulism,seealsoWeyland(1996,1999).
30. Hagopian ’simport ant work on Bra zil also shows that a keen sensitivityto varia tionatthe
subnationallevelcanhighlight howtransf ormationalnationalregimesaccommodate,rather
thanmarginalize,traditionalelites.Asintheliteratureonpopulism, workbypoliticalscien-
tistsonBrazil’sbureaucratic-authoritarianregimetendedtofocusexclusivelyonurban,met-
ropolitanareas.Inthisrespect,Hagopian(1996:xiii)notes,“Manycountrystudiesthatidentify
themselvesasnationalinscope arein factanalysesofthe mostadvancedregions—Riode

Snyder 107
JaneiroandespeciallySãoPaulo.Assuch,theyhaveacollectiveblindspotand…havepaida
pricefortheirmyopia.”SeealsoMigdal,Kohli, andShue(1994)onthevalueofsubnational
analysisinthestudyofstate-societyrelations.
31. By“illiberal”Imeannon-democraticinthepoliticalsphereandnon-free-marketintheeco-
nomicsphere.
32. OnsubnationalauthoritarianregimesinMexico,seeFox(1994)andSnyder(1999b).Onthe
important,yetoftenoverlooked,roleofnon-democraticruralelitesinsecuringpoliticaland
economicliberalizationatthenationallevelinChile,seeKurtz(1999).
33.We mightwanttoconsiderhowweshouldcodesuchacase.Isacountrycharacterizedbythe
coexistenceofa nati onal democr atic regim ea longside s ubnationa l author itarian regim es a
full-fledgeddemocra cy?
34. RemmerandWibbels(2000)alsoanalyzethedivergenteffectsofneoliberaleconomicref orms
acrossArgentineprovinces.Theyarguethatthepersistenceofilliberalpolicyregimesatthe
provinciallevelmay jeopardizenationaleconomicperformanceand thusunderminethe
sustainabilityofreformsatthenationallevel.
35. Ontherelatedtr ade-offbetweengeneralizabilityandconceptualvalidity,seeSartori(1970;
1984); Collierand Mahon (1993); Collierand Levitsky(1997); Collier andAdcoc k(1999);
andMunck(1998).
References
Alt, Jame sE. and Robert C. Lowry.1994. “Divide dGove rnment and Budget Defic its: Evidence
fromtheStates.”AmericanPoliticalScienceReview88(December):811–28.
Anderson,JeffreyJ.1992.TheTerritorialImperative:Pluralism, Corporatism,andEconomicCri-
sis.NewYork:CambridgeUniversityPress.
Anselin,Luc.1988. SpatialEconometrics:MethodsandModels.Boston:KluwerAcademicPub-
lishers.
Bates,RobertH.1997.Open-EconomyPolitics:ThePoliticalEconomyoftheWorldCoffeeTrade.
Princeton:PrincetonUniversityPress.
Collier,DavidandRobertAdcock.1999.“DemocracyandDichotomies:APragmati cApproachto
ChoicesaboutConcepts.”AnnualR eviewofPoliticalScience2:537–65.
Collier,DavidandStevenLevitsky.1997.“DemocracywithAdjectives: ConceptualInnovation in
ComparativeResearch.”WorldPolitics49(April):430–451.
Collier,DavidandJamesE. Mahon,Jr.1993.“Conceptual ‘Stretching’Revisited:AdaptingCat-
egoriesinComparativeAnalysis.”Amer icanPoliticalScienceReview87(December):845–
55.
Collier,RuthBerinsandDavidCollier.1991. Shaping thePoliticalArena:CriticalJunctures,the
LaborMovement,and RegimeDynamicsinLatinAmerica.Princeton:PrincetonUniversity
Press.
Córdoba, José. 1994. “M exico.” In The Political Economy o f Policy Refor m,ed.J.Williamson.
Washington,D.C.:InstituteforInternationalEconomics.
DiTella,TorcuatoS.1965.“PopulismandReforminLatinAmerica.”InObst aclestoChange in
LatinAmerica,ed.C.Véliz.NewYork:OxfordUniversityPress.
_____________. 1968. “TheWorkingClassinPolitics.”InLatinAmericaandtheCaribbean–A
Handbook ,ed.C.Véliz.NewYork:Praeger.
Eagles,Munroe.1995.“SpatialandContextualModelsinPoliticalResearch:AnIntroduction.”In
Spatial  and Contex tual Model s in Political Re search,ed. M. E agles. London:  Taylor and
Francis.
