ImperialFootprints:

ColonialLegacies,PartyMachinesandContemporaryVoting

KeithA.DardenYaleUniversity

DepartmentofPoliticalScienceKeith.darden@yale.edu

PaperPreparedfortheMidwestPoliticalScienceAssociationMeetingsApril24,2010

**DRAFT**

PLEASEDONOTCITEWITHOUTPERMISSIONOFTHEAUTHOR

Abstract:Thepapercombinesfieldinterviews,historicalprocess‐tracing,andaspatialdiscontinuityanalysis of the 2010 Ukrainian Presidential elections to explain why contemporary voters cleavealongonestretchofa formerimperialboundaryinUkraine. FormerlyOttomanacquisitionsoftheRussianEmpirefaceddirectrule,education,urbanization,andRussificationinthelate18thand19thCenturies,creatingadistinctivepro‐RussianregionalculturethatsuppliesthebaseofsupportfortheParty of Regions. Areas annexed from Poland in the 18th Century were subject to indirect rulethrough the Polish nobility, which led to less peasant education andmore limited cultural ties toRussia. Analyzinga60‐kmwidebandofpollingdistrictsalongthe former frontierbetweenPolishandOttomanterritoriesinUkraine,wefindsubstantiallygreatersupportforYanukovichinpollingdistricts on the Ottoman side of the former frontier. The results suggest that pre‐Communistimperiallegacieshaveasignificanteffectoncontemporaryvotingpatterns.

Voting patterns in many central European countries have taken on a

decidedlyregionalcharacter in thedecades following thecollapseofCommunism.

ThemostrecentUkrainianPresidentialElectionsrepeatedaregionalvotingpattern

inUkrainian politics that is nowquite consistent and familiar. The southern and

easternprovincesofthecountry1consistentlyvoteforthesamecandidateorparty.

Andtheydosobylargeandnearlyuniversalmajoritiesacrosseverypollingdistrict

within the region. The formerlyAustrian territoriesof Ivano‐Frankivsk,Ternopil,

and Lviv consistently vote for the rival candidate in extraordinarily high

percentages, and they consistently carrywith them the remainder of the country,

with the exceptionof Zakarpatska, Chernivtsy, andKirovohrad (where the vote is

often mixed). Again in 2010, regional voting dominated all other patterns in

Ukrainianelections.

Ukraine is not unique in this regard. Similar divides exist in Poland and

Romaniagoingbacktosomeoftheearliestpost‐Communistelections.Moreover,as

theelectoralsystemscontinuetomature,theregionalvotingpatternsappeartobe

solidifyingratherthandiminishinginsignificanceandpredictivepower.Itisnotan

exaggeration tosaywhen itcomes topredictingelectoraloutcomes in these three

countries, themost important piece of information to know about a voter is not

wheretheywork,howmuchtheyearn,orhowoldtheyare,butinwhichregionof

thecountrytheyreside.

1TheoblastsofDonetsk,Luhansk,Kharkiv,Dnipropetrovsk,Kherson,Zaporizhia,Crimea,Mykolaev,andOdessa.

More important, these electoral regions do not reflect an arbitrary

geography.Inallthreecases,theboundariesoftheregionalvotingblocsfallalong

theold anddefunct boundaries of the empires out ofwhich these countrieswere

constructed. The correlation between imperial boundaries and contemporary

voting is so close, in fact, that the best predictor of which party would win an

electoraldistrictinthe2007legislativeelectionsinPolandwaswhetherthatdistrict

fellwithinthePrussianpartitionpriorto1918.Thesamehasbeentrueforthemost

recent Romanian elections, which were dominated by a distinction between

formerlyHabsburgandformerlyOttoman(Moldavia,Walachia,Dobrogea)sections

ofthecountry. AlthoughUkraine’s imperial legaciesaremorevariedandcomplex

thananyothercountryinEurasia,allofitsPresidentialandParliamentaryelections

since2002exhibitaclearelectoraldividethatappearstofollowlooselytheareasof

historicPolish settlementand control. These remarkable imperial “footprints”on

the contemporary electoralmap of Europe beg a fundamental question: Why do

18th and 19th Century Imperial boundaries provide such a powerful predictor of

votingbehaviorinthe21stCenturyinthelargestcountriesinCentralEurope?

Toprovideapartialanswertothisquestion,thispaperfocusesontheresults

of the2010Presidentialelections inUkraine,a countrywhere the regionalvoting

patterns have gained a great deal of attention and one which offers a richer

patchworkofimperiallegaciesthananyothercountryinEurope.Tobetterisolate

the effects of a possible imperial legacy, I focus on election results within an

approximately 60‐km wide band of territory straddling the old border that

distinguished the Ottoman Empire from the Polish Commonwealth when the

RussianEmpireannexedtheterritoriesattheendofthe18thCentury.Myhopeis

that a detailed analysis of imperial legacies and contemporary voting patterns in

thisregionwillilluminatetheprocessesandmechanismsbehindthedeepimperial

footprintoncontemporaryelectionsthatwefindinUkraineandelsewhere.

My initial findings suggest that two factors play an important role in

determining the imperial voting pattern: regional political culture and the

orientation and strength of the provincial party machine. To be more precise, I

argue that the different strategies that the Russian Empire employed in the

incorporationofformerlyPolishandformerlyOttomanterritoriesledtothesetwo

areas to differ significantly in their regional political culture, orientation, and

dominantidentification—inparticulartheirattitudestowardRussianculture.Inthe

New Russia (Novorossija) province, comprising areas acquired from the Ottoman

Empire,thestrategyoftheTsaristEmpirewastocolonizethenewterritoryrapidly

andtoimposedirectrule.Intheseareas,anewandlargelymartialRussiannobility

wascreated,theschoolswereinRussianandwerewidespreadbythebeginningof

the20thCentury,andserfdomwasuncommon.Asaresult,themobilizedandfreer

peasantryhada greater attachment to and identificationwithRussia andRussian

culture. In contrast, the Russian Empire pursued a strategy of indirect rulewith

limited central influence in the areas annexed fromPoland at the end of the 18th

Century. Intheseareas,thePolishnobilityremainedinplace,preserveditspower

and privilege, and resisted the education of the peasantry and worked to block

cultural developments thatwould lead to greater associationwith Russia. These

areasbecamelessRussianinidentification,aprocessthatwasenhancedbySoviet

nationalitiespoliciesinthe1920sand1930s.

