cherry picking 2015 swiss federal elections: the influence of on …¼hlmann et al.pdf · 2017. 10....

CherryPickingatthe2015SwissFederalElections:TheInfluenceofElectoralCampaigningonPanachageandCumulation

MarcBühlmann,DavidZumbach,MarlèneGerber,AnjaHeidelberger1

Papertobepresentedatthepanel“PartiesinCongestedandContestedPoliticalArenas”,

ECPRConferenceinPrague,7‐10September2016

Abstract

Ourcontributionanalysestheinfluenceofelectoralcampaigningoncandidates’successatthe2015SwisselectionstotheNationalCouncil.Concretely,weaskwhetherandtowhatextenttheintensityandcontentofacandidate’scampaignexertsapersuasiveeffectonvoters.Indoingso,wemakeuseofaratheruniquesettingembeddedintheopen‐listPR‐systemofSwitzerland,

namelyallowingvotersnotonlytoduplicatecandidatesfromachosenlist(cumulation)butalsotoaddcandidatesfromotherpartiesandliststotheirselectedlist(panachage).Usingthe

amountofvotesacandidatereceivedfromvotersfavoringotherpartiesorlistsgivesusanideaaboutacandidate’spersuasivepotentialtogainvotesoutsidehisclassicalvotersegment.Weassessacandidate’scampaignbasedonacollectionofalmost4,000politicaladvertisementsgatheredin50importantsupraregionalandregionalnewspaperscoveringall26cantons,i.e.electoraldistricts.Theinfluenceoftheelectoralcampaignisexaminedusinghierarchicalmodels,bymodellingacandidate’selectoralsuccessforeachballotlistinhisdistrict.Thisprocedurebearstheadvantagethatweareinabetterpositiontomodelthevotesofan

individualcandidateinrelationtothealternativesavoterwasofferedonotherpartylists.Wefindthattheformaswellastheintensityofelectoralcampaigningmatterforelectoralsuccessintermsofboth,votesgainedfromcumulationaswellasvotesgainedfrompanachage.Incontrasttofindingsfromotheropen‐listPR‐systems,wefindthatchallengersprofitsubstantiallymore

fromelectoralcampaigningthanincumbentsdo.

1Theorderofauthorshasbeendeterminedbythrowingthedice.

Introduction

Theprincipleofselectingrepresentativesbycitizensviapopularelectionsliesattheveryheart

ofrepresentativedemocracies.Inmostofthecases,thechoiceofthecandidatesisdetermined

bythepartyavoterdecidestosupport.Yetinsome–andmostlyEuropean–democracies,

citizensalsohavethepossibilitytovoteforcandidatesoftheirchoice,withinorevenacross

partylists.Asaconsequence,candidate‐specificfactorsbecomemoreimportantforelectoral

successinmultimemberdistrictsthatallowforthesocalled“preferentialvoting”comparedto

districtsthatdonot(e.g.,Sartori1976;Katz1986;CareyandShugart1995;Karvonen2004;

Shugartetal.2005).

Sincevotersdemandmoreinformationonacandidatewhentheyareallowedtochoosefroma

largersetofcandidates,researchersstartedtoexaminetheeffectofcandidatespecificfactors

suchaslocaltiesandlower‐levelpoliticalexperience(e.g.,Shugartetal.2005;Tavits2010;Put

andMaddens2015),incumbency(e.g.,Moon2006;MaddensandPut2013),acandidate’s

positiononthelist(e.g.,Lutz2010;Wautersetal.2010),hismediapresence(vanAelstetal.

2008;ElmelundandHopmann2012)andtheelectoralcampaign(e.g.,Bowleretal.1996;

Maddensetal.2006;MaddensandPut2013;SpieringsandJacobs2014)onelectoraloutcomes

inpreferentialvotesystems.

Whilemostofthesefactorsarepredefinedbyacandidatescurriculumvitae,acandidatestillhas

thepossibilitytopromotehimselfandhiscapabilitiesviaelectoralcampaigning–something

whichisoftenmadeuseofinsystemsknowingpreferentialvoting(Katz1986:101;Karvonen

2004).Uptonow,however,evidenceaboutwhether“effortstocultivatepersonalvotepayoff”

(Tavits2010:216)isinconclusive(see,e.g.,Bowleretal.1996;Maddensetal.2006;Elmelund

andHopmann2012;SpieringsandJacobs2014).Furthermore,theimpactofcampaignactivities

islikelytovaryacrosscountries(SpieringsandJacobs2014:217‐18),whichcallsformore

researchindifferentcontexts.

ThispaperanalysestheimpactofcampaignactivitiesonpreferentialvotinginSwitzerlandin

theframeworkofthe2015generalelectionstotheNationalCouncil.WefocusonSwitzerland

fortwoparticularreasons.First,wemakeuseofauniquedatasetconsistingofcampaign

advertisementspublishedinmorethan50importantnationalandregionalnewspapers

(Bühlmannetal.2015).Withthehelpofthisdata,wearenotonlyinapositiontoexaminethe

impactoftheintensityofpersonalcampaigns,butalsotheeffectoftheirformandcontent.

Second,theelectoralsysteminSwitzerlandischaracterizedbystrongpreferentialvoting

(Karvonen2004:208).Inmostofthecountriesthatholdacertainformofpreferentialvoting,

onlycandidatesfromthesameparty(orlist)canbechosen.2Thusthepartychoicehasavery

predominantrole,implyingthatvoterslookatcandidates’attributesonlyafterhavingchosena

preferredpartyorlist.ThisisdifferentinSwitzerland:Swissvoterscan1)cumulateacandidate,

i.e.puthimtwiceonalist(mostlyimplyingthatanothercandidateneedstobedeletedfromthe

list),and2)andmostrelevantfortheaimofourinvestigation–includecandidatesfromanother

partyorlistontothechosenlist(socalledpanachageandagainimplyingtheexclusionofa

candidatefromachosenlist).Thistogetherwiththefactthat,afterdeterminingtheamountof

seatsapartyreceives,thedistributionofseatsisdictatedbythesumofindividualcandidates’

votes,islikelytopresentstrongincentivesforpersonalvote‐seekingbehaviour(SelbandLutz

2015;seealsoShugartetal.2005).Moreover,thisparticularityallowsustoexaminepotential

andunexploreddifferencesintheimpactofcampaigningbetween“loyal”preferentialvotes,i.e.

thecumulationofacandidatefromachosenlist,and“alien”preferentialvotes,i.e.thepanachage

ofcandidatesfromotherthanthechosenlist.

Thesubsequentsectionsofthepaperareorganizedasfollows.Inthenextsection,wediscuss

expectationsfortheimpactofelectoralcampaignsonpersonalvotes.Subsequently,wedescribe

thedataaswellasthemethod(section3)weusedinouranalyses,whichiswhatwepresentin

section4,beforewediscusstheimplicationsofourfindingsandconclude.

Campaigneffects

Broadlyspeaking,therearetwosystemsforparliamentaryelections:themajorityandplurality

versustheproportionalelectoralsystem.Themainaimofthefirsttypeistheselectionofthe

mostqualifiedcandidateinsingle‐memberdistricts.Thiswinner‐take‐all‐methodleadstothe

emergenceoftwo‐partysystems;withstrongincentivesforvoterstochooseamongcandidates

oftwostrongpartiesthatholdadequatechancestogainaseat(Duverger1959).Inthis

situation,apluralisticsocietyisonlybadlyrepresented.Adequaterepresentationofsuch

societalpluralityisthebasicaimoftheproportionalrepresentationsystemwherevotershaveto

choosebetweenseveralpartiesrepresentingdifferentideas.Inthissystemitisthepartiesthat

nominatecandidatesontheirlistsandcandidatevotingislesspronouncedthaninmajorityand

pluralitysystems.

