UCD CENTRE FOR ECONOMIC RESEARCH
WORKING PAPER SERIES
2012
Ballot Order Effects: An Analysis of Irish General Elections
John Regan, University College Dublin
WP12/16
April 2012
UCD SCHOOL OF ECONOMICS
UNIVERSITY COLLEGE DUBLIN
BELFIELD DUBLIN 4
BallotOrdereffects:AnanalysisofIrishGeneralElections
JohnRegan
UCDSchoolofEconomics
[email protected]
Keywords:Ballotordereffects,ProportionalRepresentation,FixedEffects,Irish
Elections,MalteseElections
May2011
Abstract:ThispaperpresentsevidenceofballotordereffectsinIrishGeneral
Elections,where candidatesarelistedinalphabeticalorder. Data relating to
electionsfrom1977to2011suggesttheeffectissignificantinastatisticalsense
and in magnitude. The nature of the Irish electoral system seesvoterscast
preferences for candidates, and as a result a greater level of information
regardingvotersbecomesavailable.Variousfixedeffectsareaddedtocontrolfor
constituencies,candidatesandpoliticalparties.
________________________
TheauthorgratefullyacknowledgesfinancialsupportfromtheIRCHSS.Iam
indebted to Paul Devereux for detailed comments and suggestionsandto
participantsoftheIRCHSSQSSConference2010.
INTRODUCTION
TheslimmarginofvictoryintheU.S.PresidentialelectionbetweenAlGoreand
George Bush Jr. inNovember 2000 focused media attention on ballot paper
designandwhetherbettervotingtechnologiesandmethodscouldbeadoptedto
improveaccuracy.ThesameistrueofIreland,wheretheCommission on
ElectronicVotingwasestablishedin2004todeterminewhetherelectronic
voting would improve accuracy. Large sums of money have been spent by
variousgovernmentstoimprovetheaccuracyofelections,throughelectronic
votingandvotingreform.Yet,thereremainsonesignificantproblem that
continuestobiaselectionresults‐theorderinwhichcandidatenamesappearon
aballotpaper.
Politicianshavealong‐heldbeliefthatballotorderingcaninfluencetheoutcome
ofanelection.Socialscientistshavestudiedthiseffectandmadeconclusions
aboutitsimpact,yetcourtsoflawhaverarelyactedtoremovesuchabias.
Extensiveacademicresearchhasexploredthecharacteristicsanddynamicsof
ballotordereffectsintheU.S.,findingthata“primacyeffect”(beinglistedfirst)
canhaveasignificantimpactonelectionoutcomes(Miller&Krosnick(1998),
Koppell&Steen(2004),Ho&Imai(2008)).Whenaballotpaperiscomplexor
whenmanycandidatesarecompetingforthesameposition,votersmaynotbe
wellinformedabouteachcandidateandcanthereforemakedecisionsthathave
beeninfluencedbytheballotpaperinfrontofthem.
IrishelectionsareconductedthroughProportionalRepresentationusingaSingle
TransferableVote(PR‐STV).PreviousresearchonIrishelectionshasaddressed
someaspectsofballotordereffects,buthasnotsoughttoanalysethefullextent
ofsuchaneffect(Robson&Walsh(1974)andBowler&Farrell(1991a)).
UnderthePR‐STVsystem,votersmayindicatetheirfirst,secondandsubsequent
preferencesamongthelistofcandidates.Thecandidatesontheballotpaperare
usually presentedin one column
1
. Only when an unusually large number of
candidates are declared will the candidates be presented in twocolumns
2
.

1
AsampleballotpapercanbefoundinAppendixG
2
ForadetailedanalysisofvotinginIreland,seeCoakleyandGallagher(2009).
Candidatesareplacedontheballotpaperinalphabeticalorderaccordingtotheir
surname, so a “primacy effect” would suggest a greater number of elected
politicians(TD's‐MembersoftheIrishParliament)wouldhavesurnamesnear
thebeginningofthealphabet.
The purpose of this paper is to assess whether ballot ordering influences
constituencyresults.Specifically,thispaperaskswhetheroccupyingthefirst
ballotpositionhasasignificantadvantage(inboththestatisticalandmagnitude
sense)overotherballotpositions.Thispaperwilladdtotheexistingliterature
onalphabeticalvotingbyexaminingamuchlargersampleofIrishelectionsthan
haspreviouslybeenundertaken.Usingthesummaryconstituencyresultsfrom
the11GeneralElectionsbetween1977and2011totheIrishParliament(Dáil
Éireann),thispaperwillanalysefirstpreferencevotesandsubsequent
preferences (transfer votes). The subsequent preferences of voters are an
importantbutoftenignoredpartofthePR‐STVsystem.
Voter apathy is a potential reason why ballot‐ordering effects exist. Voter
turnoutandadulteducationlevelshavechangedsignificantlyovertheperiod
coveredinthisdatasetandthiscouldimpactonthenumberofvotescastthat
wereinfluencedbyballotordering.Thispaperwillexaminewhetherballotorder
effects have changed over time in line with changes in voter turnout and
educationlevels.
Themulti‐seatnatureofconstituenciesinIrishelectionsoftenresultsinpolitical
partiesrunningmorethanonecandidatetomaximizethepoliticalparty'stotal
vote.Thispaperwillexamineintra‐partyandinter‐partyvotingbiastoassess
how“partyvoters”areinfluencedbyballotorder.Irelandisoneoftwocountries
to use PR‐STV for electionsto itsprimary legislative assembly. The Maltese
systemofPR‐STVisquitesimilartoIrelandandisusedinthispapertoprovide
supportiveevidencethatballotordereffectstranscendIrishelections.

THEORYOFBALLOTORDEREFFECTS
CognitiveCostsandBoundedRationality
Simon (1957) puts forward the idea that decision‐making by individuals is
limitedbythelevelofinformationavailabletothem,thecognitiveresources,and
theamountoftimeavailabletothemtomakeadecision.Hestates"boundedly
rationalagentsexperiencelimitsinformulatingandsolvingcomplexproblems
and in processing (receiving, storing, retrieving, transmitting) information".
Simonsuggeststhatindividualsuseheuristicmethodstohelpmakedecisions
quicklyratherthanusingastrictruleofoptimization.Examplesofheuristic
methodsincludea“ruleofthumb”,aneducatedguessorcommonsense.Theydo
thisbecauseofthecomplexityofthesituation,andtheirinabilitytoprocessand
compute the expected utility of every alternative action. Deliberation or
cognitivecostsmightbehighandtheremaybeeconomicactivitiesalsorequiring
considerationanddecision(Simon,1957).
Thetheoryofcognitivecostsisanalysedinanumberofexperimentswhere
peopleareaskedtochoosebetweenvariousalternatives.Studentswhoare
presentedwithmultiplechoiceexamsarebiasedtowardschoosinganswersnear
thetopofthelist(Cronbach,1950).Whenpeoplewereaskedtochoosebetween
foodordrinkstheyhadjusttastedtheywerebiasedtowardschoosingthefirst
onetheyhadtasted(Dean,1980).Simon(1957)suggestspeoplewill often
choosethefirstreasonableoption,especiallyiftheyconsidertheconsequences
ofsuchamistaketobesmall.
Inanelection,voterswillincursomenonzerocostwhilereadingandprocessing
theinformationpresentedtothemduringthecourseoftheelectioncampaign
andontheballotpaper.Whenvotersarefacedwithalistofelectioncandidates
theywillsearchtheirmemoriesto findreasonstovotefor eachcandidate
3
.
Votersthinklessandlessabouteachsubsequentalternative,andastheywork
throughthelistofcandidatestheirshort‐termmemorycanbecomecongested

3
RabinandSchrag(1999)suggestthatpeopleevaluatelistswithaconfirmatorybias.Itsuggests
thatpeopletendtonoticeandlookforinformationthatconfirmstheirexistingbeliefs,whilst
ignoringanythingthatcontradictsthosebeliefs.
with thoughts. Therefore, they may be more likely to generate favourable
thoughtsaboutcandidateslistedinitially,biasingthemtowardvotingforthem.
Basedonthiscognitivecostofvotingidea,itwouldbeexpectedthatvotersare
thinkinglessandlessabouteachalternativewhenconsideringwhotovotefor.If
voters are asked to rank candidates according to their preferences in an
alternativevoteortransferablevoteelectoralsystemthenitislikelythatthese
effectsareexacerbated.Asaresult,themagnitudeofballotordereffectscan
differ depending on the electoral system and whether the voter can rank
candidatesinorderofpreference.TheIrishelectoralsystemanalysedhereuses
aproportionalrepresentationsystemwherevotersrankcandidatestooutline
theirpreferences.
NameandAlphabeticalEffects
Nuttin(1985)andJohnson(1986)suggestpeopletendtohavepreferencesfor
namesthatshareinitialswiththeirownnameor“namelettereffect”.Nuttin
(1985)presentsexperimentalevidencewhichshowsthatlettersbelongingtoan
individual’sownfirstorfamilynamearepreferredaboveotherletters.Hooren
etal.(1990)findthatwhenparticipantswereaskedtochoosebetweenrandom
letters,theychosethegroupthatcontainedlettersfoundintheirownname.
Hoorenetal(1990)findthatlettersbelongingtotheparticipants’ownnames
werepreferredtoallotherletters.Byrne(1971)suggestspeoplehaveapositive
regardforpoliticalcandidateswhosharetheirowninitials,becausesimilarity
enhancesattraction.Zajonc(1968)notesthatthegreaterprevalenceofsuch
namesinthegeneralpublicresultsinanincreasedexposuretosuchcandidates,
andthisexposureenhancesliking.Pelham,etal.(2005)suggeststhatthiseffect
arisesoutof“implicitegotism”,wherebypeoplegravitatetowardsplaces,people
andsituationsthatreflectthemselves,includingperhapssimilaritieswiththeir
ownname.
Mackerras(1970),Trench(1987)andOrr(2002)notedthatpoliticalpartiesare
awareofballotordereffectsandwilloftenuseittomaximizetheparty'svoteby
choosingcandidateswhowouldtakeadvantageofapredeterminedordering
(alphabeticaletc.).Manygovernmentshaveusedrandomisationsoftheordering
ofnames,orrotationsintheprintingofballotpaperstocounteract ballot‐
orderingbias.TheCaliforniaAlphabetLottery(HoandImai,2008)isanexample
wherebotharandomisationandrotationareused.Thelettersofthealphabet
aredrawnatrandomtoformarandomizedordering,whichisusedinthefirst
electoraldistrict.Thisrandomisedorderingisthenrotatedthroughtheelectoral
districtsbymovingthecandidateatthetopofonedistricttothebottomofthe
nextdistrict.
The ordering of candidate names, whether alphabetic or randomised, may
inducebehaviouralchangesincandidates,makingitdifficulttoestimatethe
directeffectsofballotorder.Candidateswhodiscovertheyarenotplacedinan
optimalpositionafterarandomisationmaydecidetocanvassmoreaggressively
inthatdistrict.Thesameistrueofrotationswithintheelectoralarea.Depending
onthetimingofarandomisation,candidatescanstillbechosenorsubstitutedby
politicalpartiestoexploittherandomisationorthealphabet.
Somecountries,includingIreland,usealphabeticalorderingforsimplicityand
trytocounteractanycognitivecostsorballotordereffectsbyprovidingdetailed
informationonthecandidatesontheballotpaper.
InIreland,electioncandidatesmustusethenametheyarecommonlyknownby
whendeterminingtheorderingofnamesontheballotpaper,butdeedpollsmay
beusedasawayoflegallydeclaringanamechange
4
.Changingasurnametoits
equivalentintheIrishlanguagecanoftenrewardacandidate.Manysurnamesin
Irishbeginwith“Ó”and“Mac”andcandidatesmaysometimesdropthe“Ó”
altogether,orplaceitaftertheirfirstnameifdoingsowouldbeadvantageous,
examplesofthisinclude“Cuiv,ÉamonÓ”.Candidateshavealsousedmarriage
andmaidennamesifitmighthelpthemtomoveclosertothetopoftheballot
paper.
InO'ReillyversusMinisterforEnvironment(1986I.R.143)theIrishHighCourt
declaredthatalphabeticalorderonballotpaperswasconstitutionalandrejected