I contend that when there is genuine electoral competition and parties

choose to raise the salience of cultural issues, as was the case in the 2010

PresidentialelectionsinUkraine,thesedistinctiveregionalculturesthenshapethe

character of the party organizations that can succeed in the region. Issues that

pertain to Ukraine’s political and cultural relations with Russia (NATO, EU

membership,languagepolicy)havebeensalientincontemporaryUkrainianpolitics

for the past decade or more. Pro‐Russian candidates developed stronger party

organizationsintheprovincesofNewRussiaandmoreUkrainian‐andpro‐Western

partieshavedevelopedmachinesintheformerlyPolishimperialterritories.These

regional machines—through their superior organization and vote‐buying

resources—are then able to secure the vote of those voters with much weaker

cultural orientations or partisan preferences. The result is that the party

organizationsthenextendandhardenthemajoritiesforthedominantpartywithin

animperialregion,sharpeningtheregionalelectoraldivideovertimeandleadingto

a stronger “imperial footprint” in the electoral results. This interaction between

regionalpoliticalcultures,issuesalience,andpartyorganizationsappearstobethe

linkbetweenthe19thCenturyboundariesandcontemporaryvotingpatterns.

Thepaperproceedsinthreesections.SectionIdescribesthemethodbehind

thepaperandclarifiesthelogicofthespatialdiscontinuityresearchdesign.Section

IIdescribestherelevantdifferencesinRussianimperialincorporationstrategiesin

this area of the Russian empire and [partially!] traces out the processes and

mechanisms linking the 220 year‐old imperial boundary to the development of

different regional cultures evident in contemporary Ukraine. Section III then

presents the findings showing the effect of the imperial boundary on voting

behavior in the2010UkrainianPresidentialElectionsandshowshowoblast‐level

partymachinesenhanceandhardentheseregionalculturalcleavagestoproducea

substantiallystrongerregionaleffect.SectionIVconcludeswithadiscussionofour

understanding of potential causal pathways throughwhichprocesses in the fairly

distantpastcanbeconceivedascausesofcontemporaryevents(legacies).

I.SpatialDiscontinuitiesinImperialTreatment

This paper takes advantage of sharp, historically contingent, spatial

discontinuitiestodeterminetheeffectsofimperiallegaciesonvotingbehavior.The

logicofthedesignissimilartothatofnaturalexperimentsandcanbeconsidereda

subset of a broader category of discontinuity designs.2 Rather than examine all

UkrainianelectoraldistrictsIfocusonamorehomogeneousregionalsubsetofthose

polling districts that fall within a narrow band on opposite sides of a former

imperialboundary.Becauseallofthedistrictsareincloseproximitytooneanother

andsharethesamegeographicandnaturalenvironment,andbecauseofthegeneral

commonalities inpolitical institutions,socialstructure,andeconomicorganization

generatedby70yearsofSovietCommunism,thepresumptionisthatwecanmore

2 For an example of spatial regression discontinuity designs see David Card and Alan B. Krueger,“Minimum Wages and Employment: A Case Study of the Fast‐Food Industry in New Jersey andPennsylvania,”AmericanEconomicReview,vol.84(4),pages772‐93,September1994.Onregressiondiscontinuity designs more generally, see Imbens, G.W., Lemieux, T., “Regression discontinuitydesigns:Aguidetopractice,”JournalofEconometrics(2007).

confidentlyidentifytheeffectofanimperiallegacywiththisnarrowersubsetofthe

data.Ifthedistrictsoneithersideofthepreviousbordersharerelevantattributes

otherthanthosedirectlyattributabletothetreatment,thenanydifferencesseenin

the districts on either side of the border should therebymore closely reflect the

effectsoftheimperiallegacy.Becauseimperiallegaciesarespatiallydiscontinuous

variables—they have clear and known edges—a comparison of groups on either

side of a frontier approximates the assignment of similar groups randomly into

treatmentandcontrolandallowsustoidentifytheimperialeffect.

Theempiricalstrategyfirstcallsforidentifyinganareawherethepopulation

is quite homogeneous by contemporary measures, but which is divided by an

historicalimperialboundarythathaslongsinceceasedtodemarcateanypoliticalor

administrative differences. For this purpose I have chosen a stretch of territory

alongtheKodymariverinSouthwesternUkraine(seeFigure1).TodaytheKodyma

runsthroughthenorthernpartofOdessaprovince(oblast),butinthe18thCentury,

theKodymaservedasthedividinglinebetweentheOttomanEmpireandthePolish

Commonwealth,approximatelybetweenthebordertownofBalta(PolishJozefgrod,

briefly Russian Elensk) and Olviopol (Polish Bohopol, Ottoman Golta, today’s

Pervomaisk).3TheKodymaisessentiallyasmall,sandystreaminmid‐summerand

notamajorgeographicimpediment. Theslightrisetothesouthof itmightafford

some strategic advantage, but only marginally so. It does not differentiate soil3Notethatinthe18thCentury,theOttomanEmpire,PolishCommonwealth,andtheRussianEmpireeachmaintained their own border towns in the same general location for the purposes of borderprotectionandtrade. Hence“Balta”wasthenameofthetownthat fellontheOttomansideof thefrontier,whereasJozefgrodwasthePolishfortificationontheoppositesideoftheKodymariver.Itwasonly after theRussian empire annexed thePolish andOttoman territories (and in the caseofPervomaisk, only after the Bolshevik revolution) that the towns were incorporated into a singlepoliticalunitunderacommonname.

regions, climates, or transportation networks and it is not significantly different

from the rivers to the north (the Savran) or to the South. In short, the border

betweenthetwoempireswasnotinitiallyselectedonthebasisoflocalphysicalor

demographicattributesthatmightconfoundouranalysis.Theriverappearstohave

been chosen arbitrarily as a dividing line between the Polish and Ottoman

territories, possibly because it runs almost exactly from West to East and thus

servesasaconvenientNorth‐Southborder.