Yetthecandidatesthatthepartiesputonthetopoftheirlistsmaydifferfromvoters’

preferences(Katz1997;Farell2011).Whileinmostcountrieswithproportionalrepresentation

2LeaningonKarvonen(2004)wecandifferentiatetwotypesoflistproportionalrepresentationsystems:insystemswhereyoucanonlychoosethepreferredpartyandputtheirlistintotheballotbox(closedlist),candidate‐centeredvotingisoflowimportance.However,therearealsosystemsallowingvoterstoinfluencewhichcandidateswillrepresentthem(openorfreelist).Differentrulesallowforarankingofthecandidatesonthelist(e.g.Ireland),forgivingsocalledpreferencevotesforspecificcandidatesonalist(e.g.Belgium)orallowingforfreelychoosingseveralcandidatesfromapartylist(e.g.Finland).

system,votersaresimplyallowedtochoosetheirpreferredpartyandputtheirlistintothe

ballotbox(closedlistsystems),insomecountriestherearerulesgivingthevoterssome

influenceonthechoiceofthecandidates.Suchopenlistelectoralsystemstrytocombinethe

advantagesofmajoritarianandrepresentationalsystemsallowingforpartyaswellascandidate

choice,i.e.anideologicalaswellasapersonalrepresentation.

Intheirpathbreakingwork,CareyandShugart(1995)arguedthatinsuchanopenlistelectoral

system,candidateshavestrongincentivestoorienttheircampaigntowardstheirown

reputation.Ifvotershavethepossibilitytodecideoncandidates,theyaskforinformationabout

thecandidates.Thus,“candidatesmayattractsupportforwhotheyare,orwhattheyhavedone,

orwhattheymightdo,ratherthansimplybecauseofthepartytowhichtheybelong”(Marsh

2007:501).Severalstudiesshowedthatcandidate‐centredfactorsareindeedimportant

comparedtomereparty‐centredfactors(Karvonen2004;Marsh2007;Moser&Scheiner2005;

Shugartetal.2005;Swindle2002).

Thefollow‐upquestionis:whichcandidatespecificfactorshaveaninfluenceonelectoral

outcomes?Researchonthisquestionhighlightstheimpactoflocaltiesandlower‐levelpolitical

experience(e.g.,Shugartetal.2005;Tavits2010;PutandMaddens2015),incumbency(e.g.,

Moon2006;MaddensandPut2013),acandidate’spositiononthelist(e.g.,Lutz2010;Wauters

etal.2010)orhismediapresence(vanAelstetal.2008;ElmelundandHopmann2012)on

electoralsuccess(foradiscussionontheassumedeffectsofthesefactorsseebelow).

Besideshavinganinterestingcurriculum,acandidatecanalsopromotehimselfbycampaigning.

Campaignspendingisanimportantwaytoenhanceacandidate’svisibilityvisàvisthe

electorate.Severalstudiesfoundapositiveimpactofcampaignspendingonelectoraloutcome

(e.g.,Maddensetal.2006;MaddensandPut2013;SpieringsandJacobs2014;however,see

Bowleretal.1996;Elmelund‐PraestekaerandHopmann2012forcounter‐evidence).

Furthermore,thiseffectmightbestrongerforchallengers,sincevotersarelessornotatall

familiarwithnewcandidatesthantheyarewithincumbents.Thus,inthelattercase,the

“marginalreturnofcampaignexpensesshouldbelowerthanforchallengers,whohavetogain

visibilityduringthecampaignandcanonlydosothroughspending”(Maddens&Put2013:

853).ThisrelationshiphasrepeatedlybeendemonstratedintheU.S.(e.g.,Abramowitz1991;

Jacobson1978;Moon2006).Ontheotherhand,Maddensetal.(2006)couldnotdetectsuchan

effectinBelgiumwithitsproportionalelectoralsystemandsemi‐openlist.MaddensandPut

(2013:853)hypothesizethatinproportionallistsystemswithlargeconstituencies–the

averageBelgiumconstituencycounts13.6seats–theincumbentMPsarenotaswell‐knownas

theonesinsmallerconstituenciesorinmajoritariansystemswithsingle‐memberdistricts.Since

inlargeconstituencies,itismoredifficultforvoterstorecallalltheirincumbents,thecampaign

effectbetweenincumbentandchallengersislikelytodiminishinlargerconstituencies.On

average,the26Swissconstituenciesare,with7.7seatsperconstituency,ofsmallersizethanthe

Belgiumones.3

However,inourcontributionwedonotsolelyfocusontheamountofcampaignexpenses.We

relyonauniquedatasetconsistingofcampaignadvertisementspublishedinmorethan50

importantnationalandregionalnewspapers(Bühlmannetal.2015).Withthehelpofthisdata,

weaimattestingtheeffectofthevisibility,theuniquenessandthelocalanchoringofa

candidate’scampaignonhiselectoraloutcome.InSwitzerland,candidatesarenotobligedto

disclosetheirfunding,whichiswhywehavetorelyonothermeasuresofcampaigning.

Althoughweagreethatothercampaigningstrategiessuchasadvertisinginsocialmediagainin

importance,peoplestillpaymuchmoreattentiontopoliticaladvertisementsinprintmediathan

towebpagesconcerningtheelections(Selects2011).4Inthefollowing,wearguethatitisnot

onlythecampaignexpenses,i.e.amereexposureeffectthatmatterbutalsothestrategyaswell

asthecontentofacandidate’scampaign.Toattract(alien)voters,acandidatemustbevisible,he

mustshowthatheisdifferentfromothercandidatesandhemustproveconnectednesswithhis

constituents.

Visibilityincreaseseligibility:Candidatesthataremorepresentduringthecampaignhavea

comparativeadvantageovercandidatesthatdonotpresentthemselves–independentofthe

contentoftheircampaign.First,theyincreasetheirvisibilityamongthevotersandsecond,they

mightprofitfromamererepeatedexposureeffect,i.e.fromthemechanismthatrepeated

exposuretoastimulusincreasesitsaccessibilitytotheindividual’sperception(Zajonc1968:1).

Suchaneffectisparticularlypronewhenindividualvoterspursuealow‐costinformation

strategyandrelyoncuespresentedtothem(e.g.,Moonsetal.2009,seealsoSteenbergen2010).

Furthermore,visibilitycanbeincreasedbyclever‘productplacement’(e.g.,Geise&

Brettschneider2010).Acolouredadvertisementonthefrontpageofanewspaperforexampleis

supposedtoattractmoreattentionthanablack‐whitedesignedadvertisementonthe

advertorialpageinthelastpartofthenewspaper.

Inordertocaptureamereexposureeffect,wetakethefrequencyofadspublishedper

candidate.Furthermore,weincludevariablesthatcapturethevisibilityofacandidate’s

campaign.Thesearetheproportionofacandidate’sadspublishedonthefirstpageofabundled

newspaperandtheproportionofadsprintedincolour.

3YettherangeintheSwissconstituenciesisquitesubstantial:sixoutof26cantonshavemorethan10seats,twoofthemmorethan20(Bern25;Zurich35).4InSwitzerland,electoralcampaigningontelevisionandradioisnotallowed.

Individualizedcampaigninghelpstostandout:Acandidateaimingatattractingvotersthatdonot

intendtovoteforhispartyshouldshowsomedegreeofindependence.Ofcoursethereisafine

linebetweenattractingalienvotersandscaringofftheownfollowers.However,sinceparty

cohesionisassumedtobesmallerinpreferentialvotingsystems(Karvonen2004;Katz1986),

showingacertainamountofautonomycanhelptowinelections.Acandidatecandemonstrate

suchautonomybyconductingahighlypersonalizedcampaign,i.e.acampaignwherethegoalis

topromotethecandidateinsteadofthepartyheorsheisrunningfor(seee.g.,Zittel&

Gschwend2008;Selb&Lutz2015).InSwitzerland,candidatesareoftensupportedbyimportant

associations.Wearguethatthedegreetowhichacandidate’scampaignwasexternallyfunded–

weusethepercentageofacandidate’sadswherethisisthecase–accountsforthedegreeof

independencefromhisparty.Acandidatefundedbyanassociationcanpresenthimselfina

differentlightandatthesametimedistancehimselffromothercandidatesonhisownlist.Asan

additionalmeasure,weincludetheproportionofadswherethecandidatewascampaigning

withouthiscombatants,assumingthatadvertisingtogetherwithseveralcandidatesfromthe

samepartypromoteshigherpartyunity–attheexpenseofsinglecandidates.Additionally,by

showinghisconnectionwithanimportantassociation,acandidatealsogainsinuniqueness.