Evans,Peter.1995.EmbeddedAutonomy:StatesandIndustrialTransformation.Princeton:Princeton
UniversityPress.
Firmin-Sellers,Kathryn.2000.“Institutions,Context,andOutcomes:ExplainingFrenchandBrit-
ishRuleinWestAfrica.”ComparativePolitics32(April):253–72.
Fishman,RobertM.1993.“ DivergentPaths:LaborPoliticsinBarcelonaandMadrid.”InPolitics,
Society,andDemocracy:TheCaseofSpai n,ed.R.Gunther.Boulder,CO:WestviewPress.
Fox,Jonathan.1993.ThePoliticsofFoodinMexico: StatePowerandSocialMobilization.Ithaca:
CornellUniversityPress.

108 Studies in Comparative International Development / Spring 2001
_____________.1994.“LatinAmerica’sEmergingLocalPolitics.”JournalofDemocracy5(April):
105–16.
_____________. 1996. “How doe s Civil Socie ty Thicken? The  Politica lCons truction of Soc ial
CapitalinRuralMexico.”WorldDevelopment24:1089 –1103.
Gaines,BrianJ.1999.“Duverger’sLawandtheMeaningofCanadianExceptionalism.”Compara-
tivePoliticalStudies32(Oc tober):835-61.
Gers chenkron, Alexander. 1962. Economic Backwardness in HistoricalPerspective.Cambridge,
MA:HarvardUniversityPress.
Gibson,EdwardL.1997.“The PopulistRoadtoMarketReform:PolicyandElectoralCoalitionsin
MexicoandArgentina.”WorldPolitics49(April):339–70.
Gibson,EdwardL.andErnestoCalvo.2000.“FederalismandLow-Maintenance Constituencies:
TerritorialDimensionsofEconomicRef orminArgentina.”StudiesinComparativeInterna-
tionalDevelopment35,3(Fall).
Grofman,Bernard.1995.“NewMethodsforValidEcologicalInference.”InSpatialandContextual
ModelsinPoliticalResearch,ed.M.Eagles.London: TaylorandFrancis.
Hagopian,Frances.1996.TraditionalPoliticsandRegimeChangeinBrazil.NewYork:Cambridge
UniversityPress.
Heller,Patrick.2000.“DegreesofDemocracy:SomeComparativeLessonsfromIndia.”WorldPolitics
52(July):484–519.
Herrigel,Gary.1996.IndustrialConstructions: TheSourcesofGermanIndustrialPower.NewYork:
Cambr idgeUniversityPress.
JonesLuong, Pauline .2002.InstitutionalChange and PoliticalContinuityinPost-SovietCentral
Asia.NewYork:CambridgeUniversityPress.
Karl,TerryLynn.1997.TheParadoxofPlenty: OilBoomsandPetro-States.Berkel ey:University
ofCaliforniaPress.
Katz,JonathanN.andGaryKing.1999.“AStatisticalModelforMultipartyElectoralData.”American
PoliticalScienceReview93(Mar ch): 15–32.
Kenworthy,Eldon. 1973. “The FunctionoftheLittleKnownCaseinTheoryFormation, orWhat
PeronismWasn’t.”Compar ativePolitics6(October):17–46.
Kesselman,MarkandDonald Rosenthal.1974. Loc al Powerand Com parative Politi cs.Beverly
Hills:SagePublications.
King,Gary.1997.ASolutiontotheEcologicalInferenceProblem: ReconstructingIndividualBe-
haviorfromAggregateData.Princeton,NJ:PrincetonUniversityPress.
King,Gary,RobertO.Keohane,andSidneyVerba.1994.DesigningSocialInquiry:ScientificIn-
ferenceinQualitativeResearch.Princeton,NJ:PrincetonUniversityPress.
Kohli,Atul.1987.TheStateandPovert yinIndia:ThePoliticsofReform.NewYork:Cambridge
UniversityPress.
_____________. 1990.Democrac yandDiscontent:India’sGrowingCrisisofGovernability.New
York:CambridgeUniversityPress.
Kurtz,MarcusJ.1999.“FreeMarketsandDemocraticConsolida tioninChile:TheNationalPoli-
ticsofRuralTransformation.”PoliticsandSociety27:2(June):275–301.
Lijpha rt,Are nd. 1971. “Compar ative Politic sand t heC omparati veMe thod.”Amer icanPolitical
ScienceReview65: 682–93.