4
E.g.Independentcandidate“SeánD.ChristianDemocratDublinBayLoftus”priortothe1973
GeneralElection
theideathatitcreatedinequalitybetweencandidates.TheHighCourtfoundthat
thealphabeticalnatureoftheballotpapermadevotingasimpleraffairforthe
public. In an attempt to counteract these effects, political party logos and
candidatephotographswereaddedin1999and2000respectivelytoaidvoters
andreducethecognitivecostsassociatedwithvoting.

EMPIRICALEVIDENCEOFBALLOTORDEREFFECTS
Severalstudieshavefoundlargeandstatisticallysignificantballotordereffects.
These studies havefocusedlargely on the U.S., with some others looking at
Australia,MaltaandIreland.Researchcanbebrokendowndependingonthe
ordering method used on the ballot paper (Alphabetical, Rotation and
Randomisation).U.Sresearchtendstobeongoing,withnewpaperspublished
aftereachsetofstateandCongressionalelections.ResearchoutsidetheUSis
relativelyrare,withlittlesubstantiveresearchon‐goinginanyonecountry.
ResearchrelatingtoalphabeticalorderinghasfocusedonIrelandandMalta;
bothusealphabeticalvotingandproportionalrepresentationsystems.Robson
and Walsh (1974) examine the 1973 General Election in Ireland. They find
strongsupportthatalphabeticalfactorsareimportantinIrishelections;it
particularlyfavoursthefirstcandidateofapoliticalpartywheretheyrunmore
thanonecandidateinaconstituency.Bowler&Farrell(1991a)brieflyestimate
ballotpositioneffectsforthe1989IrishGeneralElection,findingsomeevidence
ofsucheffects.Ortega‐Villodres(2008)useslongitudinaldatatoresearchballot
orderinginMalta,findingthatballotorderingcanincreaseacandidate’svote
shareby8.22%vis‐à‐vistherestoftheparty'scandidates.Ortega‐Villodres
(2008)suggeststhatcompulsoryvotingandatwopartysystemmightbeits
causes.Ortega‐VillodresandDelaPuerta(2004)offerabriefanalysisofthe2002
IrishelectioninapapercomparingIrelandandMalta,bothofwhomusePR‐STV
astheelectoralsystem.Thispaperwilladdtotheexistingliteraturebyfully
examiningthePR‐STVelectoralsystem,usingtheconstituencyresults from
elevenIrishGeneralElectionsbetween1977and2011.
Rotatingnamesontheballotpaperbydistrictorwithinbundlesofballotpapers
offersasimplealternativetoalphabeticordering.“Robsonrotations”werefirst
introducedinTasmaniain1979bytheTasmanianMemberofParliamentNeil
Robsonandarenowusedinelectionsthroughouttheworld(HoandImai,2008).
Undera“Robsonrotation”nameswouldberotatedontheballotpaper,where
thenumberofrotationswouldbeequaltothenumberofcandidates,therefore
“distributing”theeffectsamongstallthecandidates.
RotationsarerathercommonacrosstheUS.MillerandKrosnick(1998)analyse
votereturnsfromthe1992Ohiostateelections.MillerandKrosnick(1998)
showedthatnameordereffectsincreasedacandidate'spercentageofvotesby
2.33percentcomparedtowhenlistedlast.Theyconcludethateventhoughthe
effecttheyfindisstatisticallysignificant,itisnotsubstantivelysignificantand
likelyhasnoimpactonelectoraloutcomes.KoppellandSteen(2004)disagree
withtheMillerandKrosnickanalysisofballotorderingeffects.Theyexamine
electionsinNewYorkCitywhererotationsarealsoused,findingthatpositional
biasrangesfrom2percentto4percentdependingontheseniorityofthepublic
office.
Despiteattemptstoremoveballotordereffects,theyremainpresent,evenwhen
thenamesarerandomised.HoandImai(2008),analysingthe2003Californian
Recallelections,findminorpartycandidatescanincreasetheirshareofthevotes
bybetween2and4percentagepointswhenlistedfirstontheballotpaper,with
noeffectsformajorpartycandidates.KingandLeigh(2009)examine the
randomised ballot order used in Australia where voting is compulsory.
Compulsoryvotingislikelyto exacerbate the ballot order effect(aswellas
increasingthenumberofinvalidballots)aspeoplewhotypicallydon'tvoteor
don'tcareaboutpoliticsareinsomeway“forced”tocomeoutandmarkthe
ballot paper
5
. They note that ballot order is orthogonal to all candidate
characteristicsduetotherandomizationofnames.Theyestimateastatistically
significanteffectof1%fortheaveragecandidate,withlargereffectsforsmaller
partiesandindependentcandidates.