Demonstratinglocalrootednesspaysoff:Severalstudiesdemonstratetheimportanceof

candidates’local‐levelpoliticalcharacteristicsforelectoralsuccess(Tavits2010).Localtiesare

seenas“acrucialpersonalvote‐earningattribute”(PutandMaddens2015:608).Occupyinga

localoffice(Tavits2010)orlivinginbigmunicipalities(PutandMaddens2015)canincreasethe

numberofvotesacandidategets.Localbirthplaceandlocal‐levelpoliticalexperiencearesignals

forfamiliaritywithlocalinterests.Therefore,votersseemtorewardcandidateswithlocalroots.

However,wearguethatacandidateshouldactivelydemonstratethestrengthofhislocalties

duringhiscampaigntoconvinceevenmorevotersthanonlythosealreadyknowinghim.To

capturewhetheracandidate’scampaignislocallyrooted,weintroducedtheproportionofads

containingoneormoretestimonialswherelocalcelebritiespromotethecandidate.

Furthermore,weusetheproportionofadscallingattentiontoeventswherevoterswereableto

meetthecandidate.Weassumethatrealcontactwithvotershasahighpotentialtoincreasea

candidate’selectoralsuccess(Karvonen2004).

Ofcourse,lookingatthecontentratherthantheextentofcampaigningshouldalsoinclude

negativecampaigning.Negativeadvertisingseemstogrowinimportance(Geer2006).Providing

voterswithinformationaboutpoorcharacteristicsofanaliencandidatecanbeaninstrumentto

alienatevotersfromthecandidatesufferingfromnegativecampaigning(FridkinandKenney

2011).Yetnegativecampaigningcanalsoprovokeabacklashforthecandidateorthepartythat

sponsoredthenegativepromotion(BrooksandMurov2012;Lauetal.2007):adherentsofthe

negativelyadvertisedcandidategetmotivatedtosupporthimevenmore.Inordertoexamine

thepotentialeffectofnegativecampaigning,weincludeabinaryvariablecapturingwhetherthe

candidatehasbeenavictimofnegativecampaigninginatleastoneadpublishedintheforefront

ofthegeneralelections.

DataandMethod

The2015generalelectionstotheNationalCouncilinSwitzerland

Switzerlandhasabicamerallegislatureandpopularelectionstakingplaceeveryfouryears,with

the26Swiss(half‐)cantonsformingthedistricts.WhileintheCouncilofStates,eachfullcanton

getstwoseatsandeachoftheformersixhalf‐cantonsreceivesone,theelectoralruleapplyingis

–withtwoexceptions–majoritarian.ThingsaredifferentintheNationalCouncil,wherethe200

seatsaredistributedaccordingtoacanton’spopulationsizeusingaproportionallistformula

(D’Hondt).AccordingtotheOSCE(2012:7),theSwisselectoralsystemis“unusuallycomplex,

[but]ithasthepositiveeffectofenhancingvoters’choice”.Itsfreelistsystemgiveseachvoteras

manyvotesasthereareseatsinhisorherdistrict(between1and35).Acitizencanerasenames

onpreprintedlists,voteformembersofdifferentpartiesorlists(panachage),givetwovotestoa

singlecandidate(cumulation)—andcandoallofthisatthesametime.Votingforalien

candidates(panachage)isquitepopular:inthe2015Swisselections,5’756’035votes(froma

totalof39’563’016votes)stemmedfrompanachage.Alreadyatthenationalelectionsin1975,

Niemetz(1977:299)detectedacertain“panachageenthusiasm”(Panaschierfreudigkeit),which

hetracedbacktoastrongerpersonalizationofthevotesratherthanthelooseningofpartyties.

Thus,theSwisssystemcombinesquiteuniquelytheadvantageofmajoritarianelectoralsystems

aimingatselectionofthemostqualifiedcandidatewiththeadvantageofproportional

representationaimingatpluralrepresentation.Thisgivesustheopportunitytotesttheimpact

ofcandidates’campaignactivitiesonboth,intra‐party(i.e.intra‐listcumulation)aswellasinter‐

partycompetition(i.e.inter‐listpanachage).Inthispaper,weundertakeanexploratory

approachandanalysecampaigneffectsonboth,“loyal”and“alien”preferencevotes.For

instance,onecouldassumethatindividualizedcampaignsincreasethevotesacandidate

receivesfromalienlistsbutnotnecessarilythevotesthatshereceivesfromvoterswhochose

herownpartylist.

Forourpurpose,wefocusonthe20cantonsthatdisposeofmorethanoneseatintheNational

Council,sincetheactofcumulationaswellaspanachagerequiresthepossibilitytovoteforat

leasttwocandidates.Intheselectedcantons,atotalof3788candidateswererunningon422

lists.1001outofthesecandidates(on220lists)placedatleastonecampaignadinoneormore

newspapersincludedinoursamplebetweenJanuary1standtheelectiondayonOctober18th.

Method

Inordertoexaminetheeffectofcampaignactivityonpersonalvotesacandidatereceives,we

ranmultilevelmodelswithrandomintercepts.UnlikeotherstudieswhichusedtheSwiss

panachagestatistics(BFS2016)atanaggregatelevel(Lutz2010),i.e.usedthecandidates’total

numberofpreferencevotescast,wefullyexploitthedata’spotentialbymodellingacandidate’s

electoralsuccessforeachballotlistinhisdistrict.Thisprocedurebearstheadvantagethatwe

areinabetterpositiontomodelthevotesofanindividualcandidateinrelationtothe

alternativesavoterwasofferedonotherpartylists.Hence,asadependentvariable,weemploy,

ontheonehand,thedegreetowhichacandidatewasabletoexploitthepanachagepotentialofa

specificlist:

Y , (1a)

wherepijisthenumberofpanachagevotesobtainedbycandidateionlistjandbdthenumberof

validballotpapersforalistd≠j.Sincethepanachagepotentialisheavilyskewedtowardsthe

right,wetakethelogofit.

Ontheotherhand,weusethedegreetowhichacandidatewasabletorealisehiscumulation

potentialasadependentvariable,whichcorrespondstotheaboveexcludedspecialcaseofd=j:

Y (2a)

Ourfocusonthelist‐specificachievementallowsustocontrolforsomecrucialcharacteristicsof

theoriginofapreferentialvote,i.e.aspecificvotersegment,andthustomodelthevotedecision

moreaccurately.Todoso,weneedtoaccountforthefactthateachlistj‐to‐listd‐combinationin

(1a)itselfisnestedinacross‐classificationofthecorrespondingparty‐to‐party‐combination

(pjpd)andtheelectoraldistrict(canton(c)).Hence,equation(1a)mustberewrittenas:

Y (1b)

Almostthesameappliestothespecialcase(2a):

Y (2b)

Sinceweassumethatunobservedfeaturesfromalltheselevelsaffectacandidate’sabilityto

exploithispersonalvotepotential,ourempiricalmodelsincludeseparatestandarderror

components.

Table1:Standarderrorcomponentsincludedintheregressionsof(1b)and(2b).

Panachagemodels Cumulationmodels

Cantons uc uc

Parties upjpd up

Lists ujd(pjpdc) uj(pjc)

Candidates uij(pjpdc)

Residual eijd(pjpdc) eij(pjc)

Takentogether,thisleadsustoacross‐classifiedrandomeffectsmodel(Snijders/Bosker1999)

whosecoefficientsareestimatedviarestrictedmaximumlikelihood(REML).