Figure1:TheStudyArea

(shaded,withtheKodymarivershowninred)

BoththePolishandOttomanterritoriesthatwereseparatedbythisborder

were acquired by the Russian Empire at approximately the same time in the last

decadeofthe18thCentury.TheRussianEmpireannexedthePolishterritorynorth

oftheKodymainthePolishpartitionof1793andacquiredtheterritorytothesouth

of the Kodyma at the conclusion of the Russo‐Turkish war in 1792 and the

dissolutionof theCrimeanKhanate. In theNorth, thePolishprovinceofBratslav

becametheRussiangovernorshipofPodolia(capitalinKamenets)atthebeginning

ofthe19thCenury. IntheSouth,anewgovernorshipofNovorossija(NewRussia)

was initially established and then divided in 1802 into the governorships of

Kherson,Ekaterinoslav,andTavrida. TheareastothesouthoftheKodymaunder

analysisherefellentirelywithintheKhersongovernorship.

II.StrategiesofImperialIncorporationandtheSourcesofRegionalCulture

The border between the Polish Commonwealth and the Ottoman empire

demarcatedsignificantdifferences inhowtheRussianEmpiregoverneditsnewly‐

acquired territories. Put in the simplest terms, the Empire pursued a strategy of

rapidcolonizationanddirect rule in theareasacquired fromtheOttomanEmpire

andastrategyof incorporationand indirectrule inareasthatwereacquiredfrom

Poland. These differences had a long‐term impact on the development of the

regionalculturesinthesetwoareas.

FormerlyOttomanTerritories:DirectRule

The goal for the territories acquired from the Ottoman empire was to

populate them as rapidly as possible, to cultivate the land for the purpose of

creating a tax base and to sustain sizable military settlements in a critical

borderlandoftheEmpire.Therewasnopre‐existingnobilitytodrawuponandthe

areaswerethinlypopulated.

To attract settlement, the Russian empire pursued a colonization strategy

focused on easy access to land, rapid upwardmobility, and direct rule (in many

regions—martiallaw).UnlikeotherareasoftheRussianempirewherelandholding

rightswerereserved for thenobility, in the territoriesannexed fromtheOttoman

EmpireandtheCrimeanKhanate,settlersofvirtuallyanybackgroundwereoffered

hereditaryprivateproperty.4AccordingtotheprinciplesoutlinedbyCatherinethe

Great,thegoalforthenewterritorieswas:

To divide the Lands amongst the Families which had none, and to

enablethemtocultivateandimprovethem.ThisDivisionoughttobe

madewithout Loss of Time, as soon as ever oneMan can be found

whowouldundertakeitonthoseTerms,thatnotaMomentmightbe

lostbeforetheWorkisbegun.5

4 JamesA.Duran,“CatherineII,Potemkin,andColonizationPolicy inSouthernRussia,”TheRussianReview,Vol.28,No.1(Jan.,1969),pages23‐24.5CatherineII,DocumentsofCatherinetheGreat,TheCorrespondencewithVoltaireandtheInstructionof1767intheEnglishTextof1768.Ed.ByW.F.Reddaway(Cambridge,England,1931)Article280,p.260)citedinDuran,“ColonizationPolicy,”page25.Capitalizationasitappearedintheoriginal.

Toincreasetheincentivestooccupyandimprovetheterritory,settlerswereoffered

taxexemptionsforperiodsof6,8,or16yearsdependingonanassessmentofhow

difficultthelandwastocultivate.Estatesofanysizeupto3888acres(48parcels)

weregrantedwithoutpayment topeopleofanyrank(zvanie). Toretain the land,

thenewownerneededonlytosettleonehouseholdperparcel.6

Nearly75%ofthenewsettlementregionwasdesignatedtosupportmilitary

settlers, who would reside in specific districts and be organized into regiments.7

Indicativeoftheremarkablepotentialforupwardmobilitythenewsettlementareas

entailed, military rank was a function of howmany settlers one could recruit to

resideonantheterritory.Amilitaryrankofmajorwastobegrantedtoanyonewho

recruited300militarysettlersor600ordinaryagriculturalsettlersfortheirestates.

Intheferventdesireformilitarysettlers,imperialdecreesestablishedamnestiesfor

fugitives,deserters,andrebelliousCossacks—thoseguiltyofanycrimeother than

murder—iftheywouldonlyreturntotheRussianempire.8

The goal was to settle, develop, and militarize the southern frontier as

rapidlyaspossible. Asaresultofthisdistinctivestrategyofcolonization,serfdom

was relatively absent from the territories acquired from the Ottoman empire

between1770and1796.9InsharpcontrasttotherestofEuropeanRussia,in1784

6Duran,“ColonizationPolicy,”page27.7 According to the decree (ukaz) of March 22, 1764, which set the policy for the occupation andsettlementofthenewterritories,anareaofsettlementwastobedividedinto70districts.These,inturn,were tobedivided into700parcelsof landof60.2or81acres. Of the70districts,52werereservedformilitarysettlers.Duran,“ColonizationPolicy,”page26‐27.8Duran,“ColonizationPolicy,”page29.9NotethatthiswastruealsofortheTiraspoluezd(futureTransdniester),whichwaspartofKhersongubernia,butnotfortherestofnorthernBessarabia(inso‐calledRussianMoldavia),whichwasstilldominatedbyMoldovanboyars.Foramapoftheterritoryseepage38ofE.I.Druzhinina,Iuzhnaia

only 2.7% of the rural population were privately­owned serfs and the area was

dominated by “state peasants,” of whom 80% were organized into military

settlements.10 Withnopriornobility, social rankwas acquiredwith relative ease

andanewRussianelitewasestablishedinshortorderandcloselyassociatedwith

militaryserviceandtheagriculturalproductionneededtosustainit.

Culturally, these territories became considerably more Russian in

orientation. In part this was due to the fact that the inhabitants were culturally

distinct.TheformerlyOttomanterritorieswerecolonizedbyimmigrants,although

many of these immigrants were Ukrainian‐speakers (fleeing serfs, resettled

Cossacks) moving southward. Throughout the 19th Century, with the spread of

urbanization and schooling, Russian culture and language came to predominate.