5
Inthe2004electionthefineforfailingtovotewasAus.$20,orapproximatelytheaverage
hourlywage
DATA&METHODOLOGY
ThedatausedinthispaperhasbeensourcedfromtheFranchiseSectionofthe
DepartmentoftheEnvironment,Community&LocalGovernmentinIreland.It
coversthe457separateconstituencyelectionstoDáilÉireannthattookplace
duringthe11GeneralElectionsbetween1977and2011.
In total there are 4,807 election candidate observations and 2,249 different
individualscontestedtheseelectionsduringthisperiod,879runningmorethan
once.Informationregardingconstituencies,thenumberofseatstobewonin
eachconstituency,thenamesofcandidatesastheyappearedontheballotpaper
andthevotesreceivedbyeachcandidatearecontainedwithinthedataset.
Table 1 details some summary statistics regarding elections to Dáil Éireann
between1977and2011.Irishpoliticsisdominatedbythreepoliticalparties
(Fianna Fáil, Fine Gael and Labour), while other parties and independents
struggletohaveasignificantimpact.Onaverage,incumbentsreceive2.5times
thenumberofvotesachallengercanexpecttoreceive.
Table2contains informationregardingnumberof candidates,constituencies
andquotasizesinthe11electionscoveredinthedataset.Turnoutvariesover
the11elections,rangingfrom62%in2002to77%in1977.Thenumberof
candidateshaschangedsignificantlyovertime,rangingfrom364in1982to568
in 2011. This may also have influenced ballot order effects as greater
competitionmeanslongerballotpapers.Asaresult,controlshavebeenaddedto
accountforthenumberofcandidatesonaballotpaper.
Table1‐DescriptiveStatistics
Mean SD Min Max
No.ofCandidates 11.53 3.33 4 24
‐FiannaFáil (N=1291) 3.05 0.78 1 5
‐FineGael(N=1115) 2.69 0.8 1 5
‐Labour(N=517) 1.47 0.62 1 3
‐Independents(N=1025) 3.95 2.36 1 14
1stPreferenceVotes 4,112 3,176 13 20,079
‐Incumbents(N=1594) 7,046 2,370 1,096 20,079
‐Challengers(N=3213) 2,656 2,430 13 17,256
‐FiannaFáil 6,235 2,700 447 20,079
‐FineGael 5,450 2,578 549 17,472
‐Labour 4,434 2,951 183 17,256
‐Independents 1,223 1,967 13 17,075
Quota 8,827 1,352 5,859 13,864
TotalValidPoll 45,176 10,476 23,434 75,539
No.ofSeat s 4.13 0.8 3 5
TotalNumberofObservations‐4807
Table2‐DescriptiveSta tistics
Election No.ofSeats
No.of
Constituencies
Averag e
Quota
Numberof
Candida tes
Turnout
1977 148 42 8466 374 77.0%
1981 166 41 8327 402 76.5%
Fe b ‐82 166 41 8084 364 73.5%
Nov‐82 166 41 8185 364 72.5%
1987 166 41 8613 466 73.0%
1989 166 41 8018 370 68.0%
1992 166 41 8351 481 68.5%
1997 166 41 8551 483 66.0%
2002 166 42 8956 463 62.0%
2007 166 43 9865 470 67.0%
2011 166 43 10698 568 70.1%
Tocaptureballotordereffectsweusethevotesreceivedbyeachcandidateinthe
baselinespecification:
Votes
ijt
0
1
(BallotPosition
ijt
)
2
(Incumbency
ijt
)
Controls
ijt
Wherei,jandtindexcandidates,constituenciesandelections.
Threevariationsofthedependentvariableareused:thenumberoffirst
preferencevotes,theshareoftotalvotesintheconstituencyandLog(Shareof
Votes).
BallotPosition
ijt
isadummyvariableforeachballotposition.Thefourteenthand
subsequent ballot positions have been grouped together as frequency gets
smaller.
Incumbency
ijt
is a set of dummiesfor candidate incumbency, being a
sittingTD;beingaGovernmentMinister;beingaleaderofapoliticalpartyand
beingapartymemberoftheoutgoinggovernment.Theseareincludedtoproxy
forhowlargeaprofilethecandidatemayhaveintheirconstituency.
Thefollowingsetofcontrolsandfixedeffectsareusedthroughout: total
candidatesintheconstituency(tocontrolforthelengthoftheballotpaperand
for the level of competition), constituency fixed effects, political party fixed
effectsandelectionfixedeffects.
Asballotpapersusealphabeticordering,itispossiblethatacandidatewithan
“advantageous” surname might occupy the top ballot position in numerous
elections.Ifsuchacandidateweretobesuccessfulinanumberofelections,the
buildingupofaloyalsupportbasewouldbiasanyestimatesofballotorder.The
cumulativeimpactofballotorderingoversuccessiveelectionsmayleadtoa
serious distortion of the composition of Dáil Éireann. To control for this,
individualcandidatefixedeffectswillalsobeused.
Thereexiststhepossibilitythattheerrortermsarenon‐independentofeach
other(clustered)ascandidatesarerunningagainsteachotherataconstituency
level.Failuretoaccountforerrorclusteringleadstounder‐estimatedstandard
errorsandconsequentlytheover‐rejectionofhypothesistests.Thispaperuses
themulti‐wayclustering(Cameron,Gelbach,andMiller,2006)forestimating
robuststandarderrors.Standarderrorsareclusteredattheconstituencylevel
becauseballotordervariesatthatlevel,butalsoattheindividuallevelasthere
arerepeatedobservationsforindividualcandidatesthroughoutthedataset.
RESULTS
1stPreferenceVotes
Figure1presentsapreliminary examinationof theshareoffirstpreference
votesreceivedbyvariousballotpositions.Thegraphsshowtheaveragenumber
offirstpreferencevotescandidatesreceiveoverandabovetheexpectedvote
(theexpectedvotehereisthetotalvalidpolldividedbythenumber of
candidates),arrangedfromballotposition1to14+.Themarkeddifference
betweenthefirstfewballotpositionsandtheremainingballotpositionswould
suggestthereisasignificantadvantagetobelocatednearthetopoftheballot
paper.Asaresult,havingasurnamethatwouldpotentially“guarantee”sucha
positionwouldgiveasignificantelectoraladvantage.Thisbiasisclearlyvisible
in four and five seat constituencies but not as obvious in three seat
constituencies.
TablesIVandVpresentresultsofthebaselinespecification,usingShareofTotal
Votesasthedependentvariable.AppendixAcontainsresultsusingtheNumber
of1stPreferenceVotes andLog(ShareofVotes)asthedependentvariables.
TableIV&Vcontaintheresultsfromfiveregressions,oneOLS,threefixed‐
effectspanelmodelsandaLogitmodel.TableIVusesthefirstpositiononthe
ballotpaperasadummyvariabletocapturetheeffectofbeingatthetopofthe
ballotpaper.TableVusesthefirstpositionontheballotpaperasthebaseline
(excluded)categoryandsoestimatesforthesubsequentballotpositionsare
relativetobeingfirstontheballot.
TheOLSspecificationisprovidedtoallowcomparisonsbetweenOLSandfixed
effects. The OLS specification over estimates the effects of ballot ordering,
comparedtothefixedeffectsmodels.
The fixed effects estimates in Table IV suggest a positive and statistically
significantballotorderingeffectof1.16percentagepoints.Referringtothefixed
effectsmodelsinTableV,theeffectofbeingonasubsequentballotposition
rangesfrom‐0.8to‐1.79percentagepointscomparedtobeingfirstontheballot.
These numbers are rather large, considering the share of votes a candidate
requirestowinaseatrangesfrom16.66%to25%.Mostoftheestimatesare
significantatthe1percentlevel.Theseresultsarebroadlyin‐linewiththe
existingballotorderliterature.Figure2presentsagraphicalanalysisofthese
fixedeffectsregressionsandincludesthe95%confidenceintervals,usingthe
twodifferenterror‐clusteringmethods.
TheseresultsareechoedinthefixedeffectsregressionsinAppendixAandB.
ThefixedeffectsestimatesinAppendixAsuggestaballotorderingeffectof508
firstpreferencevotescomparedtootherballotpositions.Referringtothefixed
effectsmodelsinAppendixB,theeffectofbeingonasubsequentballotposition
rangesfrom393to783firstpreferencevotescomparedtobeingfirstonthe
ballot.Theseeffectsaresurprisinglylargegiventheaveragequotais8,827votes.
AppendixBalsopresentsestimatesthatbeingonasubsequentballotposition
decreasesacandidate’sshareofthevote,withtheeffectrangingfrom13.8%to
31.3%dependingtheballotposition.
Asballotpapersareorderedalphabetically,itispossiblethatacandidatewithan
“advantageous” surname might occupy the top ballot position in numerous
elections.Specification(D)inTablesIVandVincludesindividualcandidatefixed
effectstoaccountforsuch“advantageous”surnamesandpopularity. This
specificationisidentifiedonlyfromwithincandidatevariation,usingthe879
candidateswhoraninmorethanoneelection.Theresultsarebroadlysimilarto
thosefromtheotherfixedeffectsregressions.
AppendixCdividesconstituenciesupbythenumberofseatstobefilledinthe
election. The results for being first on the ballot paper suggest that ballot
orderingeffectsaretobefoundin4and5seatconstituenciesonly.
ThefinalspecificationinTableIVandVpresentsalogitmodelusingthewinning
ofaseatinDáilÉireannasthedependentvariable.Beingthefirstcandidateon
theballotpaperhasapositiveeffectonwinningaseat.Thisresultissignificant
at the 5% level. This has been explored further in Appendix C by dividing
constituenciesupbythenumberofseatstobewonateachelection.Thelogit
results in Appendix C suggest ballot order effects only matter in5seat
constituencies.
Table4
AB C D E
OLS FixedEffects FixedEffects FixedEffects FixedEffectsLogit
1STBALLOTPOSITION 0.0152*** 0.0116*** 0.0116*** 0.0117** 0.320**
(0.00293) (0.00275) (0.00432) (0.00576) (0.133)
INCUMBENTTD 0.0890*** 0.0679*** 0.067 9*** 0.0203*** 2.921***
(0.00172) (0.00205) (0.00308) (0.00347) (0.0872)
GOVERNMENTPARTY 0.00608*** ‐0.0192*** ‐0.0192*** ‐0.0110*** ‐0.00581
(0.00193) (0.00198) (0.00229) (0.00224) (0.0910)
GOVERNMENTMINISTER 0.0276*** 0.0364*** 0.0364*** 0.0168 ** 0.838***
(0.00521) (0.00569) (0.00706) (0.00708) (0.302)
POLITICALPARTYLEADER 0.0652*** 0.0781*** 0.078 1*** 0.0367** 1.869***
(0.0108) (0.0108) (0.0153) (0.0147) (0.62 8)
CONSTANT 0.176*** 0.155*** 0.155*** 0.238*** ‐1.751***
(0.00274) (0.0143) (0.0230) (0.0262) (0.0867 )
Observations 4807 4807 4 807 4 807 4807
R‐squared 0.542 0 .671 0.671 0.881 0.3148(Pseudo)
ColumnsB,CandEus eConstituency,Electi on,No.ofCandida tes andPoliticalPartyfix edeffects.
ColumnDadds
IndividualCa nd ida tefixedeffects toColumnC
ColumnsA,BandEclusterattheCons ti tuencyEl ecti onlevel.CandDcl usterattheconstituencyelecti onand
candi datelevel
RobustStandarderrorsinparentheses ,***P<0.01,**P<0.05,*P<0.1
Table5
AB C D E
OLS FixedEffects FixedEffects FixedEffects FixedEffectsLogit
POSITION2 ‐0.00819** ‐0.00812** ‐0.00812* ‐0.0120** ‐0.228
(0.0 0402) (0.00 366) (0.00482) (0.00587) (0.196)
POSITION3 ‐0.0138*** ‐0.0112*** ‐0.0112** ‐0.0107 ‐0.240
(0.0 0362) (0.00 337) (0.00464) (0.00713) (0.189)
POSITION4 ‐0.0144*** ‐0.0127*** ‐0.0127*** ‐0.0131* ‐0.252
(0.0 0383) (0.00 358) (0.00492) (0.00785) (0.191)
POSITION5 ‐0.0181*** ‐0.0142*** ‐0.0142*** ‐0.0165** ‐0.216
(0.0 0388) (0.00 359) (0.00480) (0.00837) (0.199)
POSITION6 ‐0.0169*** ‐0.0118*** ‐0.0118** ‐0.0197** ‐0.218
(0.0 0392) (0.00 372) (0.00520) (0.00912) (0.194)
POSITION7 ‐0.0176*** ‐0.0106*** ‐0.0106** ‐0.0164* ‐0.0759
(0.0 0382) (0.00 349) (0.00480) (0.00942) (0.199)
POSITION
8 ‐0.0158*** ‐0.0126*** ‐0.0126*** ‐0.0229** ‐0.189
(0.0 0375) (0.00 342) (0.00456) (0.00978) (0.201)
POSITION9 ‐0.0167*** ‐0.0119*** ‐0.0119*** ‐0.0233** ‐0.689***
(0.0 0387) (0.00 350) (0.00461) (0.010 4) (0.229)
POSITION10 ‐0.0153*** ‐0.0133*** ‐0.0133*** ‐0.0192* ‐0.385*
(0.0 0410) (0.00 375) (0.00471) (0.011 0) (0.226)
POSITION11 ‐0.0143*** ‐0.00780* ‐0.00780 ‐0.0116 ‐0.0251
(0.0 0467) (0.00 434) (0.00534) (0.011 4) (0.270)
POSITION12 ‐0.0186*** ‐0.0120*** ‐0.0120** ‐0.0233* ‐0.347
(0.0 0450) (0.00 391) (0.00511) (0.012 5) (0.263)
POSITION13 ‐0.0220*** ‐0.0179*** ‐0.0179*** ‐0.0273** ‐0.639**
(0.0 0484) (0.00 440) (0.00555) (0.013 0) (0.311)
POSITION14+ ‐0.0140*** ‐0.01000** ‐0.01000* ‐0.0153 ‐0.0928
(0.0 0459) (0.00 442) (0.00531) (0.013 8) (0.295)
INCUMBENTTD 0.0888*** 0.0679*** 0.0679*** 0.0202*** 2.495***
(0.00173) (0.00205) (0.00309) (0.00347) (0.0998)
GOVPA RTY 0.00595*** ‐0.0193*** ‐0.0193*** ‐0.0111*** ‐0.69 2***
(0.00193) (0.00197) (0.00229) (0.00223) (0.0930)
GOVMINISTER 0.0276*** 0.0362*** 0.0362*** 0.0168** 0.965***
(0.0 0521) (0.00 568) (0.00703) (0.00710) (0.3 06)
PARTYLEADER 0.0660*** 0.0785*** 0.0785*** 0.0369** 2.611***
(0.01 08) (0.01 08) (0.0154) (0.0149) (0.712)
CONSTANT 0.189*** 0.165*** 0.165*** 0.242*** ‐0.211
(0.00350) (0.0148) (0.0240) (0.0260) (0.248)
Obser vations 4807 4807 4807 4807 4807
R‐squared 0.544 0.6 72 0.672 0.882 0.3869(Pseudo)
ColumnsB,CandEuseConstituency,Electi on,No.ofCandidatesandPoliti c alPar tyfixedeffects
.
ColumnDaddsIndi vidual Ca nd idatefixedeffectstoColumnC
ColumnsA,BandEclusterattheCons ti tuencyEl ectionlevel.ColumnsCandDclusteratthe
cons ti tuencyel ecti onandcandi datelevel
RobustStandarderrors inparenth eses,***P<0.01,**P<0.05,*P<0.1
Figure2
TransferVotes
Ananalysisoftransfervotesisrequiredtofullyunderstandthe nature of
electionsinIreland.TransfersareanimportantpartofvotingunderthePR‐STV
system. When a candidate has been elected, or eliminated, theirvotesare
transferred according to the second and subsequent preferences marked on
individualballotpapers.Eliminatedcandidateshavealltheirvotestransferred,
while elected candidates have their surplus votes transferred (total votes
receivedoverandabovethequotatobedeemedelected).
Withtheelectionandeliminationofcandidates,thefirstcandidatecanchangeat
eachcount.Ifthepersonatthetopoftheballotpaperiselectedonthefirstcount
andhasasurplustodistribute,theninthesecondcountthepersonwhowas
second on the ballot paper becomes the first. The same idea applies to
candidateswhohavebeeneliminated.Arrangingthedatathiswayensuresthat
thereisalwaysacandidatedescribedasfirstineachcount.
Aspecificationsimilartothebaselinespecificationisapplied,howeverinsteadof
lookingatthetotalnumberofvotesacandidatehasineachcount,thechangein
totalvotesacandidatereceives(i.e.thetransfervotes)isusedasthedependent
variable. The number of votes being transferred is typically small, so the
regressionsusingrawvotesasthedependentvariablehavebeendroppedin
favourofshareofthetransfersandlog(ShareofTransfers).