Besidethevariablesoftheassumedcampaigneffectsasdiscussedabove,weincludeseveral

additionalvariables.Atthecantonallevel,wecontrolfortheamountoflistspresentedtothe

voter,assumingthatahigheramountoflistsincreasesthechancesforavotertofindalistthat

matcheshispreferences,whichthuslowerstheamountofpersonalvotes.Furthermore,we

controlforthenumberofseats,ormandates,percanton(or,district).Furthermore,weinclude

twovariablesthatcapturewhetheracantonhas,comparedtothepreviouselectionanddueto

itsdemographicdevelopment,oneseatmoreorlesstooccupy.Althoughthesevariablesarenot

supposedtobedirectlyrelatedtothedependentvariable,theymightstronglyinfluencethe

degreeofcompetitionandthusthelevelofelectoralcampaigningwithinacanton.Wethus

includethesefactorsinordertoaccountforapotentialomittedvariablebias(seebelow).

Atthelevelofparties(or,party‐combinationsinthecaseofpanachage),weaccountforthe

overallvoteshareinthepreviouselections(2011)ofthepartythecandidatebelongsto,and–in

thecaseofthepanachagemodels–alsoforthe2011voteshareofthepartyreceivinga

(panachage)vote(=recipient).Incaseofthepanachagemodelswherecandidatesreceivevotes

fromoutsidetheirownlist,wealsoincludetheoverallvoteshareofthepartywhoselistisused

(donor)inordertoaddan“alien”candidate(whichgoeshandinhandwiththedonorparty

losingavote).Furthermore,weaddabinaryvariablethataccountsforthefactwhetherthe

donorandtherecipientlistbelongtothesamepartyornot.5

5InSwitzerland,partiesareallowedtorunonseverallistsandsublists(i.e.listswithfemalecandidatesandmalecandidatesonly,listswithyoungcandidates,etc.).

Attheleveloflists(or,list‐combinationsinthecaseofpanachage),wecontrolwhethera

candidatereceivingavote(=recipient)ispresentedonthemainorthejuniorlistofaparty

(comparedtootherlistsofaparty,suchasSwissabroad,greypanthers,etc.)aswellasthe

numberofmandateswonbythepartyin2011inordertomonitorthestrengthoftherecipient

listinagivencanton.Incaseofthepanachagemodels,weincludeexactlythesamevariablesfor

thedonorparty,i.e.thepartywhoselistwaschosenandwhereanexternalcandidatewasadded

to.Here,weassumethatcandidatesfrompowerfullistsarelikelytobeenlistedonother,less

powerfullists.Inaddition,wecontrolfortheorderoftherecipient’slist(assumingthatfront

listsreceivemoreattention)andwhetherthechosenlist(donor)disposesoffreespots,which

enhancesthepotentialforcumulationorpanachagesincenocandidateneedstobereplaced.

Furthermore,wecontrolforthefactthatpartiesareallowedtoallywithotherpartiesin

“combinedlists”(apparentments)andalsotocreate“sub‐lists,”e.g.foryoungorfemale

candidatesonly.Here,weassumethatvotersaremorelikelytoenclosecandidateswithina

combinedlist.

Atthecandidatelevel,wecontrolforalargesetofpotentialpersonalvoteearningattributes

(Shugartetal.2005).Onegroupofvariablesisintroducedtoaccountforthepoliticalexpertiseof

acandidate.Oneofthestrongestfactorsfoundinvirtuallyallstudiesonpreferencevotingis

incumbency(Carsonetal.2007;Cox&Katz1996;Gelman&King1990;Maddens&Put2013;

Put&Maddens2015),sinceincumbentsdisposeoftraceablepoliticalexperienceandemanate

familiarity.Furthermoreandforasimilarreason,weincludeabinaryvariablecapturing

whetherthecandidateisalsorunningfortheprestigioussecondchamberoftheCouncilof

States.However,politicalexperiencecanalsobegainedatthelocallevel.Thetypicalcareerofa

(Swiss)politicianbeginsatthelocalandendsatthenationallevel.EvenifthisOchsentour

(drudgery),asitiscalledinSwitzerland,isnolongeraswidespreadasitwasinthe20th

century,holdingapoliticalofficeatthelocallevelcanbeasignofpoliticalexpertise.

Additionally,alocalofficeimplieslocalrootsthatcouldattractvoterslookingforcandidates

whopromiserepresentingtheconstituency(Tavits2010).Therefore,wecontrolforwhethera

candidateholdsapoliticalofficeatthelocalorcantonallevel.Furthermore,localtiesmaybe

connectedtoconstituency‐orientationandthusbearhigherincentivesforvoterstoconsidera

localcandidate(Blaisetal.2003;Tavits2010),whichiswhywealsoconsiderwhethera

candidateisrunningforofficeinthesamecantonwherehishometownislocated.Lastbutnot

least,acandidate’spersonalcharacteristicsmayinfluenceitsvotingpotential,sincevotersmight

striveforenhancingdescriptiverepresentation(Mansbridge1999;BühlmannandSchädel

2012).Therefore,weincludecontrolsforgender,ageandprofession(farmers,workers,and

academics).6

Furthermore,wecontrolforadditionalfactorslocatedatthecandidatelevel:Severalstudies

highlighttheimportanceofballotposition(foranoverviewseeSpierings&Jacobs2014).The

positiononthepartylisthasproventobeanimportantdeterminantofelectoralsuccess,in

Switzerlandaswellaselsewhere(e.g.,Lutz2010;Tavits2010).Sincecampaigningisstrategic,

i.e.dependentontheexpectedchancesofelectoralsuccess,weneedtoidentifyfactorsthat

directlyaffecttheendogeneousregressor(campaigning)withoutdirectlyaffectingthe

dependentvariable,inordertoprecludetheproblemofomittedvariablebias(Moon2006).One

suchfactoriswhetheracandidateoccupiesawaiting‐listposition.Here,weexpectcandidates

thatholdthepositiononthelistthatwouldsignifyagaininseatforapartytohaveparticularly

strongincentivestoinvestinhiscampaignsincehischancestosucceedarerealbutnotcertain.

Asanexample,considertheSwissPopulistPartyinthecantonofZurichwhoheld11seats

duringthe2011/15legislature.Sinceinmostofthecantons,theorderofthecandidatesis

determinedbytheirpoliticalexperienceandincumbencystatus,thecandidateonthe12th

positionofthelisthasahigherincentivetogatherpersonalvotesthanthecandidatesplaced

beforeandafterhim.

Results

Wefirstpresentourmodelsofthepanachagepotential,i.e.determinantsofpreferentialvotes

receivedfromoutsideacandidate’sownpartylist.Turningtoourcontrolvariables(Model1,

Table2),resultsareasexpected:candidatesbelongingtolargerpartiesalsoexhibitahigher

panachagepotentialthancandidatesfrommoremarginalizedpartiesorlists.Furthermore,

closenesspaysout:candidatesaresubstantiallymorelikelytobeaddedtootherlistsofthe

samepartyor,tosomedegree,alsotoconnectedlists.Wealsofindtheassumedeffectofthe

ballotposition:thefurtherupacandidateisplacedinthelist,themorevotesshereceivedfrom

panachage.Furthermore,politicalexpertiseishighlyvaluedbyvoters.IncumbentMPsarefar

morelikelytoreceivepanachagevotesthantheirchallengers.Alsoholdingapoliticalofficeon

themunicipalityorthecantonallevelbearsacomparativeadvantageandsodoesrunning

simultaneouslyforthemoreprestigioussecondchamber,theCouncilofStates.Interestingly,

womenandyoungercandidateshaveasignificantlyhigherpanachagepotentialthantheirpeers,

presentingafirsthintthatthepanachagevotealsoservestoenhancedescriptiverepresentation.

6Inthesocalled„militiasystem“inSwitzerlandrepresentativesarenotprofessionalpoliticiansbuttheypursueaprofessionbesidetheirpoliticalmandate.Thus,mostofthecandidates(andevenincumbents)mentiontheiroccupation.Ofcourse,nobodycontrolsifthecandidatesreallypractisethementionedjobortowhatlevel.Thereareevenmentionedsuchthingsas“politician”or“artistoflive”.However,wearguethatthementionofanoccupationisanimportantsellingargumentforacandidate.