ThelanguageofthecitieswasRussian,aswasthelanguageandthecontentofthe

schools. 11 Moreover, as is common in areas of diverse settlement (Germans,

Moldovans,Jews,andSerbswereencouragedtosettleintheseareasinaneffortto

populatethemquickly)thisledtotheuseofRussianasalinguafranca.Druzhinina

alsonotesthatmanyoftheearlysettlersweredrawnfromotherpartsoftheempire

and that the regionwasquitedistinctive in its openness toprivateproperty, new

Ukraina v 1800­1825 [Southern Ukraine from 1800‐1825] (Moscow: Nauka, 1970). For thepopulationdataonTiraspoluezdseeDruzhinina,pages86‐87.10Duran,“ColonizationPolicy,”page35.Druzhininaprovidesfiguresfrom1801showingthatoutofatotalpopulationof488,965ofNovorossiskaiaGubernia,451,812weresmallagriculturalproducers.Of these, 151,573 were in military settlements. Only 28,166 were serfs. (Druzhinina, IuzhnaiaUkraina,Table1,page70‐71).Thereducednumberofmilitarysettlementsweredue,inpart,tothefact that theBugCossacksandtheOdessaGreekdivisionwerereleased fromservice in the1790s.Bothreturnedtoserviceby1819.11AnenormousgymnasiumandteachertrainingcollegewasconstructedinAnanivandservedtheseareasofnorthernKherson.

economicmodelsandeducation. Laterarrivalsassimilatedtoexistingresidents.12

Asaresult,theseweresomeofthemosteducatedprovincesintheRussianempire

by 1910 and thus the population was more directly exposed to and attached to

RussianhighculturethanothersintheWesternborderlandsoftheempire.

ThePolishRightBank:Indirectrule.

IntheareasthattheRussianempireacquiredfromthepartitionsofPoland,

therewasawell‐establishedsystemofserfdominplaceandanextremelypowerful

Polish landowning classwith a coherent sense of corporate identity, the szlachta.

TheseareaswestoftheDnieperandNorthoftheKodymariverhadinsomecases

been under the control of the Polish nobility for centuries. Leading up to the

partition, theRussianempirehadsupported thesePolishnobles in their struggles

withthePolishmonarchy,hadnegotiatedseparatetreatieswiththem,andinitially

didnothingtochallengetheirstatuswhentheywereincorporatedintotheRussian

empire. The Polish landowners in the newly‐acquired territories were given the

samerightsasRussiannobleswithintheRussianempire,andweregivende facto

control over their regions. The language of administration in these regions

remained Polish.13 The Polish nobility retained control over the schools, the

textbooks and curriculum remained as they were before the partition, and

educationwasextendedonly to theirowngentry.14 Insummary, theempirekept

12Druzhinina,IuzhnaiaUkraina,p.69.13AlekseiMiller,TheUkrainianQuestion,page143.14N.Hans,“PolishschoolsinRussia1772‐1831,”SlavonicandEastEuropeanReview,Vol.38,No.91(June1960),pages399‐400.

the previous social order intact and ruled indirectly, entirely through the Polish

nobility.15

After the firstPolishuprising in1830‐31, theRussian imperialgovernment

formally changed its approach to the Polish nobility in theWestern borderlands.

AccordingtoMoon:

Following the revolt of 1830‐31 the Russian authorities tried to

address the 'Polish' and 'peasant' questions simultaneously. They

wanted to reduce the power and influence of the unreliable Polish

nobility, and to gain the support of the Lithuanian, Belorussian and

Ukrainian peasantry, lest they be won over to the Polish national

cause. The social andpolitical situation in thewesternprovinces, in

particular thedisaffectionamong thePolishnobility,gaveNicholas I

andhis advisers theopportunity to introduce the typeof regulatory

reformofserfdomtheywerecontemplatingfortheempireasawhole.

Theydidnotneed toworryaboutalienating thenobility sincemost

Poleswerealreadydisaffected.16

But the Russian imperial governmentmade little headway in uprooting or

supplanting the szlachta’s dominance in the region. Decrees were signed which

were to transfer Polish landholdings to the state as punishment for the rebellion,15Moon,“InventoryReform,”page655.16Moon,“InventoryReform,“page655.

but for themost part thesewere never broadly implemented.17 Throughbribery

andinformalmeans,thePolishnoblespreservedboththeirlandholdingsandtheir

localpoliticaldominance.ThelatereffortsoftheRussianempiretotrytomobilize

thepeasantsandlowerclassestosupplantthePolishnobility18werestymiedbythe

successoftheszlachtainblockingtheprogressofeducationandsocialmobilization

thatwere soadvanced in theNewRussian territoriesandalsoextending toother

parts of the empire.19 The spread of the zemstvos (local peasant self‐governance

institutions) to theSouthwestern (Polish) regionwasachievedonly in1911. And

since thezemstvoswereproviding for the largest shareof theeducationexpenses

for peasant schools, spending on primary education per capita was three times

lowerinthoseareasoftheEmpirewherenozemstvoswereinplace.20Accordingto

Miller, “the assimilation potential of the poor, scattered Russian gentry, with a

weaklydevelopedcorporatespirit,couldhardlyoutweighthedominanceoftheage‐

oldPolishszlachta,unitedbythecommoncauseofresistancetothosemeasuresof

the imperial government that violated the property rights of the Polish

landowners.”21PeasantsinthePolishareasweresimplynotRussifiedtonearlythe

sameextent.

Moreover, towards the end of the 19th Century the imperial government

came to be more concerned about the dangers of politicized peasants than its

17Moon,“InventoryReform,”page655.18 To some extent the szlachta responded in kind, however, fostering the publication of books inUkrainianandcultivatingadistinctiveUkrainian identitybeforesucheffortswerecutshortby theValuevcircular(censoringUkrainianbooksandpreventingtheiruse inschools)andtheEmsUkaz(banningthepublicationofUkrainianliterature).SeeMiller,TheUkrainianQuestion,19N.Hans,“PolishschoolsinRussia1772‐1831,”SlavonicandEastEuropeanReview,Vol.38,No.91(June1960),pages404‐405.20Miller,TheUkrainianQuestion,page149.Eklof,RussianPeasantSchools,p.89,94.21Miller,TheUkrainianQuestion,page140.

disloyalPolishnobility,soeventheseeffortstomobilizeandRussifythepeasantry

tocounterbalancePolishinfluencewereabandoned.Strikingly,outofatotalof720

thousandserviceestatesinEuropeanRussia,Polesconstituted48%.22Insummary,

the Russian empire initially ruled through the Polish nobility, and Polish nobles

continuedtousetheirentrenchedpowertoblockeffortstoeducate,toRussify,and

to mobilize their serfs out from under them. As a result, literacy rates were

considerably lower in formerly Polish territory and the attachments to Russian

identityandculturewererelativelyweakintheSouthwesternRegion(Figure2).