isasetofdummyvariablesforeachcount
6
,addedtoaccountforthe
differentroundsoftransfervotes.Themodifiedmodelisasfollows:














TableVI presents thefixed effectsestimates.Being firstontheballotpaper
increasesacandidate’s shareofthetransfer votesby1.2percentagepoints.
However,thisnumberhidesasignificantportionofthevariationbetweenballot
positionsandthelowerpanelofTableVIshowssignificantvariationbetween
ballotpositions(relativetobeingfirst).Beingsecond,thirdorfourthonthe
ballotpaperhasastatisticallysignificant(negative)effectcomparedtobeing
first. Using Log(Share of Transfer Votes) as the dependent variable in
specification2producessimilarresults.Beingfirstcandidateontheballotpaper
increasesacandidate’sshareofthetransfersby6.4percent.Again,thesecond
panelofTableVIunmaskssignificantvariationinthisnumber.Beingsecondand
thirdontheballotpaperhasastatisticallysignificant(negative)effectcompared
tobeingfirst.

6
excludesthe1stcount
Table 6
Dependent Var Share of Votes Share of Votes LogShare LogShare
Clustering
Election Count
Election Count &
Candidate
Election Count
Election Count &
Candidate
1st POSITION
1.231*** 1.231** 0.0636*** 0.0636**
(0.430) (0.483) (0.0223) (0.0273)
INCUMBENT TD
3.288*** 3.288*** 0.327*** 0.327***
(0.273) (0.340) (0.0177) (0.0228)
GOV PARTY
-1.709*** -1.709*** -0.0943*** -0.0943***
(0.375) (0.370) (0.0263) (0.0282)
GOV MINISTER
-0.326 -0.326 0.00504 0.00504
(0.819) (0.733) (0.0455) (0.0473)
PARTY LEADER
6.481*** 6.481*** 0.541*** 0.541***
(1.634) (2.249) (0.0853) (0.152)
Constant
37.51*** 37.51*** -1.727*** -1.727***
(0.478) (0.737) (0.0802) (0.0940)
R-squared
0.235 0.235 0.237 0.237
Dependent Var Share of Votes Share of Votes LogShare LogShare
Clustering
Election Count
Election Count &
Candidate
Election Count
Election Count &
Candidate
POSITION 2
-1.196** -1.196* -0.0705*** -0.0705**
(0.567) (0.633) (0.0272) (0.0322)
POSITION 3
-2.109*** -2.109*** -0.102*** -0.102***
(0.538) (0.592) (0.0287) (0.0327)
POSITION 4
-1.250** -1.250** -0.0539* -0.0539
(0.522) (0.571) (0.0299) (0.0361)
POSITION 5
-0.508 -0.508 -0.00528 -0.00528
(0.540) (0.592) (0.0328) (0.0382)
POSITION 6
-0.624 -0.624 -0.0245 -0.0245
(0.528) (0.589) (0.0351) (0.0410)
POSITION 7
-0.925* -0.925 -0.0659* -0.0659
(0.539) (0.582) (0.0379) (0.0453)
POSITION 8
-1.100** -1.100* -0.0768* -0.0768
(0.549) (0.602) (0.0440) (0.0522)
POSITION 9
-0.548 -0.548 -0.0333 -0.0333
(0.587) (0.633) (0.0502) (0.0596)
POSITION 10
-1.329** -1.329* -0.0851 -0.0851
(0.608) (0.691) (0.0578) (0.0666)
POSITION 11
-1.431** -1.431** -0.161** -0.161**
(0.702) (0.716) (0.0708) (0.0790)
POSITION 12
-0.805 -0.805 -0.0517 -0.0517
(0.783) (0.847) (0.0883) (0.103)
POSITION 13
-0.918 -0.918 -0.0493 -0.0493
(0.964) (0.983) (0.118) (0.122)
POSITION 14+
-1.560** -1.560* -0.196** -0.196*
(0.714) (0.816) (0.0863) (0.106)
INCUMBENT TD
3.323*** 3.323*** 0.329*** 0.329***
(0.273) (0.339) (0.0177) (0.0226)
GOV PARTY
-1.697*** -1.697*** -0.0935*** -0.0935***
(0.375) (0.370) (0.0263) (0.0282)
GOV MINISTER
-0.398 -0.398 0.00116 0.00116
(0.818) (0.731) (0.0455) (0.0471)
PARTY LEADER
6.530*** 6.530*** 0.543*** 0.543***
(1.629) (2.185) (0.0850) (0.149)
Constant
38.92*** 38.92*** -1.653*** -1.653***
(0.564) (0.806) (0.0810) (0.0950)
R-squared 0.236 0.236 0.237 0.237
Number of Observations - 18,692
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
The following fixed effects are used throughout: Election Count, Total Candidates in that
count, Political Party, Election, Constituency & No. of Seats
TimeVariation
Voterturnoutvariessignificantlyoverthe11electionscoveredinthedata.If
ballotorderingeffectscomeaboutbecauseofvoterapathythenwewouldexpect
ballotordereffectshavechangedalongwithvoterturnoutoverthepastthirty
years.Ballotordereffectsmayalsocomeaboutbecauseofalackofinformation
aboutelections,votingorpolitics.Ifthisistrue,thenchangingeducationlevels
overthepastthirtyyearswouldsuggestballotordereffectswouldfluctuateover
time.
Totestwhetherthisisthecase,aninteracttermiscreatedbetweentheballot
positiondummyvariableandeachelectioninthesample.Theinteracted
coefficientsshowthemagnitudeofballotordereffectsinthatelection.















TheresultsfromthisspecificationarecontainedinAppendixD.Regardlessof
whichdependentvariableisused,thefirstpositionispositive(butnotalways
significant)inallbuttheFeb.1982and2011GeneralElections.Ballotordering
appearstohavehadlittleornoimpactonelectionsduringthe1980's.Thiswasa
timeofsignificantinstabilityinIrishpoliticsoverall,andthismayhaveresulted
intheeffectbeingnon‐existentormoredifficulttoquantify.
Totestthehypothesisthatballotpositioneffectsdonotchangeovertime,anF
testisconductedtotestwhetherthe11coefficientsareequal.TheFtestsfailto
reject the hypothesis that the first position effect is the same in all eleven
elections.