_____________. 1975.“ TheCompara ble-Case sStrategyinCom parativeRese arch.”Comparative
PoliticalStudies8(July): 158-77.
Linz,JuanJ.1986.ConflictoenEuskadi.Madrid:EspasaCalpe.
Linz, Jua nJ .a ndAmando de Mi guel. 1966. “Within- Nation Differ ences andC omparis ons: The
Eight Spains.”InCompar ing Nations:TheUseof Quantita tiveData inCross-NationalRe-
search,ed.R.L.MerrittandS.Rokkan.NewHaven:YaleUniversityPress.
Lipset,SeymourMartin.1950.AgrarianSocialism:TheCooperativeCommonwealthFederationin
Saskatchewan .Berkeley:UniversityofCaliforniaPress.
Lipset,Seymour Martin.1959.“SomeSocialRequisitesof Democracy:Economic Development
andPoliticalLegitimacy.”AmericanPoliticalScienceReview53(March):69–105.
Locke,RichardM.1995.RemakingtheItali anEconom y.Ithaca:CornellUniversityPress.
Locke,RichardM.andWadeJacoby.1997.“TheDilemmasofDiffusion:SocialEmbeddednessandthe
ProblemsofInstitutionalChangeinEasternGermany.”PoliticsandSociety25(March):34–65.

Snyder 109
Lowry,RobertC.,JamesE.Alt,andKarenE.Ferree.1998.“FiscalPolicyOutcomesandElectoral
AccountabilityinAmericanStates.”AmericanPoliticalScienceReview92(December):759-74.
Migdal,JoelS.,AtulKohli,andVivienneShue,eds.1994.StatePowerandSocialForces:Domina-
tionandTransformationintheThirdWorld.NewYork:CambridgeUniversityPress.
Miles,William.1994. Hausaland Divided.Ithaca:CornellUniversityPress.
Montero,AlfredP.2000.“DelegativeDilemmasandHorizontalLogics:SubnationalIndustrialPolicy
inBrazilandSpain.”PaperpresentedattheThirdMeetingofInternationalWorkingGroupon
SubnationalEconomicGovernanceinLatinAmericaandSouthernEurope.SanJuanPuerto
Rico,August26-28.
Munck,Gerardo.1998.“CanonsofResearchDesigninQualitativeAnalysis.”StudiesinCompara-
tiveInternationalDevelopment33(Fall):18-45.
Naroll,Raoul.1961.“TwoSolutionstoGalton’sProblem.”PhilosophyofScience28(January):15-39.
_____________.1966.“ScientificComparativePoliticsandInternationalRelations.”InApproaches
toComparativeandInternationalPolitics,ed.R.Farrell.Evanston,IL:Northwester nUniver-
sityPress.
Nelson, J oan M. 1990. “Intr oduction: The Poli tics of Economic Adjustm ent in Developing Na-
tions.”InEconomicCrisisandPolicyChoice:ThePoliticsofAdjustmentintheThirdWorld,
ed.J.Nelson.Princeton:PrincetonUniversityPres s.
O’Donnell, Guillermo.1973.Moder nizationandBureaucratic-Authoritarianism:StudiesinSouth
AmericanPolitics.Berkeley:InstituteofInternationalStudies,UniversityofCalif ornia,Ber-
keley.
_____________. 1999.Counterpoints:SelectedEssaysonAuthoritarianismandDemoc ratization.
NotreDame,IN:UniversityofNotr eDamePress.
Paige,JeffreyM.1997.CoffeeandPower:RevolutionandtheRiseofDemocracyinCentralAmerica.
Cambridge:HarvardUniversityPress.
Peters,B.Guy.1998.ComparativePolitics:TheoryandMethods.NewYork:NewYorkUniversity
Press.
Portes,Alejandro.1983. “TheInformalSector:Definition,Controversy,andRelationtoNa tional
Development.”Review7(Summer):151-74 .
Przeworski,AdamandHenryTeune.1970. The Logicof ComparativeSocialInquiry.NewYork :
JohnWileyandSons.
Putnam,RobertD.,withRobertLeonardiandRaffaellaY.Nanetti.1993.MakingDemocracyWork:
CivicTraditionsinModernItaly.Princeton:PrincetonUniversityPress.
Remm er,Kare nL. andE rikWibbel s.2000. “ TheSubna tional Polit icsof Econom icAdjust ment:
ProvincialPoliticsandFiscalPerformanceinArgentina.”ComparativePoliticalStudies 33
(May):419-51 .