Figure 2 Provinces with Literacy Above 50% in 1917

In Areas With Predominantly Russian Schools (1939 oblast boundaries)

22Miller,TheUkrainianQuestion,page143.

Tosumup,thecontrastincolonizationstrategiesandtheresultantcultural

differencesbetweenNovorossijaandtheSouthwesternregioncouldnotbestarker.

In Novorossija we see colonization and direct rule, with the establishment of

predominantly military settlement, high upward mobility, and a newly‐created

Russianelite that fostersacommonRussianculturethroughschools,urbanization

andcommerce.IntheareasoftheRussianempireacquiredatthesametimefrom

Poland,weseeastrategyofindirectrulerelyingontheexistingPolishnobility.The

landed aristocracy in these areas was able to preserve its dominance and to

successfully blockRussian imperial efforts to centralize the state and build direct

linkstothepeasantrythrougheducation.Thepopulationinhabitingthepreviously

Polishimperiallandshadlimitededucation,limitedattachmenttobroaderRussian

culture,andremainedlargelyUkrainianinitslanguage.23

Thecausalpathwaylinkingthebordertodistinctiveimperialtreatmentsand

theresultantculturaldifferencesisshowninFigure3below.

23Theparticularconfigurationofthetriadicrelationshipbetweenthemonarchy,thenobility,andthepeasantry appears to have had a significant impact on subsequent national identity among thepeasantry. In areas of EuropeanRussiawhere therewas anon‐Russiannobility—Germans in theBalticandPolesintheWesternborderlands—thenobilityblockedRussificationofthepeasantry.Inareaswhere thenobilitywasRussianOrthodox—regardless ofwhether thesewerenewly‐createdelites as in Novorossija or more established nobility on the Left Bank of the Dnieper river—Russification of the peasantry proceeded with considerably less resistance. On the NorthwestterritoriesseeWeeks2001,Thaden(ed.)1981.

Figure3

RussianColonizationStrategiesandRegionalPoliticalCulture

HowConsistentandDurableistheTreatment?

There are three areas of concernwith the use of this imperial border in a

quasi‐experimentaldesign.First,itisnotclearthattheoldimperialfrontierdefines

theboundariesof the imperial treatment, since Ihavenotbeenable todetermine

how heavily the border region was populated. Maps from the final years of the

PolishCommonwealthshowthedensityofroadsandsettlementsgrowingthinner

as one approaches the border region. Settlement appears to have concentrated

further to the north (the capital of the Polish province was Bratslav, which falls

outsideoftheareaofstudytotheNorth)andtheprimarytransportationroutesrun

tothecitiesofJozefgrod(partoftoday’sBalta,inOdessaoblast)andBohopol(part

of today’sPervomaisk, inMikolaevoblast).24 This appears tohave remained true

during the Russian Empire, asmaps of Podolia Guberniya from 1900 continue to

24 TheprimaryPolishmilitary and trading road, theKuczmanski Szlak, passed just to theWest ofBalta.SeeW.Faden,“AMapoftheKingdomofPolandandGrandDutchy[sic]ofLithuaniaincludingSamogitia and Curland, Divided according to their dismembermentwith the Kingdom of Prussia.”(London: W. Faden, Geographer to the King, 1799). (Note that this map has rather significantinaccuraciesinmanyrespectsandthereforemaynotbereliable.)

showthesouthernreachesoftheprovincetobemorethinlypopulated.25Itmaybe

thecasethattheareasclosestotheoldfrontierdidnotreceiveasstronganimperial

treatmenton thePolish side. It is almost certain that the imperial treatmentwas

nothomogenous.26Inthisway,measurementatthefrontierwouldbeexpectedto

underestimatetheimperialeffectontheformerly‐Polishside,sincewecanassume

theborderregionreceivedaweakertreatment.

Second,thereisnowaytoreliablyestimatethefidelityofthetreatmentand

controlgroupssincetherewassomemovementacrosstheborder.Thisconcernis

amelioratedsomewhatbythefactthatmobilitywassomewhat limitedbothinthe

RussianempireandintheSovietUnion,but220yearsisalongtimeandcertainly

interminglingandresettlementacrosstheboundarytookplace.Peasantsfromthe

countryside moved into the towns across the course of the 20th century, and

sometimes this meant moving across the former imperial frontier. Military

settlement continued in the area, which was home to several intercontinental

ballisticmissilesilos.Wewouldexpectmobilityandmigrationtodiminishevidence

of the imperial effect in the areas closest to the former frontier, since the edges

would“blur”overtime.

25 See Kudravtsev, “Karta Podolskoi Gubernij,” (1900). Downloaded on March 29, 2010 fromhttp://upload.wikimedia.org/wikipedia/commons/1/15/Podoliskaia_gubernia.jpg26 The story of Balta itself illustrates this problem. The fortified town on the Polish side of theKodymawascalledJozefgrodandwasaprivateholding,alongwithalloftheterritorybetweentheSavran and the Kodyma, of the Polish Count Liubomirski.When the Russian Empire annexed thisterritory,themilitaryactuallypurchasedJozefgrodfromLiubomirskiandrenamedthetownElensk.In1797,byorderofEmperorPavelI,theformerlyOttomantownofBalta(oppositeJozefgrodontheother side of the Kodyma) was absorbed into Elensk and the whole town, now straddling theKodyma,wasnamedBalta and incorporated into thePodolianguberniya. As a result,wehaveanareawherethePolishnobilitywasdisplaced,butwithinthePodoliangubernia.ThismeansthatforBalta, at least, the former imperial border does not demarcate the boundary of the treatment.BogdanSushins’kij,Balta:Misto,OsviacheneVichnistiu,istorichniese.(Odessa:Druk,2005),pages94‐97.