InterandIntraPoliticalPartyeffects
The prominence of the large political parties in Irish elections cannot be
underestimated.Votersarebetterinformedaboutlargerpartycandidatesthan
smallerpartycandidates.Onereasonwhyballotordereffectsmightdifferisthe
highvisibilityoflargepoliticalpartiesduringelectioncampaigns;theydistribute
moreliterature,spendmoremoneyandreceivemoremediacoverage.
Politicalpartiescanbedescribedas“mainstream”and“marginal”parties,each
havingdifferenttypesofvotersandpoliticalviews.Irishpoliticsisdominatedby
threelargepoliticalparties,FiannaFáil,FineGaelandtheLabourParty.Their
votersareoftendescribedas“partyvoters”.“Partyvoters”are indifferent
betweencandidatesoftheirfavouredpoliticalparty,and socouldbe biased
towardsvotingforthefirstpartycandidateontheballotpaperifballotordering
effectsexistbetweencandidates.
Toformallytestwhetherballotordereffectsdifferforeachpoliticalpartyan
interactiontermiscreatedbetweentheballotpositiondummyvariableandeach
politicalparty.Theinteractedcoefficientsshowthemagnitudeofballotorder
effectsforcandidatesofthatpoliticalparty.Thesmallerpoliticalpartieshave
beengroupedtogetherforsimplicity.AnFtestisalsoundertaken,whichtests
whether the political party coefficients are equal. The results of these
regressionsarecontainedinAppendixEusingthesamedependentvariablesas
before.Smallerpoliticalpartieshavebeengroupedtogether
7
.














Regardlessofwhichdependentvariableisused,theresultssuggestthatballot
ordereffectsvarygreatlybetweenpoliticalparties.Theeffectsarepositive(but
notalwayssignificant)forallofthepartygroupings,withtheexceptionofthe
LabourParty.ThedifferencebetweenFiannaFáilandFineGaelcandidatesis

7
ThesmallerpoliticalpartiesincludetheGreenParty,ProgressiveDemocrats,DemocraticLeft,SocialistParty,
SocialistWorkersParty,Workersparty,SinnFein,FathersRightsandResponsibilities,H‐BlockCandidatesandthe
ChristianSolidarityParty
striking,despitebothpoliticalpartiesbeingsimilarinideology.Thiscouldbedue
to candidate selection procedures within the two parties, with Fianna Fail
selecting candidates to take advantage of ballot ordering. Independent
candidatesmayalso be influenced by ballot ordering.The F testsrejectthe
hypothesisthatballotorderingeffectsarethesameforeachofthepolitical
parties.
Thelargerpoliticalpartiesoftenaimtowinatleastoneseatinaconstituency(or
twoinalargerconstituency)andmayrunmorethanonecandidateeachto
achievethis goal.Itislikelythereforethataballotorderingeffectwillexist
betweencandidatesofthesamepoliticalparty.Thereisoftenmorethanone
independentcandidaterunninginaconstituency.Totestthisintra‐partyballot
orderinghypothesisthesampleisrestrictedtothetwomainpoliticalparties,
FiannaFáilandFineGael,andIndependentcandidatestodeterminewhether
ballotorderingexistswithinpoliticalpartiesandgroupsofindependents.
Regressionsforthethreesub‐samplesareperformedinasimilarmethodtothe
originalmodel.Errorsareclusteredattheindividualcandidateandconstituency
level.Indoingso,theresultingestimatesarerelativetoothercandidatesinthat
politicalpartyorindependentgrouping,ratherthanallcandidates.Theresults
arecontainedinAppendixF.Theregressionsconfirmthatintrapartyballot
orderingexistsforFiannaFailcandidates,butnotforFineGaelorindependent
candidates.BeingthefirstFiannaFáilcandidateontheballotpapercanincrease
acandidatesvoteby5801stpreferencevotes(or11.5%)abovethatoftheother
FiannaFáilcandidates.

PRSTVinMalta
InordertotestwhethertheseeffectsareuniquetoIrishelections,ananalysisof
Malteseelectionsisundertakenbelow.Thispaperwillattempttoexploitthefact
thatbothcountrieshaveusedasimilarelectoralsysteminordertoestablish
whethertheseeffectstranscendIreland.
TheMaltesePR‐STVruleswerebroadlysimilartoIrelandpriorto1971.From
1921to1971,allcandidateswerelistedontheballotalphabetically,regardless
oftheirparty.Since1976,theyhavebeengroupedbypartyandlisted
alphabeticallywithintheirpartygroup.Forthepurposesofthis paper, only
thoseelectionsfrom1921to1971havebeenexamined
8
.Ballotordereffectsfor
these elections have been analysed by Ortega‐Villodres (2008), finding
significanteffectswhicharethenattributedtocompulsoryvotingandatwo‐
partysystem.
TableVIIdetailssummarystatisticsforMalteseelections.Thesummarystatistics
reflectthefactthatthepopulationofMaltaismuchsmallerthanIreland.The
averagefirstpreferencevoteissignificantlysmallerinMaltathaninIreland(642
and 4,137 respectively). This is also reflected in the quota size in the two
countries(2,132votesforMaltaand8,583forIreland).
Table7‐SummaryStat istics
Mean SD Min Max
No.ofCandidate s (N=1,889) 21.05 8 7 38
1stPreferenceVotes 642 766 1 6,137
‐Incumbents(N=491) 1,291 1,011 14 6,137
‐Challengers(N=1,398) 414 484 1 3,786
Quota 2,132 6,767 372 3,182
ConstituencyTotalpoll 12,792 4,431 1,857 19,357
No.ofSeatsintheConstituency 5 0.465 4 6
Source:DeclarationsofResultspublished byElectoralCommission
ThemethodologyusedtoanalyseIrishelectionsisusedheretoallowforeasy
comparisonsbetweenIrishandMalteseelections.Thefollowingsetofcontrols
andfixedeffectshavealsobeenadded:totalcandidatesintheconstituency,no.