Likewise,farmersarefarmorelikelytobeaddedtoexistinglists,whilethecontraryappliesto

workers,althoughthisrelationshipisweakerandonlymarginallysignificant.Moreover,

candidatesrunningincantonswithahighernumberofseatsdisplayalowerpanachage

potential,leavinguswiththeassumptionthatthehigherthenumberofcandidatestobeelected,

thebetterthelistofcandidatesalreadyreflectsthevoters’willandthelesspanachageis

necessary.Lastbutnotleast,candidatesalsoreceivesignificantlymorepanachagevotesfrom

listswithemptylines.Apparently,voterstendtomaximizetheirvotingpotentialbyadding

additionalcandidatestoemptyspotsonthelist.

Table2:Determinantsofpanachagesuccess    Model 1 Model 2  Model 3

  Intercept  ‐5.449 (0.224)** ‐5.346 (0.23)**  ‐5.346 (0.229)**

Canton  No. of mandates  ‐0.051 (0.009)** ‐0.048 (0.008)**  ‐0.047 (0.008)**  Additional mandate  0.191 (0.187)  0.186 (0.186)  Mandate Reduction  ‐0.039 (0.149)  ‐0.031 (0.148)  No. of lists  ‐0.002 (0.012) ‐0.010 (0.012)  ‐0.010 (0.012)

Party  Recipient: Vote share 2011  0.007 (0.005) 0.002 (0.005)  0.002 (0.005)  Donor: Vote share 2011  ‐0.030 (0.005)** ‐0.030 (0.005)**  ‐0.030 (0.005)**  Same party  1.213 (0.184)** 1.198 (0.183)**  1.197 (0.183)**

List  Recipient: main  0.478 (0.041)** 0.410 (0.041)**  0.402 (0.041)**  Recipient: junior  ‐0.077 (0.044)° ‐0.057 (0.043)  ‐0.049 (0.043)  Recipient: position  ‐0.003 (0.002) ‐0.001 (0.002)  ‐0.001 (0.002)  Recipient: No. of mandates 2011  0.088 (0.010)** 0.077 (0.010)**  0.076 (0.010)**  Donor: main ‐0.591 (0.025)** ‐0.592 (0.025)**  ‐0.592 (0.025)**  Donor: junior 0.146 (0.025)** 0.146 (0.025)**  0.146 (0.025)**  Donor: No. of mandates 2011  ‐0.086 (0.008)** ‐0.086 (0.008)**  ‐0.086 (0.008)**  Donor: No. of empty lines on ballot 0.049 (0.002)** 0.049 (0.002)**  0.049 (0.002)**  Combined lists 0.237 (0.029)** 0.237 (0.029)**  0.236 (0.029)**

Candidate  Ballot position ‐0.018 (0.001)** ‐0.018 (0.001)**  ‐0.018 (0.001)**  Waiting‐list position  ‐0.179 (0.078)*  ‐0.170 (0.078)*  Incumbent  1.408 (0.042)** 1.240 (0.046)**  1.435 (0.054)**  Age  ‐0.002 (0.001)* ‐0.002 (0.001)**  ‐0.002 (0.001)**  Woman  0.106 (0.017)** 0.108 (0.017)**  0.109 (0.017)**  Runs for Council of States  0.653 (0.054)** 0.521 (0.054)**  0.537 (0.054)**  Holds political office on municipality level 0.157 (0.031)** 0.157 (0.030)**  0.157 (0.030)**  Holds political office on cantonal level 0.340 (0.035)** 0.299 (0.035)**  0.278 (0.035)**  Locally anchored  0.084 (0.017)** 0.079 (0.016)**  0.077 (0.016)**  Academic  0.026 (0.019) 0.030 (0.019)  0.029 (0.019)  Farmer  0.313 (0.043)** 0.346 (0.042)**  0.349 (0.042)**  Worker  ‐0.072 (0.037)° ‐0.070 (0.036)°  ‐0.068 (0.036)°

Campaign  No. of advertisements  0.012 (0.002)**  0.023 (0.002)**  Share of advertisements on front page 0.155 (0.062)*  0.146 (0.061)*  Share of individual advertisements 0.158 (0.052)**  0.141 (0.052)**  Share of advertisements in colour  0.156 (0.035)**  0.119 (0.036)**  Share of externally funded advertisements ‐0.022 (0.043)  ‐0.040 (0.043)  Share of testimonial advertisements 0.150 (0.089)°  0.123 (0.089)  Share of event advertisements  0.262 (0.107)*  0.231 (0.107)*  Negative campaigning  ‐0.117 (0.089)  ‐0.113 (0.088)

Interactions  (Incumbent)x(No. of advertisements)   ‐0.021 (0.003)**

Random effect  Cantonal‐Level 0.236 0.216   0.215SDs  Party‐Level  0.636 0.636   0.635  List‐Level  0.658 0.656   0.656  Candidate‐Level  0.454 0.443   0.441  Residual  0.608 0.608   0.608

Notes:REML‐estimatesfromtheregressionoflogpanachagesuccessondifferentmeasures.Nobservations=94174,NCandidates=3788,NList.‐combinations=9314,NParty‐combinations=336,NCantons=20;**99%;*95%;°90%.

Turningtoourcampaignvariables(Model2,Table2),wefindthatthemoreoftenacandidate

waspresentedonadvertisements,thehigherhispanachagepotential.7Alsotheadditional

campaignvariablesintroducedtocapturethevisibilityofacandidate’scampaign(shareof

colouredadsandtheshareofadspresentedonthefrontpageofanewspaper)provetobe

significant.Furthermore,anindividualizedstyleofcampaigningapparentlymakesacandidate

moreattractiveforvoterswhochoseanotherpartylistthanthecandidate’s:theshareof

externallyfundedcampaignadsissignificantlyandpositivelyrelatedtothepanachagepotential.

Theshareofcampaignadsdisplayingacandidate’snametogetherwiththeinvitationtoa

campaigneventgeneratesadditionalvoterpotential,too.However,displayinglocalrootedness

bylettingspeaktestimonials(i.e.usuallypotentialvoters)inone’scampaignissupposedtobe

onlymarginallysignificant.Furthermore,victimsofnegativecampaigningdonotnecessarily

havetofearnegativeeffects:althoughthepostulatedrelationtothepanachagepotentialis

negative,itisnotsignificant.

Inmodel3ofTable2,weinteractedtheincumbencystatuswiththenumberofcampaignadsin

ordertoexplorewhetherelectoralcampaigningpaysoutdifferentlyforchallengersand

incumbents.AsFigure1shows,thisisindeedthecase.Whileincumbentsdobarelyseemto

profitfrom(extensive)electoralcampaigning,challengerscan.Inthissense,ourresultsconcur

withevidencefromtheU.S.(e.g.,Abramowitz1991;Jacobson1978;Moon2006)whilethey

standincontrasttofindingsfromBelgium(MaddensandPut2013).Asdiscussedabove,

MaddensandPut(2013:853)tracetheirnon‐findingbacktothesizeoftheconstituency:in

largeconstituencies,itmightbemoredifficulttorecallallincumbentMPs,whileitiseasierin

smallerconstituenciesorinmajoritarianelectoralsystemswithsingle‐memberdistrictssuchas

theU.S.TheaverageSwissconstituencyis,with7.7seatsperunit,almosthalfasbigasthe

averageBelgiumconstituency(13.6).Byinteractingthechallenger’scampaigneffectwith

districtsize,futureresearchmightshedmorelightonthismatter.

7Ofcourse,ourmodelsruntheriskofoverestimatingthecampaigneffectsinceirrelevantchallengersareincludedaswell(seee.g.,Maddens&Put2013:853).Yetwealsoranmodelswhereweonlyincludedcandidateswhowereondisplayofatleastonecampaignad(Ncandidates=1001),assumingthatonlypotentiallyrelevantchallengersinvestedincampaignspending.However,resultsdidnotchangesubstantially(resultsavailableonrequest).