Third,unliketheidealdiscontinuitydesignorrandomizedfieldexperiment,

theanalysisofthetreatmentis“retrospective”ratherthan“prospective,”sincethe

assignmentofdistrictstotreatmentandcontroltookplaceinthe18thcentury.The

attributes of the two populations are only subsequently homogenized by a set of

commonhistoricalexperiencesthatdonoteradicatethepersistentculturaleffectof

thetreatmentthatisofinteresttous.Assuch,thedesignispredicatedontheidea

that the homogenizing experience of Communism combined with contemporary

physical proximity standardized many of the underlying conditions that are not

directlyaffectedbythe“treatmentvariable”,theimperiallegacy.Itmustbenoted,

however, that all of the control variables canonlybemeasuredpost‐treatment in

thiscaseandtheircurrenthomogeneityistakenasanassumption.Inthisrespect,

thedesignisratheratypical.

Althoughatypical,astrongcasecanbemadethatinthiscasetheassumption

is reasonable precisely because Soviet Communism was such an atypically

significanthistorical intervention. TheSoviet legacy indeederadicateddifferences

inallrespectsexceptfortheculture,whichmakestheassumptionofhomogeneity

reasonable in this case. Soviet Communism certainly standardized many of the

societal featuresthatmighthaveinfluencedthevotein2010. Intermsofpolitical

institutions, both experienced the same Soviet establishment of single‐party rule

and the same formal governing institutions. Agriculture on both sides of the

Kodyma was collectivized in the 1930s and privatized and de‐collectivized in

preciselythesamefashioninthe1990s.Bothsidesoftheriverhadsomeindustrial

factories, which were largely non‐operative or greatly scaled back by the time

electionswereheldin2010. Atthetimeoftheelections,villagesoneithersideof

theKodymariverweresimilartooneanotherinsizeandlayout.Becausebothsides

of the riverwere in the same oblast (Odessa), they experienced the same oblast‐

level influences. WithinOdessaoblast, thepopulationswereexposed to thesame

radioandtelevisionbroadcasts.TheysharedthesamelegacyaspartsoftheSoviet

Union and they were not permitted to express many of their differences in the

Sovietperiod.Inthisway,noneofthepartisan‐reinforcingtendenciesreportedby

Converse, for example, could be expected to operate.27 In essence, I rely on the

standardizingtemplateofSovietCommunismtoprovideexogenouspost‐treatment

homogeneityinabroadrangeofpotentiallyrelevantcausalfactorssothatwemight

better identify the effects of the persisting regional cultural differences that stem

fromthedifferentpre‐communistcoloniallegacies.

FollowingJasonWittenberg’spioneeringworkonthepersistenceofpolitical

party loyalties in Hungary, we might expect local church institutions to play a

particularly important role in preserving pre‐Communist traditions.28 Yet church

institutionsonbothsidesoftheborderweredecimatedduringtheearlyyearsofthe

Bolshevik regime. Moreover, because there were sharp restrictions on church

attendance, with very few exceptions the churches in the region were

decommissioned and converted to other functions or, more commonly, were

physically destroyed. The large Franciscan monastery in Chechelnik (Vinnitsa

oblast),forexample,wasconvertedintoagymnasiumandthenintoanautomotive

27Converse1968.28 Jason Wittenberg, Crucibles of Political Loyalty: Church Institutions and Electoral Continuity inHungary(CambridgeUniversityPress,2006).

repair shop under Communism. Villages on both sides of the border rebuilt new

churches following the Communist collapse, but there was no institutional

continuityofChurchinstitutionsinthispartofUkraineinthewaythattherewasin

HungaryorinotherareasofCentralEurope.

Moreover,theculturaleffectsoftheinitial“treatment”wouldbeexpectedto

survivecommunism.Priorresearchhasshownthatdifferencesinnationalloyalties

and other pre‐Communist cultural legacies of the type we are discussing here

persisted,preservedwithinfamiliesandlocalcommunities,despitetheexperience

ofCommunismandtheSovietstate’ssuccessineradicatingthemfrompubliclife.29

In this respectwemight think of the Soviet Communist experience as a powerful

sieveinwhichallpublic,institutional,andsocio‐economicfeaturesofasocietywere

all caughtup, radically transformed,andstandardized,butcertainprivatecultural

attributes slipped through and were preserved relatively unaltered through the

moresubtlesocializinginfluencesofhomeandfamily.

III.RegionalCultures,PoliticalPartiesand2010UkrainianElections

What isthe linkbetweenthesecultural legaciesandsupport forcandidates

in the 2010 elections? As in previous elections, issues surrounding the place of

Russian culture in Ukrainian life played a central role in the campaign and in

differentiatingthecandidates.ThePartyofRegions,headedbyViktorYanukovich,

was initially foundedonaplatformsupportingtheuseofRussian languageandof

preserving ties to Russian culture. Yulia Timoshenko was not an anti‐Russian

29Darden,ResistingOccupation,forthcoming.DardenandGrzymala‐Busse,“TheGreatDivide,”2007.

candidatebyanymeans,butshewasassociatedwiththeOrangecoalitionandwith

agovernmentthatcontinuedtopursueUkrainianizationofpubliclife.Ininterviews

conducted in the summer of 2009 prior to the elections, respondents clearly

associatedTimoshenkowiththeOrangecoalitionthatcametopowerfollowingthe

elections of 2004 and, in Odessa oblast, shewas associatedwith policies like the

mandatetodubforeignfilmsintoUkrainianandotherdivisiveculturalpolicies.In

short, cultural issueswere salient in the 2010 Presidential elections, just as they

wereinthe1994Presidentialelections,the2002Parliamentaryelections,the2004

Presidentialelections,andthe2006Parliamentaryelections.

Inpart,culturalissuescontinuetobesalientbecausethedevelopmentofthe

mainrivalparties,theBYuTandPOR,tookplaceinacontextinwhichtheseissues

werecontinued tobe theprimarymeans fordifferentiating the twomainpolitical

blocs. Thedifferences in the stated economicpolicies of the twoblocs cannot be

easily discerned, the continued competition between the parties maintains the

salience of the cultural cleavage as away that they are able tomobilize support.

Whileitisclearlythecasethattherearedifferentbusinessfactionsassociatedwith

eachpartyandwitheachpresidential candidate, inorder topreserve thepolitical

powerneededtomaintaintheireconomicholdingsthesebusinessfactionsneedto

winelections. Whichpartyorcandidatewinsinagivenpollingdistrictappearsto

bea functionof the cultural inclinationsof thevoter (which isdeterminedby the

imperial legacy) and which party is dominant in the oblast (which is also

determinedbytheimperiallegacyandtheboundarylinesoftheoblasts).