8
ThisdatahasbeensourcedfromtheDeclarationsofResultsofPollpublishedbytheElectoral
CommissionintheGovernmentGazette
ofseatsintheconstituency,constituencyfixedeffectsandelectionfixedeffects.
Multi‐wayclusteringhasbeenusedinthesamewayasbefore,clusteringby
individualelectionlevelandattheindividualcandidatelevel.Theresultsofthe
regressionusing1stpreferencevotesandtheshareofthe1stpreferencevotes
asthedependentvariablesarecontainedbelow.
BasedontheresultsinTable8,asignificantballotordereffect appears in
MalteseGeneralElections.Theseeffectsaresurprisinglylargegiventheaverage
quotaisjust2,132votes.UsingLog(Shareof1stPreferenceVotes) as the
dependentvariablesuggeststheeffectofbeingonasubsequentballotposition
decreasesacandidate’sshareofthevotebybetween24%and63%depending
theballotposition.
Theseeffectsaremuchlargerinmagnitudethan theeffectspresentinIrish
elections.Ortega‐Villodres(2008)suggeststhisispotentiallyduetocompulsory
votinginMaltaoralargelytwopartysystem(MaltaLabourParty and the
Nationalist Party). Ireland has a largely two party system also, but no
compulsory voting. Given the compulsory nature of voting in Malta andthe
cognitivecostsassociatedwithvoting,itislikelythatballotordereffectsare
exacerbated.
Table8
Coeff SE(A) SE(B) Coeff SE(A) SE(B)
Positi on2 ‐229.5 (76.29)*** (72.96)*** Posi ti on2 ‐0.441 (0.136)*** (0.136)***
Positi on3 ‐160.1 (76.75)** (71.37)** Posi ti on3 ‐0.296 (0.121)** (0.129)**
Positi on4 ‐193.2 (75.45)** (84.09)** Posi ti on4 ‐0.366 (0.128)*** (0.134)***
Positi on5 ‐193 (89.51)** (102.3)* Posi ti on5 ‐0.423 (0.136)*** (0.144)***
Positi on6 ‐153.8 (81.66)* (87.02)* Posi ti on6 ‐0.243 (0.139)* (0.146)*
Positi on7 ‐257.5 (67.95)*** (74.15)*** Posi ti on7 ‐0.558 (0.149)*** (0.152)***
Positi on8 ‐158.6 (73.20)** (82.75)* Posi tion8 ‐0.474 (0.147)*** (0.155)***
Positi on9 ‐208 (71.95)*** (81.60)** Posi tion9 ‐0.498 (0.145)*** (0.153)***
Positi on10 ‐114.3 ‐84.25 ‐102.5 Position10 ‐0.462 (0.156)*** (0.166)***
Positi on11 ‐
68.09 ‐99.34 ‐124.5 Posi ti on11 ‐0.448 (0.158)*** (0.165)***
Positi on12 ‐174.6 (90.85)* (99.73)* Position12 ‐0.544 (0.179)*** (0.183)***
Positi on13 ‐223.8 (107.7)** ‐137.1 Posi ti on13 ‐0.637 (0.201)*** (0.216)***
Positi on14+ ‐185 (61.05)*** (85.30)** Pos i ti on14+ ‐0.635 (0.116)*** (0.143)***
Incumbent 876.8 (54.81)*** (114.6)*** I ncumbent 1.38 (0.0667)*** (0.0982)***
Cons tant 382.2 (64.89)*** (76.76)*** Constant ‐2.985 (0.160)*** (0.173)***
RSquared 0.359 0.359 RSquared 0.326 0.326
Numberofobs ervations:1889,Robuststandarderrorsinpare nthe ses , ***p<0.01,**p<0.05,*p<0.1
SE(A)me ansStandErrorsareclusteredatconstituencylevel
SE(A)me ansStandErrorsuse multi‐wa yclusteringatconstitue ncyandIDCodelevels
Thefollowingfixedeffects areuse d:Ele ctionCount,Tota lCandidat e s,Election,
Consti tuency&No.ofSeats
ShareofVotes Log(Share ofVotes)
Conclusions
Thispaperdemonstratesthatalphabeticorderingeffectsexistinthe1977‐2011
Irish general elections. The effect is significant, in both a statistical and
substantivesense.Theestimatedeffectofbeinglistedfirstonanalphabetical
ballotpaperinanIrishgeneralelectionisapproximately544firstpreference
votes or 1.27 percentage points for the average candidate. This estimate is
broadlyin‐linewith RobsonandWalsh(1974)who foundtheeffectslightly
higher,rangingfrom784to968firstpreferencevotesduringthe1973General
Election.Evidenceisalsofoundthatsubsequentpreferencesarebiasedbythe
alphabeticalballotpaper.Thispaperalsosuggeststhatballotordereffectsexist
attheintraandinterpartylevel.
In general, the estimates of ballot order effects found in Irish elections are
smallerthanthosefoundintheUnitedStatesbyMillerandKrosnick(1998)(2.3
percentagepoints)andHoImai(2008)(2to4percentagepoints).Theestimates
aresimilartothosefoundinAustraliabyKingandLeigh(2009)(1percentage
point)andinMaltabyVillodres(2008)(0.74percentagepoints).
Ballot order effects of the magnitudes outlined in this paper could have
potentiallyserious implicationsfortheIrishpoliticalsystem.Theideaofan
incumbencyadvantageinasubsequentelectionhasbeenwelldocumentedand
researched(Lee,2008).Thisadvantagehasbeenexplainedbyanumber of
factors,suchasabilitytofundraise,accesstohiredhelporlargermediaprofile.
Ballotordereffectsmayalsomatteringettingachallengerelectedinitially,and
thusanincumbencyadvantagemaybeconferreduponthiscandidate.Additional
researchisrequiredtoinvestigatetherelationshipbetweenballotorderingand
theincumbencyadvantage.
Beyondthe direct effectsof ballot ordering, significant indirecteffects could
potentiallyexist.Thenumberofvotesacandidatereceivesisofteninterpretedas
themandateofthatpoliticianortheideastheysupport.Ballotorderingeffects
mightsuggestthatthismandateortheideasofthispoliticianmaygetgreater
attentionatanationallevelastheir“mandate”isincreased.
Theresultsofthispaperalsohaveimplicationsforthedemocracy of the
electoralsystem.Thepresenceofballotordereffectscouldbetakentosuggest
thattheelectionoutcomesarenotthetruewillofthepeople.Ifthisisthecase,
voterinformationandvotereducationcampaignsarerequired.Electionsmust
beconductedinawaythatisseenasfair,democraticandunderstandable.Ifthe
publicareinfluencedortheirpreferencesaredistortedbytheelectoralsystem,
thentheelectoralsystemitselfgetscalledintoquestion.Theseeffectshavebeen
thesubjectofmanylegalproceedingsandseveralfamouscourtcasesexistinthe
U.S.surroundingtheGorev.BushPresidentialelectionin2000. In O'Reilly
versusMinisterforEnvironment(1986‐I.R.143)theIrishHighCourtdeclared
thatalphabeticalorderonballotpaperswasconstitutionalandrejectedtheidea
thatitcreatedinequalitybetweencandidates.TheHighCourtfoundthatthe
alphabetical nature of the ballot paper made voting a simpler affair for the
public.
Thepresenceofastatisticallysignificantballotordereffectshouldbeofconcern
to policy makers. Rational voter models suggest voters will reward those
candidateswhoreflecttheirownpoliticalviewsandhaveperformedwellinthe
past.Rationalvotersshouldnotbeinfluencedbythedesignoftheballotpaper.
Redesigningoftheballotpaperfromalphabeticalorderingwouldthereforebea
desirablepolicyrecommendation.
However,aredesigningoftheballotpaperwouldatbestreducetheordering
effects,butnoteliminatethem.Asimplerandomizationofthenamesbeforethe
ballotpaperswereprintedwouldnotremoveballotordereffects,butrather
continuetobenefitthecandidatewhowasplacedatthetopoftheballotpaper.
Alternatively,a“Robsonrotation”systemcouldbeintroducedwherenamesare
rotatedontheballotpaper,distributingtheeffectsamongstallcandidates.
References
1. Bagley,C.R,"DoesCandidates’PositionontheBallotPaperInfluence
Voters’ Choice? A Study of the 1959 and 1964 British General
Elections.,"ParliamentaryAffairs,19(1966),162‐174.
2. Bain, Henry M., and Donald S. Hecock, Ballot Position and Voter’s
Choice(Detroit,MI:WayneStateUniversity,1957).
3. Bowler,ShaunandDavidFarrell.1991a.‘PartyLoyaltiesinComplex
Settings:STVandPartyIdentification’,PoliticalStudies.39:350‐62.
4. Bowler,ShaunandDavidFarrell.1991b.‘VoterBehaviourunderSTV‐
PR:SolvingthePuzzleoftheIrishPartySystem’,PoliticalBehaviour.
13: 303‐30. Byrne, Donn, The Attraction Paradigm(NewYork:
AcademicPress,1971).
5. Campbell,Angus,PhilipE.Converse,WarrenE.Miller,andDonaldE.
Stokes,TheAmericanVoter(NewYork:Wiley,1960).
6. Coakley,J.andGallagher,M.(2009).PoliticsintheRepublicofIreland,
5thed.RoutledgePSAIPress,London.
7. Cronbach,LeeJ.1950.“FurtherEvidenceonResponseSetsandTest
Design.”EducationalandPsychologicalMeasurement,10,pp.3‐31
8. Dean,MichaelL.1980.“PresentationOrderEffectsinProductTaste
Tests”.JournalofPsychology,105:107‐110
9. Geys,B.&Heyndels,B,"BallotLayoutEffectsinthe1995Electionsof
theBrussels'Government,"PublicChoice,116(2003).
10. Gold,David,"Anoteonthe“Rationality”ofAnthropologistsinVoting
forOfficers,"AmericanSociologicalReview,17(1952),99‐101
11. Graber,Doris,"TheMassMediaandElectionCampaignsintheUnited
StatesofAmerica,"inFrederickJ.Fletcher,ed.,
Media, Elections and
Democracy(Toronto:DundurnPress,1991),pp.139–177.
12. Ho,DanielE.,andKosukeImai.,"EstimatingtheCausalEffectsofBallot
OrderfromaRandomisedNaturalExperiment:theCaliforniaAlphabet
Lottery,1978‐2002,"PublicOpinionQuarterly,72(2008),216‐240
13. Hoorens, V., Nuttin, J. M., Herman, I. E., & Pavakanun, U. (1990).
Masterypleasureversusmereownership:Aquasi‐experimentalcross‐
culturalandcrossalphabeticaltestofthenamelettereffect.European
JournalofSocialPsychology,20(3),181‐205
14. Jacobson,Gary,"TheEffectsofCampaignSpendinginCongressional
Elections: New Evidence for Old Arguments.," American Journal of
PoliticalScience,34(1990),334‐362.
15. Jacobson, Gary, The Politics of Congressional Elections(Boston:
Longman,2004).
16. Johnson,MitziS.1986.“TheInitialLetterEffect:EgoAttachmentor
MereExposure?”PhD.dissertation,OhioStateUniversity.
17. King,A.andAndrewLeigh,"AreBallotOrderEffectsHeterogeneous?,"
SocialScienceQuarterly,90(2009).
18. Koppell,JonathanandJenniferA.Steen,"TheEffectsofBallotPosition
onElectionOutcomes,"JournalofPolitics,66(2004),267–281.
19. Lee, David S., 2008. "Randomized experiments from non‐random
selectioninU.S.Houseelections,"JournalofEconometrics,Elsevier,
vol.142(2),pages675‐697,February.
20. Lijphart,Arend,andRafaelLopezPintor,"AlphabeticBiasinPartisan
Elections:PatternsofVotingfortheSpanishSenate,1982and1986,"
ElectoralStudies,7(1988),225‐231
21. Mann,ThomasE.,andRaymondE.Wolfinger,"CandidatesandParties
in Congressional Elections," American Political Science Review, 74
(1980),617–632.
22. Miller,JoanneM.,andJonA.Krosnick,"TheImpactofCandidateName
Order on Election Outcomes," Public Opinion Quarterly, 62 (1998),
291–330.
23.
Miller, Warren E., and J. Merrill Shanks, The New American Voter
(Cambridge,MA:HarvardUniversityPress,1996).
24. Nuttin,JozefM,"NarcissismbeyondGestaltandAwareness:TheName
LetterEffect,"EuropeanJournalofSocialPsychology,15(1985),353–
361.
25. Orr, Graeme. 2002. “Ballot Order: Donkey Voting in Australia.”
ElectionLawJournal,1:4,pp.573‐78
26. OrtegaVillodres,Carmen.2008."GenderandPartyDuopolyinaSmall
State:BallotPositionEffectsundertheSingleTransferableVote in
Malta,1947‐2008."SouthEuropeanSocietyandPolitics,13:4,pp.435
‐56.
27. OrtegaVillodres,Carmen&delaPuerta,Belen.2004."Positioneffects
underSTV:IrelandandMalta."Representation,41:1,pp.3‐14.
28. Pelham,B.W.,Carvallo,M.,&Jones,J.T.(2005).Implicitegoism.
CurrentDirectionsinPsychologicalScience,14(2),106–110.
29. Rabin, M. & J. L. Schrag, "First Impressions Matter: A Model of
ConrmatoryBias,"QuarterlyJournalofEconomics,114(1999),37‐82.
30. Robson,Chris,andB.Walsh,"TheImportanceofPositionalVotingBias
intheIrishGeneralElectionof1973,"Political Studies, 22 (1974),
191–203.
31. Schwarz,Norbert,H.J.Hippler,andE.Noelle‐Neumann,"ACognitive
ModelofResponseOrderEffectsinSurveyMeasurement,"inNorbert
Schwarz and Seymour Sudman, ed., Context Effects in Social and
PsychologicalResearch(NewYork:Springer,1992),pp.187–201.
32. Simon,Herbert,ModelsofMan(NewYork:Wiley,1957).
33. Sudman, Seymour, Norman M. Bradburn,andNorbertSchwarz,
Thinking about Answers: The Application of Cognitive Processes to
SurveyMethodology(SanFrancisco:Jossey‐Bass,1996).
34. 34.Trench,B.etal.1987“Thealphabetadvantage”,inMagillBookof
Irishpolitics:ElectionsFebruary87, ed.V. Browne,Magill,D
ublin,
pp23.
35. Villodres,Carmen,"GenderandPartyDuopolyinaSmallState:Ballot
PositionEffectsundertheSingleTransferableVoteinMalta,1947‐
2008,"SouthEuropeanSocietyandPolitics,13(2008),435‐456.
36. Villodres,CarmenanddelaPuerta,BelenMorataGarcia,"Position
effectsunderSTV:IrelandandMalta,"Representation,41(2004),3‐
14.
37. Zajonc,RobertB.,"AttitudinalEffectsofMereExposure,"Journal of
PersonalityandSocialPsychology,9(1968),1‐27.