Roberts,Kenneth.1995.“NeoliberalismandtheTransformationofPopulisminLatinAmerica:The
PeruvianCase.”WorldPolitics48(October):82-116.
Robinson,WilliamS.1950. “EcologicalCorrela tionsandtheBehaviorofIndividuals.”American
SociologicalReview15(June):351-57.
Rokkan,  Stein. 1970. C itiz ens, Elec tions, Parties :Approaches to the C omparative  Study  of the
ProcessesofDevelopment.NewYork:DavidMcKayCompany.
Rubin,JeffreyW.1997.DecenteringtheRegime:Ethnicity,Radicalism,andDemocracyinJuchitán,
Mexico.Durham:DukeUniversityPress.
Rustow,DankwartA.1968.“Moder niza tionandComparativePolitics:ProspectsinResearchand
Theory.”ComparativePolitics1(October):37-51.
Sam uels, David J . 1998. “ Caree rism and its Conse quences : Federa lism, Ele ctions, and Poli cy-
MakinginBrazil,”DoctoralDissertation,DepartmentofPoliticalScience,UniversityofCali-
fornia,SanDiego.
Samuels,DavidJ.andRichardSnyder.2001.“TheValueofaVote:MalapportionmentinCompara-
tivePerspective.”BritishJournalofPoliticalScience31,4(October).
Sar tori, Giovanni. 1 970. “Conc ept Mis forma tion in Compara tive Resea rch.”American Political
ScienceReview64:1033-53.
Sartor i,Giovanni,ed.1984.SocialScienceConcepts:ASystematicAnalysis.BeverlyHills:Sage.
Snyder,Richard.1999a.“AfterNeoliberalism:ThePoliticsofReregulationinMexico.”WorldPoli-
tics51(January):173-204 .

110 Studies in Comparative International Development / Spring 2001
_____________. 1999b.“Afterthe State Withdr aws:Neolib eralismandSubnat ionalAuthorita rian
RegimesinMexico.”InSubnationalPoliticsandDemocratizationinMexico,ed.W.Cornelius,
T.Eisenstadt,andJ.Hindley.LaJolla,CA:CenterforU. S.-MexicanStudies,Universityof
California,SanDiego.
_____________. 2001.  Politic saft er Neoliberali sm: Reregulation in Me xico.NewYork: Cam-
bridgeUniversityPress.
Snyder, Rich ard and Davi d Samuel s. 2001. “Devaluing the Vote in Lati nAmer ica.” Journal of
Democracy12(J anuary):146–59.
Stavenhagen,Rodolfo.1968.“SevenFallaciesAboutLatinAmerica.”InLatinAmerica:Reformor
Revolution?eds.J.PetrasandM.Zeitlin.Greenwich,CT:FawcettPublications.
Stoner-Weiss,Kathr yn. 1997. LocalHeroes:The PoliticalEconomyof RussianRegional Gove r-
nance.Princeton:PrincetonUniversityPress.
Tarrow,SidneyG.1967.PeasantCommunisminSouthernItaly.NewHaven:YaleUniversityPress.
Tarrow,Sidney.1976.FromCentertoPeriphery:AlternativeModelsofNational-Local PolicyIm-
pactandanApplicationtoFranceandItaly.Ithaca,NY:WesternSocietiesProgram,Cornell
University.
Tendler,Judith.1997.GoodGovernmentintheTropics.Baltimor e:JohnsHopkinsUniversityPress.
Tendler,Judith(withRodrigoSerrano).1999. “TheRiseofSocialFunds:WhatareTheyaModel
Of?”PreparedforTheMIT/UNDPDecentraliza tionProject,ManagementDevelopmentand
GovernanceDivision,UnitedNationsDevelopmentProgramme,NewYork.
Tilly,Charles.1964.TheVendée.Cambr idge,MA:HarvardUniversityPress.
Varshney,Ashutosh.  2001. Ethnic C onflict and CivicL ife: Hindus  and Musl ims in Indi a. New
Haven:YaleUniversityPress.
Weyland,Kurt.1996.“NeopopulismandNeoliberalisminLatinAmerica:UnexpectedAffinities.”
StudiesinComparativeInternationalDevelopment31(Fall): 3–31.
Weyland, Kurt.1999.  “Neoliber alPopulis mi nLatinAm erica andEas ternEur ope.”Comparative
Politics31(July): 379–401.