Ifwethenassumethatthetreatmentimpliesthatpolingdistrictssouthofthe

KodymashouldbemoreinclinedtosupportthecandidacyofViktorYanukovichand

pollingdistrictstotheNorthoftheKodymashouldbemoreinclined,ceterisparibus,

to support the candidacy of Yulia Timoshenko, the results in favor of an imperial

effect are quite positive. Despite the long period of time since the initial (pre‐

Communist) treatment and the potential for cross‐contamination of populations

acrosstheinitialboundarybetweenPolishandOttomanterritories,wefindthatthe

differenceinpoliticalsupportinthetwoformerlyimperialterritoriesinsubstantial.

TheresultsareshownbelowinTable1.

Table1ImperialLegaciesandElectoralResultsinthe2010Elections

(withina60kmbandincludingnorthernOdessaandsouthernVinnitsaoblasts)

Turnout Yanukovich Timoshenko

Polish Territories 69.4% 31.2% 64.7%

New Russia 62.4% 66.0% 29.8%

Asshowninthetable,withinthisapproximately60kmwidestripalongtheKodyma

river, and dividing the population into the two categories of formerly Polish

territoriesandtheformerly‐OttomanareasthatbecametheprovinceofNovorossija,

we see a strong association between imperial legacy and voting behavior. In the

formerly‐Polishareas,TimoshenkobeatsYanukovich2‐1.Theresultsarereversed

fortheareasthatwereoncepartofNewRussia.

Uponcloserobservation,itisclearthatthestrengthoftheseresultsstemsin

part from differences in the oblasts into which these districts fall. South of the

Kodyma,allof thepollingdistricts fall inOdessaoblast. Immediatelynorthof the

KodymaanduptotheSavranriver,thedistrictsalsofallwithinOdessaoblast,but

this is a fairly narrow strip of land containing only 51 polling districts. The

remainingdistrictsnorthoftheKodymafallinVinnitsaoblast,whichisdominated

byBYuT.

Toeliminate thepossibilityofbothoblast‐levelrather thanpurely imperial

effects, it is useful to control for oblast‐level effects by comparing the results for

different imperial territoriesonlywithinOdessaoblast, i.e. the51pollingdistricts

north of theKodyma to the 45 closest districts to the South of theKodyma. The

Kodyma in thiscasealsopartiallysplits tworaions (BaltaandLiubashiv), soeven

localgovernanceisheldconstanttoacertainextent. Inmanycasesthesedistricts

are walking distance apart, with towns occupying opposite sides of the river.

Barring any effects of the imperial legacy, we would expect no difference in the

electoralbehaviorofthesetwogroups.

Yet even herewe find a significant effect. Both themedian and themean

percentage of the vote for Yanukovich in districts within Odessa oblast that are

South of the Kodyma river are approximately 10 percentage points higher than

thoseNorthoftheKodyma.

Table2:MedianVoteShareforCandidatesinthe2010UkrainianPresidentialElections(secondround)withinOdessaOblast,comparingacrossthe51pollingdistrictsnorthoftheKodymatothe45nearestdistrictstotheSouth(narrowband)

TURNOUT Yanukovich Timoshenko South of the Kodyma (New Russia) 63.9% 65.2% 30.3% North of the Kodyma (Polish) 69.4% 53.6% 42.2%

The effect is certainly not as strong as when we include the polling districts in

Vinnitsa oblast, and Yanukovich wins in both districts. Yet the difference is still

substantialandYanukovichactuallylostthesedistrictsNorthoftheKodymainthe

thirdroundofthe2004Presidentialelections.

PartyMachinesandtheOblasteffect

These findings are particularly striking given the close proximity of these

districts, our anticipation that the treatment effects might be somewhat weaker

rightalongtheformerimperialboundary,and,particularlygiventheimportanceof

oblast‐levelfactorsonthevote. Thestrengthofoblastleveleffectsisclear. Inthe

past decade, both the Party of Regions andBYuThave developed extensive party

organizationsattheoblastlevel.Withfewexceptions,onecandidatenotonlywins

theoblastasawhole,butineverysingleelectoraldistrict(andthevastmajorityof

pollingstations)withintheoblast.Whymightthisbethecase?

Field interviewsthatIconductedpriortotheelections inJulyof2010with

voters, party strategists and agitators revealed that each of the parties had a

hierarchically‐organized campaign structure, centered in the regional capital, and

linkedthroughachainofcommandtoatleastoneactivist(agitator)ineverysingle

poling district. At this lowest level, the agitator was responsible for delivering

voterstothepolls, fordistributingmaterialsandpersuadingvoterstosupportthe

candidate,and,ifnecessary,vote‐buying(withpricesrangingfromapproximately4

USDto10USDdependingonthearea,ageandprofessionofthevoter).Becausethe

strengthandresourcesofthepartyorganizationisconcentratedattheoblastlevel,

itmeantthatthemostpowerfulorganizationattheoblastlevelwastypicallybetter

able to mobilize voters in all districts throughout the oblast. Agitators with the

dominantpartyreceivedhighersalaries,hadmoreresourcesavailabletothem,and

the partieswere better able to recruit themost talented cadres towork on their

behalf.Inmanyways,thestrengthofthepartyorganizationlaymoreinitspowerto

draw the most respected local authorities into the ranks of the party than in

mobilizingvoterstothepollsor inbuyingvotes.30 Thiswascertainlythecasefor

PORinOdessaoblastandforBYuTinVinnitsaoblast.31Itwouldbeverysurprising

30Theresourcesavailableforvote‐buyingweresimplytoosmalltobeabletoaccountforthesizableoblast‐leveldifferencesinvoting.AccordingtointerviewswithonePORagitatorspeakingaboutthe2004elections,eachagitatorresponsible forapollingdistrict (village)wasgiven1200hryvnia (atthetime,approximately$300)forthecampaign. Thiswastocovertheirpersonalsalaryaswellasthe resources tobeallocated topurchase the supportofvoters. Votes costapproximately40hrn.With such limited funds, the agitators were unable to buy a substantial enough portion of theelectoratetoaccountforthedifferencesacrossoblasts.31Iamnotspeakinghereoftheuseofso‐called“administrativeresources”toinfluencetheelectionoutcomes,i.e.theuseoftheapparatusofthestateadministrationtocampaignandapplypressureonbehalf of a candidate or party. The use of administrative resources has declined considerably inUkrainianelectionsfollowingtheelectionsof2004.