AppendixA
DependentVar:1stPrefe renceVotes
ABCD
OLS Fixe dEffects FixedEffects FixedEffects
1s tPOSITION 655.0*** 507.9*** 50 7.9*** 4 98.4**
(120.5) (112.5) (166.8) (252.2)
INCUMBENTTD 3955*** 2870*** 2870*** 774.3***
(68.85) (79.51) (123.9) (151.2)
GOVPARTYTD 318.9*** ‐833.4*** ‐833.4*** ‐465.5***
(84.94) (85.09) (95.00) (97.86)
GOVMINISTER 1030*** 1521*** 1521*** 743.0**
(242.0) (250.1) (305.6) (324.1)
POLITI CALPARTYLEADER 2 963*** 3640*** 3640*** 1701**
(525.3) (497.9) (657.9) (749.5)
Cons tant 4496*** 3621*** 3621*** 6986 ***
(209.8) (677.1) (977.5) (1732)
Observations 4807 4807 4807 4807
R‐s qua red 0.462 0.651 0.651 0.868
DependentVar:Log(Shareof1stPreferenceVot es)
ABCD
VARIABLES OLS Fixe dEffects FixedEffects FixedEffects
1s tPOSITION 0.228*** 0.175*** 0.175*** 0.0789
(0.0428) (0.0370) (0.0542) (0.05 50)
INCUMBENTTD 1.300*** 0.795*** 0.795*** 0.180***
(0.0221) (0.0251) (0.0426) (0.03 15)
GOVPARTYTD 0.391*** ‐0.235*** ‐0.235*** ‐0.0986***
(0.0283) (0.0246) (0.0275) (0.02 25)
GOVMINISTER ‐0.0390 0.212*** 0.212*** 0.1 09**
(0.0369) (0.0403) (0.0494) (0.05 03)
POLITI CALPARTYLEADER 0.636*** 0.720*** 0.720*** 0.271**
(0.0778) (0.0859) (0.112) (0.120)
Consta nt ‐2.631*** ‐3.166*** ‐3.166*** ‐1.482***
(0.0603) (0.322) (0.463) (0.540)
Observations 4807 4807 4807 4807
R‐s qua red 0.389 0.663 0.663 0.955
ColumnsBandCuseCons tituency,Election,No.ofCandidatesandPol iti c alPa r tyfixed
effects .
ColumnDaddsI ndividualCandidatefixedeffectstoColumnC
ColumnsAandBclusterattheConstituenc yElectionlevel.CandDclusteratthe
consti tuencyel ecti onandcandidatelevel
RobustSta ndarderrorsinpar entheses,***P<0.01,**P<0.05 ,*P<0.1

AppendixB
ABCDABCD
OLS FE FE Fe OLS FE FE FE
Posi ti on2 ‐382.7** ‐393.3*** ‐393.3** ‐516.4** ‐0.120* ‐0.138*** ‐0.138** ‐0.0831
(166.4) (151.4) (185.5) (259.8) (0.0628) (0.0505) (0.0612) (0.0553)
Posi ti on3 ‐611.0*** ‐507.6*** ‐507.6*** ‐466.1 ‐0.191*** ‐0.153*** ‐0.153** ‐0.0802
(149.6) (137.9) (183.6) (308.1) (0.0606) (0.0508) (0.0663) (0.0712)
Posi ti on4 ‐567.8*** ‐508.1*** ‐508.1*** ‐538.5 ‐0.169*** ‐0.160*** ‐0.160** ‐0.0658
(161.2) (147.1) (192.0) (342.3) (0.0642) (0.0515) (0.0673) (0.0762)
Posi ti on5 ‐773.2*** ‐613.2*** ‐613.2*** ‐732.3** ‐0.220*** ‐0.168*** ‐0.168** ‐0.139
(162.0) (147.4) (187.1) (360.0) (0.0641) (0.0543) (0.0697) (0.0857)
Posi ti on6 ‐722.3*** ‐498.6*** ‐498.6** ‐880.2** ‐0.201*** ‐0.114** ‐0.114 ‐0.163*
(160.3) (150.0) (203.3) (407.8) (0.0597) (0.0530) (0.0710) (0.0903)
Posi ti on7 ‐763.1*** ‐451.7*** ‐451.7** ‐755.2* ‐0.295*** ‐0.150*** ‐0.150** ‐0.133
(160.1) (144.8) (190.6) (418.7) (0.0689) (0.0570) (0.0753) (0.0978)
Posi ti on8 ‐655.5*** ‐523.4*** ‐523.4*** ‐993.8** ‐0.256*** ‐0.206*** ‐0.206*** ‐0.206*
(162.1) (146.6) (193.0) (432.8) (0.0678) (0.0563) (0.0697) (0.106)
Posi ti on9 ‐756.6*** ‐554.7*** ‐554.7*** ‐1114** ‐0.283*** ‐0.211*** ‐0.211*** ‐0.220**
(169.1) (152.0) (197.1) (468.0) (0.0732) (0.0597) (0.0730) (0.110)
Posi ti on10 ‐695.9*** ‐624.1*** ‐624.1*** ‐995.2* ‐0.262*** ‐0.256*** ‐0.256*** ‐0.166
(179.7) (164.8) (201.4) (510.8) (0.0795) (0.0657) (0.0776) (0.117)
Posi ti on11 ‐605.2*** ‐319.2 ‐319.2 ‐619.1 ‐0.376*** ‐0.264*** ‐0.264*** ‐0.122
(216.8) (200.1) (234.4) (524.1) (0.0999) (0.0802) (0.0931) (0.128)
Posi ti on12 ‐822.6*** ‐548.5*** ‐548.5** ‐1162** ‐0.322*** ‐0.261*** ‐0.261*** ‐0.256*
(212.1) (182.6) (224.6) (562.9) (0.104) (0.0783) (0.0912) (0.139)
Posi ti on13 ‐964.6*** ‐783.1*** ‐783.1*** ‐1286** ‐0.382*** ‐0.313*** ‐0.313*** ‐0.262*
(228.7) (203.7) (244.2) (585.3) (0.116) (0.0928) (0.103) (0.156)
Posi ti on14+ ‐608.5*** ‐456.5** ‐456.5* ‐739.4 ‐0.288** ‐0.217** ‐0.217** ‐0.222
(226.2) (214.5) (245.8) (641.6) (0.115) (0.0958) (0.110) (0.165)
IncumbentTD 3947*** 2867*** 2867*** 772.9*** 1.297*** 0.793*** 0.793*** 0.181***
(69.22) (79.65) (124.0) (151.1) (0.0224) (0.0251) (0.0426) (0.0314)
GovParty 311.5*** ‐836.0*** ‐836.0*** ‐469.4*** 0.387*** ‐0.237*** ‐0.237*** ‐0.0987***
(84.46) (84.96) (94.96) (97.25) (0.0282) (0.0247) (0.0275) (0.0226)
GovMinis te r 1032*** 1514*** 1514*** 740.1** ‐0.0367 0.210*** 0.210*** 0.107**
(242.2) (250.4) (305.1) (324.9) (0.0365) (0.0396) (0.0486) (0.0509)
Pa rtyLea der 2999*** 3653*** 3653*** 1719** 0.646*** 0.718*** 0.718*** 0.274**
(527.2) (497.6) (657.9) (758.6) (0.0787) (0.0865) (0.114) (0.124)
Cons ta nt 5059*** 4048*** 4048*** 7087*** ‐2.450*** ‐3.018*** ‐3.018*** ‐1.479***
(233.0) (693.9) (1007) (1720) (0.0691) (0.324) (0.463) (0.534)
Observ ations 4807 4807 4807 4807 4807 4807 4807 4807
R‐s qua red 0.464 0.652 0.652 0.869 0.392 0.664 0.664 0.956
RobustSta ndarderrors inpa renthese s,***P<0.01,**P<0.05,*P<0.1
1st Preference Votes Log(Share of 1st Preference Votes)
Colu m nsBandCuseCons tit uenc y,Election,No.ofCa ndidatesandPoli tica lPa rtyfixedeffects.
Colu m nDa ddsIndividua l Ca ndi date fixedeffe cts toColumnC
Colu m nsAandBcluste rattheConsti tue ncyEle ctionleve l.CandDclus terattheconsti tue ncyelection&ca ndidatelevel

AppendixC
ABCABC
FE FE logit FE FE logit
1s tPos i ti on 0.0113* 0.0113 0.0510 0.0130** 0.0130** 0.300
(0.00616) (0.00882) (0.221) (0.00543) (0.00602) (0.293)
IncumbentTD 0.0933*** 0.0933*** 2.814*** 0.0590*** 0.0590*** 2.1 81***
(0.00549) (0.00821) (0.211) (0.00354) (0.00465) (0.172)
Gov.Pa r ty ‐0.0214*** ‐0.0214*** ‐0.643*** ‐0.0196*** ‐0.01 96*** ‐0.622***
(0.00495) (0.00533) (0.186) (0.00406) (0.00430) (0.172)
GovMi ni s ter 0.0345** 0.0345* 0.612 0.0318*** 0.0318*** 0.799
(0.0152) (0.0182) (0.558) (0.0111) (0.0121) (0.559)
Pa r tyLeader 0.0880*** 0.0880*** ‐ 0.0999*** 0.0999*** 4.088**
(0.0328) (0.0251) ‐ (0.01 70) (0.023 6) (1.629)
Constant 0.107*** 0.107*** ‐0.148 0.0245*** 0.0245*** ‐1.01 8***
(0.0299) (0.0328) (0.370) (0.00662) (0.00838) (0.236)
Observati ons 1265 1265 1236 1655 1655 1652
R‐s quared 0.672 0.672 0.651 0 .655 0.6 55 0.651
ABC
FE FE logit
1s tPositi on 0.01 45*** 0.0145** 0.471**
(0.00419) (0.00596) (0.214)
IncumbentTD 0.0567*** 0.0567*** 2.526***
(0.00278) (0.00369) (0.157)
Gov.Pa r ty ‐0.0157*** ‐0.0157*** ‐0.820***
(0.00293) (0.00310) (0.144)
GovMi ni s ter 0.0389*** 0.0389*** 1.531***
(0.00878) (0.00940) (0.509)
Pa r tyLeader 0.0623*** 0.0623*** 2.440*
(0.0178) (0.0183) (1.452)
Constant 0.0461** 0.0461* ‐1.326***
(0.0198) (0.0257) (0.248)
Observati ons 1887 1887 1861
R‐s quared 0.651 0.651
4 SeaterConstituencies
5SeaterConstitue ncies
RegressionsuseConstituen cy,Election,
No.ofCandidatesandPoliti calPar tyfixedeffects .
ColumnBaddsIndi vi dualCandidatefixedeffectstoColumnA
Robust Standarderrorsinparentheses,***P<0.01,**P<0.05,*P<0.1
3Sea tConstituencies