Figure1:Effectofcampaignintensityonelectoralsuccess(panachageexploitation)

Notes:ThisfigureisbasedontheinteractioneffectincludedinModel3,displayedinTable2.Campaignintensityiscapturedbyacandidate’stotalnumberofcampaignads.

Inthesecondsetofmodelsweexplorethecumulationpotentialofacandidate,i.e.thevoteshe

orshegetsfromhisownpartylist(Table3).Here,weareforemostinterestedinthedifferences

betweenintra‐andinter‐listsuccess.Comparedtothepanachagemodels(Table2)itstrikesthat

womenandyoungcandidatesarenotmorelikely(althoughalsonotlesslikely)tobecumulated

thantheirpeers.Thus,ourresultspresentafirsthintthatvotersmightstrivefordescriptive

representationviapanachage:seemingly,therearemanyvoterswhocompletetheirlistswith

womenandyoungcandidatesaimingatimprovingtheir(under‐)representationinthenational

parliament.Ontheotherhand,farmersare–again–notonlymorelikelytoreceiveadditional

votes,workersarealsosignificantlylesslikelytobeselected.Furthermore,beinganacademic

seemstoincreasetheelectoralsuccessintermsofvotesreceivedfromcumulation.Allinall,a

candidate’sprofessionorlevelofeducationthusseemstomattersomewhatmoreforintra‐list

competition.

Table3:Determinantsofcumulationsuccess    Model 1 Model 2  Model 3

  Intercept  ‐2.888 (0.306)** ‐2.636 (0.308)**  ‐2.635 (0.307)**

Canton  No. of mandates  ‐0.021 (0.013) ‐0.023 (0.012)°  ‐0.023 (0.012)°  Additional mandate  0.478 (0.269)°  0.475 (0.268)°  Mandate Reduction  ‐0.1 (0.213)  ‐0.096 (0.212)  No. of lists  ‐0.005 (0.006) 0.029 (0.017)  0.029 (0.017)

Party  Vote share 2011  0.208 (0.052)** ‐0.007 (0.006)  ‐0.008 (0.006)

List  Main  0.157 (0.051)** 0.168 (0.053)**  0.164 (0.053)**  Junior  0.010 (0.003)** 0.158 (0.051)**  0.162 (0.051)**  Position  0.045 (0.004)** 0.01 (0.003)**  0.010 (0.003)**  No. of empty lines on ballot  0.053 (0.016)** 0.044 (0.004)**  0.044 (0.004)**  No. of mandates 2011  ‐0.010 (0.001)** 0.05 (0.016)**  0.050 (0.016)**

Candidate  Ballot position ‐0.005 (0.006) ‐0.010 (0.001)**  ‐0.010 (0.001)**  Waiting‐list position  ‐0.138 (0.044)**  ‐0.135 (0.044)**  Incumbent  0.614 (0.024)** 0.530 (0.026)**  0.591 (0.031)**  Age  0.000 (0.000) ‐0.001 (0.000)  ‐0.001 (0.000)  Woman  0.011 (0.010) 0.012 (0.010)  0.012 (0.010)  Runs for Council of States  0.297 (0.031)** 0.228 (0.031)**  0.231 (0.031)**  Holds political office on municipality level 0.074 (0.018)** 0.075 (0.017)**  0.076 (0.017)**  Holds political office on cantonal level 0.179 (0.021)** 0.161 (0.021)**  0.155 (0.021)**  Locally anchored  0.031 (0.010)** 0.030 (0.009)**  0.029 (0.009)**  Academic  0.022 (0.011)° 0.024 (0.011)*  0.023 (0.011)*  Farmer  0.068 (0.024)** 0.088 (0.024)**  0.089 (0.024)**  Worker  ‐0.062 (0.021)** ‐0.060 (0.021)**  ‐0.060 (0.021)**

Campaign  No. of advertisements  0.005 (0.001)**  0.009 (0.002)**  Share of advertisements on front page 0.093 (0.042)*  0.089 (0.042)*  Share of individual advertisements 0.117 (0.033)**  0.109 (0.033)**  Share of advertisements in colour  0.003 (0.025)  ‐0.007 (0.025)  Share of externally funded advertisements 0.065 (0.026)*  0.057 (0.026)*  Share of testimonial advertisements 0.120 (0.064)°  0.101 (0.064)  Share of event advertisements  0.144 (0.066)*  0.134 (0.066)**  Negative campaigning  0.042 (0.055)  0.046 (0.055)

Interactions  (Incumbent)x(No. of advertisements)   ‐0.007 (0.002)**

Random effect  Cantonal‐Level 0.348 0.315   0.312SDs  Party‐Level  0.181 0.189   0.190  List‐Level  0.269 0.269   0.270  Residual  0.260 0.256   0.255

Notes:REML‐estimatesfromtheregressionoflogcumulationsuccessondifferentmeasures.NCandidates=3788,NLists=422,NParties=20,NCantons=20;**99%;*95%;°90%.

Intermsofcampaignadvertisements,wecannotdetectstrongdifferencesbetweenthetwosets

ofmodels,althoughingeneral,thecampaigneffectsappeartobeslightlylesscertain.Yet

particularlydisplayinglocaltiesbyinvitingvoterstopre‐electionpartiesorbyincluding

testimonialstopromoteone’svoteseemstoexertlessaneffectonloyalvotersthanonvoters

whochooseanotherpartylistthanthecandidatetheyaddedtothelist.Thesameholdstruefor

oneaspectofthevisibilityofacandidate’scampaign:theimpactofcolouredadvertisementsis

lessimportantforcumulationthanforpanachage.Finally,eveniftheimpactisnotsignificant,a

lookatthefindingsforthenegativeadvertisementsisnoticeablebecauseofthereversalofthe

sign.Whilenegativecampaigningtendstonegativelyinfluencepanachagesuccess,itratheris

rewardingforintra‐partychoice.Wetentativelyassumethatitisthemostextremecandidatesof

agivenlistthatarenegativelyportrayedbyotherparties.Itisthesecandidateshowever,that

mostattractstheownvoters.Ofcourse,theresultsarenotsignificantandthereisonlyveryfew

negativecampaigninginSwitzerland.However,giventhestabilityofallotherresultsthechange

ofsignisworthnoting.

Wemayonlyspeculateaboutthereasonsforthese–allinall–quitesmalldifferences.One

mightbethat,giventhe–onaverage–rathersmalldistrictsizeofSwisscantons,votersalready

knowthecandidatesoftheirpreferredpartylistwhilecandidatesfromotherlistsstillhavethe

possibilitytoattractanalienvoters’interestbymakingthemselvesknownvia(local)electoral

campaigning.Howeverandagainstourexpectations,individualizedcampaigningalsoenhances

thecumulationpotential.Apossiblereasonforthisfindingmightbethewayweoperationalized

ourvariable.Campaignadvertisementssponsoredbyassociationsmightnotnecessarily

underminetheintra‐listpopularityofacandidate,particularlynotiftheassociationstands

ideologicallyclosetothepartylistchosen.Here,amorefine‐grainedoperationalizationof

individualizedcampaigningmighthaveledtodifferentresults.

Discussion

Thispaperfollowsthetraditionofanalysisofelectoralperformanceinansweringthe

fundamentalquestionofwhogetselected.Theopenlist‐systeminSwitzerlandallowsforboth,

gatheringadditionalvotesfrominsideacandidate’sownlist(intra‐list‐votes;socalled

cumulation),aswellasadditionalvotesfromoutsidehisownlist(inter‐list‐votes;socalled

panachage).Ifweconsiderchoosingalistastheideologicalactofvoting,theanalysisofthe

additionalvotes(duetocumulationand/orpanachage)allowsustoexaminethevoting

preferencescontrollingforideologicalaffiliation.Inthissense,weaskwhichfactorsbesides

ideologyvotersexplicitlyorimplicitlyconsidertobeimportantwhenselectingtheircandidates.

Morespecifically,weareinterestedintheeffectsofelectoralcampaigningonthevoterpotential

andadoptanexploratoryapproachtodetectpotentialdifferencesbetweencampaigneffectsfor

cumulativeandpanachagevotes.

