Ellge , Fabio
A icle — Published Ve sion
The Mobilizing E ec o Pa y Sys em Pola iza ion.
E idence F om Eu ope
Compa a i e Poli ical S udies
P o ided in Coope a ion wi h:
WZB Be lin Social Science Cen e
Sugges ed Ci a ion: Ellge , Fabio (2024) : The Mobilizing E ec o Pa y Sys em Pola iza ion. E idence
F om Eu ope, Compa a i e Poli ical S udies, ISSN 1552-3829, SAGE Publishing, Thousand Oaks, CA,
Vol. 57, Iss. 8, pp. 1310-1338,
h ps://doi.o g/10.1177/00104140231194059
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/307747
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
A icle
Compa a i e Poli ical S udies
2024, Vol. 57(8) 1310–1338
© The Au ho (s) 2023
A icle euse guidelines:
sagepub.com/jou nals-pe missions
DOI: 10.1177/00104140231194059
jou nals.sagepub.com/home/cps
The Mobilizing E ec o
Pa y Sys em
Pola iza ion. E idence
F om Eu ope
Fabio Ellge
1
Abs ac
Does pa y sys em pola iza ion mobilize o e s? Pola iza ion is inc easingly
shaping democ a ic compe i ion ac oss Eu ope. While o en pe cei ed o be
nega i e, pola iza ion can be an e ec i e emedy agains o e disengagemen .
This pape in es iga es wo dis inc , bu o en confla ed mechanisms, which
could explain why pola iza ion leads o mobiliza ion. Spa ial pola iza ion o
pa ies di e sifies elec o al op ions a he ballo , while a ec i e pola iza ion
mobilizes based on emo ional conside a ions. This a icle hen shows he link
be ween pola iza ion and u nou ac oss 22 Eu opean coun ies. The esul s
a e complemen ed by a di e ence-in-di e ences analysis o Ge man local
elec ions. Howe e , o ing esul s alone do no in o m abou he mechanism
a play. Su ey da a is used o show ha nega i e a ec appea s o be he main
d i e o o e pa icipa ion. Pa y pola iza ion hus has ambi alen conse-
quences o democ acies: I mobilizes he elec o a e, bu i s e ec is d i en by
nega i e emo ions.
Keywo ds
pola iza ion, o ing beha io , o e u nou , ep esen a ion
1
T ans o ma ions o Democ acy Uni , WZB Be lin Social Science Cen e , Be lin, Ge many
Co esponding Au ho :
Fabio Ellge , T ans o ma ions o Democ acy Uni , WZB Be lin Social Science Cen e ,
Reichpie schu e 50, Be lin 10785, Ge many.
Email: [email p o ec ed]
Da a A ailabili y S a emen included a he end o he a icle
In oduc ion
Wha did he gene al elec ions o Ge many 2017, Spain 2015, and Aus ia
1995 ha e in common? A a fi s glance, he dynamics a ound elec ion day
ha e been qui e dissimila : The Ge man igh -wing pa y “Al e na i e o
Ge many”(A D) gained ep esen a ion o he fi s ime in 2017 and became
he la ges ac ion in opposi ion. In 2015, Spanish an i-aus e i y mo emen s
joined o ces as he newly o med Podemos pa y was success ully en e ing
pa liamen on a adical le pla o m. Finally, he es ablished F eedom Pa y o
Aus ia FP ¨
Ohad only ecen ly o e hauled i s p ofile in 1995 unde i s new
eccen ic leade : J¨
o g Haide abandoned mode a e posi ions in exchange o
populis and na ionalis appeals. These changes in Eu opean pa y sys ems
o igina ed om di e en ac ions and dynamics, such as pa y o ma ions o
leade ship change. Bu all h ee elec ions expe ienced a significan inc ease in
u nou as pa ies we e able o mobilize la ge sha es o he elec o a e. This
pape a gues ha he e has been a common cause: an inc ease in pola iza ion
o he pa y sys ems.
These examples also e eal ha pola iza ion can be ola ile in sys ems o
p opo ional ep esen a ion, which a e associa ed wi h highe deg ees o
pola iza ion in he fi s place. To es he a gumen o o e mobiliza ion
h ough pola iza ion, I conduc a ange o c oss-sec ional and panel da a
analyses.