ifdifferencesinthepartyorganizationsaloneaccountfortheenormousoblast‐level

differences,butthisiscertainlypartoftheexplanation.32

Yet party strength at the oblast level cannot be easily separated from the

regional cultures and the imperial legacy. The initial strength of the party

organizationsattheoblastlevelwasitselflikelyafunctionoftheconsistencyofthe

Partyplatformwiththegeneralculturalorientationofthemajorityofvotersinthat

oblast. Giventhatregionaloligarchswishtobeassociatedwiththewinningparty,

andthewinningpartyistheonewhoseappealsalreadyresonatewithvoters(hence

requiring fewer vote‐buying resources), these trends lead to strong pressures

towards the dominance of one party at the oblast level. One can see this in the

electoralresults.Thereareveryfewoblastswhereonepartyorcandidatedoesnot

winalloftheelectoraldistrictswithinanoblast,andinwhichoneparty/candidate

doesnotwinby a largemargin. The initial cultural orientationof thepopulation

appearstohavedeterminedwhichpartywouldbedominantinagivenprovince,but

thishasmeantthatthepartymachinessubsequentlyhaveadditionalresourcesand

canextendandexpandtheirdominance.

VI.ConclusionsandImplications

Howbesttointerpretthesefindings? Ithinkthatthedatasuggestastrong

link between Russian imperial policies, contemporary cultural differences, and

32NordoestheboundarybetweenVinnitsaoblastandOdessaoblastdoesnotdemarcatea

sharplinguisticboundary.WhileitistruethattherearemoreRussiansintheraionsontheOdessasideoftheOdessaVinnitsaborder(approximately5‐6%ofthoseraions,asopposedto1‐3%ontheVinnitsaside),theseminordifferencesaresodwarfedbythedifferencesinvotingbehaviorthattheyareunlikelytoaccountformuchofit,ifanyatall.

patterns of voting in Ukraine. Voting largely cleaves along former imperial lines.

Even when we only examine polling districts within Odessa oblast, there are

substantialandsignificantdifferencesbetweentheresultsofdistrictsdependingon

whether they fell within the territory thatwas once called “NewRussia” and the

formerlyPolishareasoftheRussianempire.Butwhy?

I have suggested one possible process thatmight be driving this effect, i.e.

thattheregionalelectoralcleavagereflectsaculturalcleavage,andthatthiscultural

cleavage can ultimately be traced to different strategies of incorporation initially

pursued by the Russian empire in the 18th Century and carried through the

development of these regions in the 19th and early 20th Centuries. North of the

Kodyma, the local nobility empowered by indirect rule was more interested in

creating impediments to peasant power and peasant education. As a result,

peasantsinareaswherethePolishszlachtaretainedlocalpowerweremorelikelyto

be educated only after the Bolshevik revolution, during a period of “nativization”

thatwouldimbuethemwithUkrainianratherthanRussianculture.33Theopposite

was true in the formerly Ottoman areas, where a new Russian elite was created

following the annexation, one that fostered rapid development in education,

urbanization, and the spread of Russian culture and loyalties. In this way, the

33YuriSlezkine,“TheUSSRasaCommunalApartment,orHowaSocialistStatePromotedEthnicParticularism,”SlavicReview53,2(Summer1994),414‐452.TerryMartin,AffirmativeActionEmpire:Nations and Nationalism in the Soviet Union, 1923­1939 (Ithaca: Cornell University Press, 2001).Provincial officials interviewed along the study areamentioned that Party communicationswithinVinnitsa oblast, for example,were always in Ukrainianwhereas theywere typically in Russian inOdessaoblast. TerryMartin’sworkalsosuggeststhattherewasresistancetokorenizatsiaandtheforceduseofUkrainianmetconsiderableresistanceinthemoredevelopedsouthernandurbanpartsof the Ukrainian SSR. I suspect that further research would bear out the argument that regionalculturaldifferencespersistedunderCommunism,and thata strongattachment toRussian identityandculturepersistedintheareasofNewRussiaevenunderCommunism.

specificmechanismthatIhavesuggestedlinkingtheimperialpastwiththeelectoral

presentdrawsonmypreviousresearchlinkingschoolingwiththedevelopmentof

regional cultures, andwhich suggests that such culturaldifferences arepreserved

despite themany changes that takeplaceunderSoviet rule. Theargument about

imperiallegaciespresentedherejustdeepensthatcausalchaintoshowthecolonial

originsofdifferencesinhowpeasantswereeducated.

The explanation that I have provided here is consistentwith the available

data, but it certainly does not exclude the possibility of alternative processes and

accounts. Precisely what is generating the imperial effect, however, demands

further research and while the available data. Given the spatial discontinuity

researchdesignemployedhere,wecannotknowwhichimperialexperiencesmight

bedrivingthecontemporarydifferences(onlythattheimperiallegacy—takenasa

whole—appearstohaveasignificanteffect).

There were many differences between the (formerly Polish) Podolian

governorship and the “New Russian” province of Kherson within the Russian

empire. Thelatterwaspopulatedalmostentirelybymigrants. Evenaslateasthe

1897censusapproximatelyhalfofthepopulationoftheprovincehadnotbeenborn

intheprovince. ItsnobilitywasRussianandtheterritoryhad, for themostpart,

neverknownserfdom.ThisiscontrasttothePolish‐dominated,lessurbanized,less

educated,anddeeplyenserfedpopulationofPodolia. Given thecomplexityof the

differences between these two imperial provinces, we have difficulty identifying

preciselywhichimperial“treatment”mightbedrivingtheeffect. Furtherresearch

is needed to determinewhether othermechanismsmight be atwork, and also to

determine whether there were similar mechanisms at work in other areas of

Ukraine(Poltavaoblast,forexample).

Nonetheless,thefindingspresentedhereareusefulindirectingourattention

totheimperialandlargelyculturalsourcesofthecontemporaryelectoralvariation.

They suggest that empires leave cultural footprints that have significant

implications for explaining contemporary voting in ways unanticipated in most

models of electoral behavior, and which significantly impact both the domestic

politicalcleavagesandinternationalrelationsofthecountriesinquestion.