Besidesthepotentialtoenrichthediscussionontheconditionsofelectoralsuccessinopen‐list

proportionalsystems,thecomparisonhasamethodologicaladvantage.Otherthanprevious

studies,wefullyexploitthedata’spotentialbymodellingthevotesofanindividualcandidatein

relationtothealternativesavoterwasofferedonthesameaswellasonotherpartylists.

Thecombinationofouruniquedatasetconsistingofcampaignadvertisementspublishedin

morethan50importantnationalandregionalnewspapers(Bühlmannetal.2015)withthe

officialstatisticsofpanachageandcumulation(BFS2016)showthatcampaigningindeedisan

importantfactorforincreasingtheprobabilityofgettingapanachageoracumulationvote.Not

onlythenumberofadvertisementsbutalsotheircontent,i.e.whethertheyshowthecandidate

aloneorpromoteaneventtogettoknowthecandidate,significantlyincreasebothkindsof

additionalvotes.Thesefactorsthereforehelpacandidatestandoutagainstherpartycolleagues

aswellasagainstcandidatesfromotherlists.

Theseresultsareinaccordancewithpreviousstudieswhichhaveshownthatstronger

campaigningactivitiesincreasethefamiliaritywiththecandidates(Biersacketal.1993,

Bonneau2007,Jacobson1990).Thisalsoexplainsthehigherbenefitofcampaignspendingfor

challengerscomparedtoincumbents:theformerarelesswellknownandcanthusintroduce

themselvesviacampaigning.Campaigningthusallowsvoterstolookbeyondwhattheyalready

know,i.e.toconsideralienlistsandlessfamiliarcandidates.

Tosomeextent,however,ourresultsalsochallengehithertofindingsorourownexpectations.

Wehighlightthreepoints:

(1) Whiletheresultconcerningthedifferentimpactofcampaigningforincumbentsand

challengersconcurwithfindingsfromtheUSelections,theydeviatefromfindingsfrom

Belgium(Maddensetal.2006)orBrazil(Samuels2001).ThedifferencesbetweenBelgium/

BrazilandtheUScannotfullybeexplainedbythevotingsystem,giventhatBelgiumand

BrazilaswellasSwitzerlandarecharacterizedasopen‐listPRsystems.Sincethecountry

studiesdifferinseveralterms,suchasdistrictsizeoroperationalizationofourcampaign

variable,werefrainfromattemptstoexplainthesedifferencesatthispoint.Nevertheless,

theseresultsclaimforfurtherinvestigation,atthenationalaswellasattheinternational

level.

(2) Wedidnotdetectlargedifferencesofcampaigninfluencewhencomparingvotesreceived

frompanachage(inter‐listcompetition)totheonesreceivedfromcumulation(intra‐list

competition).Yetagain,itistooearlytodrawdefinitiveconclusions.Alternativemeasures

ofthecontentofacampaign,forexampleasregardsthedegreeofindividualized

campaigning,mighthaveledtodifferentresults.Furthermore,districtsizemightmakea

difference:whilewithincreasingdistrictsize,itbecomesmoredifficulttorememberall

candidatesfromachosenlist,itmightbecomeevenmoredifficulttoremember“alien”

candidates(whichmakesitcomparativelyeasierforthelattertomakethemselvesknown

viacampaigning).

(3) Yettheimportanceofhavingthechoicetoaddcandidatesfromtheselectedaswellas

candidatesfromalienlistswasdemonstratedforothernon‐campaignrelatedfactors.Most

interesting,youngorfemalecandidatesgetsignificantlymorepanachagevotesthanolder

ormalecandidates,whiletherearenosuchdifferencesconcerningcumulationvotes.Many

votersthusactivelysearchforyoungorfemalecandidatesonalienliststocompletethelist

oftheirfirst(ideological)choice.Weassumethatthiscanbeattributedtothe

underrepresentationofsuchcandidatesonmostmainlists.Thisfindingshowsone

advantageofthefullyopenlistsysteminSwitzerland.Whenvotersareallowedtochoose

amongcandidatesnotonlyfromtheirselectedlistbutamonganycandidatefromanylist,

theybenefitfromtheadvantagesofboth,thecandidatecentredmajoritarianaswellasthe

ideologycentredproportionalsystem.Suchasystemseeminglyallowsforboth,descriptive

aswellasideologicalrepresentation.

Ourcontributionsuggestsseveralavenuesforimprovement.First,whileweincludedafair

amountofimportantfactors,therestillaresomeaspectsthatarenotyetconsideredinour

models.Theseareadditionalideologicalfactors:itwouldbeinterestingtocontrolforthe

ideologicaldistancebetweentherecipientanddonorlists(althoughwealreadyslightly

approachthisbycontrollingforlistapparentments)aswellasforthemainissuesofanelection

campaign.Thiswouldofferapossibilitytocomparetheimpactofideologicalaswellasnon‐

ideologicalfactorsonpersonalvotesreceived.Second,itwouldbeworthwhiletoexplorethe

hypothesizedinfluenceofdistrictsizeontheimpactofelectoralcampaigns,asforexampleby

comparingthecampaigneffectsfortheelectionstotheFirstandtheSecondChamber–or

interactingcampaigneffectswithdistrictsize.Third,wecapturedelectoralcampaigningvia

advertisementsplacedinnationalandregionalnewspapers.AlthoughinSwitzerland,

newspapersofferstillanimportant,ifnotthemostimportant,arenaforelectoralcampaigning,

otherchannelsofcampaigninggaininimportance.Mostnotably,socialmedia(seee.g.,Spierings

&Jacobs2014)offerstheopportunityforengaginginanewtypeofcampaigningthatmaynot

onlybecharacterizedbylowercostsbutaswellbyahigherintensityofpoliticaladvertising.

Thus,moreresearchisneededtoinvestigatetheinterplaybetweenideology,candidatefactors

andcampaignstrategiesforvotechoiceandelectoralsuccess.Nevertheless,ourpapershows

thathavingtheopportunity,manyvoterspicktheircherriesnotonlyfromonebutfrom

differenttrees.

Literature

Abramowitz,AlanI.1991.„Incumbency,CampaignSpending,andtheDeclineofCompetitioninU.S.HouseElections“.TheJournalofPolitics53(1):34–56.

vanAelst,Peter,BartMaddens,JoNoppe,andStefaanFiers.2008.„PoliticiansintheNews:MediaorPartyLogic?MediaAttentionandElectoralSuccessintheBelgianElectionCampaignof2003“.EuropeanJournalofCommunication23(2):193–210.

Biersack,Robert,PaulS.Herrnson,andClydeWilcox.1993.„SeedsforSuccess:EarlyMoneyinCongressionalElections“.LegislativeStudiesQuarterly18:535–51.

Blais, André et al. 2003. „Does the Local CandidateMatter? Candidate Effects in theCanadianElection of 2000“. Canadian Journal of Political Science/Revue canadienne de sciencepolitique36(3):657–664.

Bonneau,ChrisW.2007.„TheEffectsofCampaignSpendinginStateSupremeCourtElections“.PoliticalResearchQuarterly60:489‐499.

Bowler, Shaun, David M Farrell, and Ian McAllister. 1996. „Constituency campaigning inparliamentary systemswith preferential voting: Is there a paradox?“Electoral Studies15(4):461–76.

Brooks,DeborahJordan,andMichaelMurov.2012.„AssessingAccountabilityinaPost‐CitizensUnited Era The Effects of Attack Ad Sponsorship by Unknown Independent Groups“.AmericanPoliticsResearch40(3):383–418.

Bühlmann, Marc, Marlène Gerber, Laura Salathe, and David Zumbach. 2015. PoliticalAdvertisements in the2015FederalElections [Database]. Bern: Année Politique Suisse,InstituteofPoliticalScience,UniversityofBern.

Bühlmann,Marc, and Lisa Schädel. 2012. „RepresentationMatters: The Impact of DescriptiveWomen’sRepresentationonthePoliticalInvolvementofWomen“.Representation48(1):101–14.