AppendixD
(1) (2) (3)
Votes Share ofVotes Log(ShareofVotes)
1977El ec tion 856.8 *** 0.0240*** 0.30 8***
(310.6) (0.00921) (0.104)
1981El ec tion 401.7 0.00940 0.130
(428.5) (0.010 4) (0.144)
Feb1982El ecti on ‐106.7 ‐0.00563 0.0665
(345.4) (0.00858) (0.105)
Nov1982Electi on 511.9 0.0126 0.148
(373.3) (0.00985) (0.125)
1987El ec tion 77 7.0* 0 .0158 0.247*
(457.8) (0.011 6) (0.144)
1989El ec tion 474.0 0 .0115 0.0397
(434.2) (0.012 2) (0.169)
1992El ec tion 801.6** 0.0215** 0.318**
(349.8) (0.00895) (0.124)
1997El ec tion 800.0** 0.0205** 0.227*
(314.9) (0.00812) (0.129)
2002El ec tion 398.6 0.00799 0.110
(437.0) (0.010 2) (0.154)
2007El ec tion 76 4.5* 0.0183** 0.250**
(408.9) (0.00877) (0.109)
2011El ection ‐144.5 ‐0.00839 0.0402
(478.8) (0.010 1) (0.143)
I numbentTD 2875*** 0.0681*** 0.796***
(125.1) (0.00310) (0.0420)
GovernmentPa r ty ‐824.4*** ‐0.0190*** ‐0.232***
(94.66) (0.00226) (0.0274)
GovMi ni s ter 1525*** 0.03 65*** 0.210***
(300.8) (0.00694) (0.0491)
Pa r tyLeader 3634*** 0.0779*** 0.718***
(646.0) (0.014 9) (0.111)
Constant 3633*** 0.155*** ‐3.133***
(985.0) (0.023 2) (0.462)
Observati ons 4807 480 7 4807
R‐s quared 0.652 0.673 0.663
Chi 2(10) 9.98 15.01 8.1
Prob>chi 2 0.442 1 0.1317 0.6192
Stan darderrors
inparenthes es
***p<0.01,**p<0.05,*p<0.1

AppendixE
(1) (2) (3 )
Votes ShareofVotes Log(Sha reofVotes)
Fiann aFai l 816.5*** 1.832*** 0.132***
(180.0) (0.435 ) (0.0356)
FineGael 223.4 0.89 1* 0.0238
(187.6) (0.484 ) (0.0415)
LabourPa rty ‐796.0* ‐1.723* ‐0.250*
(409.2) (0.987 ) (0.149)
Independents 653.1** 1.118 0.629***
(280.1) (0.706 ) (0.152)
SmallPar ties 699.5** 1.345** 0.250**
(295.9) (0.636 ) (0.104)
IncumbentTD 2847*** 6.717*** 0.77 8***
(75.21) (0.194 ) (0.0232)
GovernmentTD ‐824.1*** ‐1.885*** ‐0.228***
(80.63) (0.187 ) (0.0233)
GovernmentMi ni s ter 1428*** 3.441*** 0.18 9***
(224.3) (0.505 ) (0.0338)
Pa r tyLeader 3491*** 7.519*** 0.66 5***
(458.9) (0.990 ) (0.0758)
Constant 6348 *** 20.06*** ‐1.912***
(284.0) (0.485 ) (0.0940)
Observati ons 4807 4807 4807
R‐s quared 0.639 0.666 0.654
Robuststandarderrorsinparenthes es
***p<0.01,**p<0.05,*p<0.1

AppendixF
ABCABC
ShareofVotes 1stPreferenceVotes Log(Share) ShareofVotes 1stPreferenceVotes Log(Share)
1s tPosi ti on 0.0366*** 579.9** 0.115*** 0.0236 280.5 0.0697*
(0.0127) (230.8) (0.0393) (0.0151) (173.6) (0.0405)
IncumbentTD 0.136*** 2385*** 0.503*** 0.211*** 2669*** 0.632***
(0.0124) (218.0) (0.0441) (0.0174) (198.0) (0.0481)
Gov.Pa rty ‐0.121** 1192 ‐0.217 ‐0.0987** ‐2649*** ‐0.253**
(0.0554) (1197) (0.178) (0.0411) (580.2) (0.122)
GovMinis ter 0.131*** 2406*** 0.382*** 0.183*** 2392*** 0.537***
(0.0272) (540.6) (0.0765) (0.0393) (585.8) (0.110)
Pa rtyLe a der 0.359*** 7493*** 0.991*** 0.196*** 4804*** 0.574**
(0.100) (2431) (0.322) (0.0701) (1080) (0.242)
Consta nt 0.138*** 4135*** ‐2.290*** 0.153*** 4084*** ‐1.820***
(0.0181) (557.9) (0.155) (0.0159) (349.7) (0.0725)
Obs e rvati ons 1291 1291 1291 1115 1115 1115
R‐s qua red 0.706 0.647 0.635 0.750 0.751 0.685
ABC
ShareofVotes 1stPreferenceVotes Log(Share)
1s tPosi ti on 0.0147 82.62 0.103
(0.0332) (158.0) (0.135)
IncumbentTD 0.632*** 4967*** 2.659***
(0.0548) (700.5) (0.213)
Consta nt 0.109* 1198** ‐2.960***
(0.0568) (586.1) (0.259)
Obs e rvati ons 1025 1025 1025
R‐s qua red 0.615 0.585 0.490
Robu s tSta nd ard e rrors inpa renthe s es ,***P<0.01,**P<0.05,
*P<0.1
FiannaFáilCandidates FineGaelCandidates
IndependentCandidates
Regressionsus eConsti tuency,Election,No .ofCandidatesandPol i ti ca l Pa rtyfixedeffects.
Col u mnBaddsI ndi vi dua lCandidatefixe d effectstoCol umnA

AppendixG
UCD CENTRE FOR ECONOMIC RESEARCH – RECENT WORKING PAPERS
WP11/18
Vincent Hogan, Patrick Massey and Shane Massey: 'Late Conversion:
The Impact of Professionalism on European Rugby Union' September 2011
WP11/19
Wen Fan: 'Estimating the Return to College in Britain Using Regression
and Propensity Score Matching' September 2011
WP11/20
Ronald B Davies and Amélie Guillin: 'How Far Away is an Intangible?
Services FDI and Distance' September 2011
WP11/21
Bruce Blonigen and Matthew T Cole: 'Optimal Tariffs with FDI: The
Evidence' September 2011
WP11/22
Alan Fernihough: 'Simple Logit and Probit Marginal Effects in R'
October 2011
WP11/23
Ronald B Davies and Krishna Chaitanya Vadlamannati: 'A Race to the
Bottom in Labour Standards? An Empirical Investigation' November 2011
WP11/24
Wen Fan: 'School Tenure and Student Achievement' November 2011
WP11/25
Mark E McGovern: 'Still Unequal at Birth - Birth Weight, Socioeconomic
Status and Outcomes at Age 9' November 2011
WP11/26
Robert Gillanders: 'The Mental Health Cost of Corruption: Evidence
from Sub-Saharan Africa' November 2011
WP11/27
Brendan Walsh: 'Well-being and Economic Conditions in Ireland'
December 2011
WP11/28
Cormac Ó Gráda: 'Fetal Origins, Childhood Development, and Famine:
A Bibliography and Literature Review' December 2011
WP12/01
Mark E McGovern: 'A Practical Introduction to Stata' January 2012
WP12/02
Colm McCarthy: 'Ireland’s European Crisis: Staying Solvent in the
Eurozone' January 2012
WP12/03
Colm McCarthy: 'Improving the Eurosystem for Old and New
Members' January 2012
WP12/04
Ivan Pastine and Tuvana Pastine: 'All-Pay Contests with Constraints'
February 2012
WP12/05
David Madden: 'Methods for Studying Dominance and Inequality in
Population Health' February 2012
WP12/06
Karl Whelan: 'ELA, Promissory Notes and All That: The Fiscal Costs of
Anglo Irish Bank' February 2012
WP12/07
Olivier Bargain, Eliane El Badaoui, Prudence Kwenda, Eric Strobl and
Frank Walsh: 'The Formal Sector Wage Premium and Firm Size for Self-employed
Workers' March 2012
WP12/08
Brendan Walsh: 'The Influence of Macroeconomic Conditions and
Institutional Quality on National Levels of Life Satisfaction' March 2012
WP12/09
Ronald B Davies and Rodolphe Desbordesz: 'Greenfield FDI and Skill
Upgrading' March 2012
WP12/10
Morgan Kelly and Cormac Ó Gráda: 'Change Points and Temporal
Dependence in Reconstructions of Annual Temperature: Did Europe Experience a
Little Ice Age?' March 2012
WP12/11
Morgan Kelly and Cormac Ó Gráda: 'The Waning of the Little Ice
Age' April 2012
WP12/12
Morgan Kelly and Cormac Ó Gráda: 'Agricultural Output, Calories and
Living Standards in England before and during The Industrial Revolution' April
2012
WP12/13
Arnaud Chevalier and Orla Doyle: 'Schooling and Voter Turnout - Is
there an American Exception?' April 2012
WP12/14
David Madden: 'The Relationship Between Low Birthweight and
Socioeconomic Status in Ireland' April 2012
WP12/15
Robert W Fairlie, Kanika Kapur and Susan Gates: 'Job Lock: Evidence
from a Regression Discontinuity Design' April 2012
UCD Centre for Economic Research Email econom[email protected]