Carey,JohnM,andMatthewSobergShugart.1995.„Incentivestocultivateapersonalvote:Arankorderingofelectoralformulas“.ElectoralStudies14(4):417–39.

Carson,JamieL.,ErikJ.Engstrom,andJasonM.Roberts.2007.„CandidateQuality,thePersonalVote, and the Incumbency Advantage in Congress“. American Political Science Review101(2):289–301.

Cox,GaryW., and JonathanN.Katz. 1996. „WhyDid the IncumbencyAdvantage inU.S.HouseElectionsGrow?“AmericanJournalofPoliticalScience40(2):478–97.

Dowling, ConorM., andAmberWichowsky. 2015. „Attackswithout Consequence? Candidates,Parties, Groups, and the Changing Face of Negative Advertising“. American Journal ofPoliticalScience59(1):19–36.

Duverger,Maurice.1959.PoliticalParties:TheirOrganizationandActivity in theModernState.London:Methuen.

Elmelund‐Præstekær,Christian,andDavidNicolasHopmann.2012.„DoesTelevisionPersonaliseVotingBehaviour?Studying theEffectsofMediaExposureonVoting forCandidatesorParties“.ScandinavianPoliticalStudies35(2):117–40.

Farrell, David M. 2011. Electoral Systems: A Comparative Introduction. New York: PalgraveMacmillan.

Fridkin,KimL.,andPatrickKenney.2011.„VariabilityinCitizens’ReactionstoDifferentTypesofNegativeCampaigns“.AmericanJournalofPoliticalScience55(2):307–25.

Geer, John. 2006. In Defense of Negativity: Attack Ads in Presidential Campaigns. Chicago:UniversityofChicagoPress.

Geise, Stephanie, and Frank Brettschneider. 2010. „Die Wahrnehmung und Bewertung vonWahlplakaten: Ergebnisse einer Eyetracking‐Studie“. In Information–Wahrnehmung–Emotion. Politische Psychologie in derWahl‐ und Einstellungsforschung, hrsg. ThorstenFaas, Kai Arzheimer, und Sigrid Rossteutscher. Wiesbaden: VS Verlag fürSozialwissenschaften,71–95.

Gelman, Andrew, and Gary King. 1990. „Estimating Incumbency Advantage without Bias“.AmericanJournalofPoliticalScience34(4):1142–64.

Jacobson,GaryC.1978.„TheEffectsofCampaignSpendinginCongressionalElections“.AmericanPoliticalScienceReview72(2):469–491.

Jacobson,GaryC.1990.„TheEffectsofCampaignSpendinginHouseElections:NewEvidenceforOldArguments“.AmericanJournalofPoliticalScience34:334–62.

Karvonen,Lauri.2004.„PreferentialVoting:IncidenceandEffects“.InternationalPoliticalScienceReview25(2):203–26.

Katz, Richard S. 1986. „Intraparty Preference Voting“. In Electoral Laws and their PoliticalConsequences,eds.BernardGrofmanandArendLijphart.NewYork:AgathonPress,85–103.

———.1997.DemocracyandElections.NewYork:OxfordUniversityPress.

Lau,RichardR., Lee Sigelman, and IvyBrownRovner. 2007. „TheEffects ofNegativePoliticalCampaigns:AMeta‐AnalyticReassessment“.JournalofPolitics69(4):1176–1209.

Lutz,Georg.2010.„FirstCome,FirstServed:TheEffectofBallotPositiononElectoralSuccessinOpenBallotPrElections“.Representation46(2):167–81.

Maddens,Bart,undGert‐JanPut.2013.„OfficeEffectsandCampaignSpendinginaSemi‐OpenListPRSystem:TheBelgian/FlemishFederalandRegionalElections1999–2010“.ElectoralStudies32(4):852–63.

Maddens,Bart,BramWauters,JoNoppe,andStefaanFiers.2006.„EffectsofCampaignSpendinginanOpenListPRSystem:The2003LegislativeElections inFlanders/Belgium“.WestEuropeanPolitics29(1):161–68.

Mansbridge, Jane. 1999. „Should Blacks Represent Blacks and Women Represent Women? AContingent“Yes”“.TheJournalofPolitics61(3):628–657.

Marsh,Michael.2007.„CandidatesorParties?ObjectsofElectoralChoiceinIreland“.PartyPolitics13(4):500–527.

Moon,Woojin. 2006. „The Paradox of Less Effective Incumbent Spending: Theory and Tests“.BritishJournalofPoliticalScience36(4):705–21.

Moons,WesleyG.,DianeM.Mackie,andTeresaGarcia‐Marques.2009.„TheImpactofRepetition‐Induced Familiarity on Agreement with Weak and Strong Arguments“. Journal ofPersonalityandSocialPsychology96(1):32–44.

Moser,RobertG.,andEthanScheiner.2005.„StrategicTicketSplittingandthePersonalVoteinMixed‐MemberElectoralSystems“.LegislativeStudiesQuarterly30(2):259–76.

Niemetz, Heinz. 1977. „Die Wahl des schweizerischen Nationalrates vom 26. Oktober 1975“.ZeitschriftfürParlamentsfragen8(3):292–303.

OSCE. 2012. Federal Assembly Elections 23 October 2013: Election AssessmentMission Report.Warsaw:ODIHR,OfficeforDemocraticInsitutionsandHumanRights,OSCE.

Put,Gert‐Jan,undBartMaddens.2015.„TheEffectofMunicipalitySizeandLocalOfficeontheElectoral Success of Belgian/Flemish Election Candidates: A Multilevel Analysis“.GovernmentandOpposition50(4):607–628.

Samuels,David.2001.„Money,Elections,andDemocracyinBrazil“.LatinAmericanPoliticsandSociety43(2):27–48.

Sartori, Giovanni. 1976. Parties and Party Systems: Volume 1: A Framework for Analysis. CUPArchive.

Selb,Peter,andGeorgLutz.2015.„LoneFighters:IntrapartyCompetition,InterpartyCompetition,andCandidates’VoteSeekingEfforts inOpen‐BallotPRElections“.ElectoralStudies39:329–37.

Selects. 2012. Befragung der Wählerinnen und Wähler nach den Wahlen ‐ 2011 [Dataset].Lausanne:FORS.

Shugart,MatthewSøberg,MelodyEllisValdini, andKati Suominen. 2005. „Looking for Locals:Voter InformationDemandsandPersonalVote‐EarningAttributesofLegislatorsunderProportionalRepresentation“.AmericanJournalofPoliticalScience49(2):437–49.

Spierings, Niels, and Kristof Jacobs. 2014. „Getting Personal? The Impact of Social Media onPreferentialVoting“.PoliticalBehavior36(1):215–34.

Steenbergen, Marco R. 2010. „Decomposing the Vote: Individual, Communal, and CantonalSourcesofVotingBehaviorinSwitzerland“.SwissPoliticalScienceReview16(3):403–424.

Swindle, Stephen M. 2002. „The Supply and Demand of the Personal Vote TheoreticalConsiderations and Empirical Implications of Collective Electoral Incentives“. PartyPolitics8(3):279–300.

Tavits,Margit.2010.„EffectofLocalTiesOnElectoralSuccessandParliamentaryBehaviourTheCaseofEstonia“.PartyPolitics16(2):215–35.

Wauters, Bram, Karolien Weekers, and Bart Maddens. 2010. „Explaining the Number ofPreferentialVotesforWomeninanOpen‐ListPRSystem:AnInvestigationofthe2003FederalElectionsinFlanders(Belgium)“.ActaPolitica45(4):468–90.

Zajonc,RobertB.1968.„AttitudinalEffectsofMereExposure“.JournalofPersonalityandSocialPsychology9(2):1–27.

Zittel,Thomas,andThomasGschwend.2008.„IndividualisedConstituencyCampaignsinMixed‐Member Electoral Systems: Candidates in the 2005GermanElections“.WestEuropeanPolitics31(5):978–1003.