1
Fi s , I show ha pola iza ion is associa ed wi h highe le els o
u nou in 22 pa liamen a y democ acies ac oss Eu ope. Second, I make use o
he une en pa icipa ion o he newly o med A D a Ge man local elec ions
a ound 2013 o implemen a di e ence-in-di e ences design. The esul s
show ha he “Al e na i e o Ge many,”as a newly o med igh -wing pa y
boos ed pa icipa ion in locali ies whe e i was on he ballo . This empi ical
case s udy nicely complemen s he ini ial esul s. Bo h analyses sugges ha
pa y pola iza ion and o e u nou a e closely linked, bu e eal li le in-
o ma ion on he mechanism ha connec s bo h phenomena. In a hi d
analysis, I he e o e in es iga e dynamics wi hin c oss-coun y su ey da a o
disen angle spa ial om a ec i e pola iza ion. I con ibu e o he g owing
li e a u e on pola iza ion and o ing beha io by ope a ionalizing spa ial and
a ec i e pola iza ion as dis inc mechanisms. Upon close analysis, nega i e
a ec appea s o be he mos ele an p edic o o u nou .
O e all, pola iza ion can be conside ed an e ec i e emedy agains po-
li ical disengagemen . Howe e , inc easing o e in ol emen comes a a
subs an ial cos : I u nou is d i en by nega i e emo ions, pola iza ion can
u he unde mine he legi imacy o democ a ic decision-making. In he
concluding discussion, I elabo a e on po en ial implica ions o hese findings
o he s udy o pola iza ion and u nou in Eu ope and beyond. Impo an ly,
his s udy highligh s he a ec i e componen o mobiliza ion, which so a
ecei ed li le a en ion in compa a i e esea ch. In pa , his is due a lack o
Ellge 1311
consensus a ound a clea concep ualiza ion and measu emen o a ec i e
pola iza ion in mul ipa y democ acies (see Boxell e al., 2022;Wagne , 2021,
o aluable con ibu ions). This pape unde sco es ha due o he di e en
ways in which pola iza ion e ol es unde p opo ional ep esen a ion, simple
adap a ions o concep s and measu emen s om he US con ex ha e some
sho comings. Building on p e ious wo k, I also p opose a new way o
di e en ia e posi i e om nega i e a ec in a compa a i e analysis o
cons i uen a i udes. Wi h ega d o u nou , he main implica ion o his
a icle is qui e sobe ing. A e decades o pe sis en u nou decline, many
es ablished democ acies in Eu ope ecen ly expe ienced mo e pa icipa ion.
The o e all findings o my analyses sugges ha his spike could a leas
pa ially be explained by ising nega i e a ec ac oss pa ies and hei
suppo e s. E en s like he 2015 e ugee influx, he elec o al success o
populis pa ies, o dispu e abou he Co id- ela ed policies ha e ein o ced
socie al ension. This can become ha m ul i poli ical ac i ism ueled by
animosi y mo i a es illici beha io such as iolen p o es s o he a ge ed
ha assmen o democ a ic ep esen a i es. Ul ima ely, inc easingly pola ized
pa y sys ems ha e he po en ial o unde mine he poli ical cohesion o es-
ablished democ acies.
Pola iza ion and Vo e Tu nou
So a , pola iza ion has mos ly been seen as an undesi able end in es ablished
democ acies. Especially in he US, schola s end o blame eli e pola iza ion o
a ange o poo ou comes. This includes lacking le els o ep esen a ion and
eli e-cons i uen alignmen (Fio ina & Ab ams, 2012;Poole & Rosen hal,
1984), legisla i e g idlock (Jones, 2001), o he splin e ing and poli iciza ion
o he media landscape (P io , 2013).
This pape zooms in on he consequences o pa y sys em pola iza ion in
es ablished democ acies. This concep o pola iza ion implies ha ideological
posi ions ac oss pa ies d i apa (Dal on, 2008). I connec s o wo k by
schola s who emphasize ha pa y eli es ake inc easingly di e ging posi ions
on salien issues whene e pola iza ion is high (He he ing on, 2001;Poole &
Rosen hal, 1984). Bu no only did he ideological dis ance among pa y eli es
inc ease. Vo e s also s a ed o place hei own p e e ences close o hose o
hei a o able pa y (Ab amowi z & Saunde s, 2008;Lenz, 2013). Unde he
labels o “so ing”(Baldassa i & Gelman, 2008;Fio ina & Ab ams, 2008;
Le endusky, 2009)o “ ealignmen ”(Ab amowi z & Saunde s, 1998),
schola s o US poli ics desc ibe ha pa y eli es and hei pa isan o e s ally
close a ound ce ain co e policies. Pola iza ion he e o e is also associa ed
wi h inc easing ideological homogenei y wi hin pa ies, dis inc policy p o-
files, and less o e lap ac oss pa y lines.
1312 Compa a i e Poli ical S udies 57(8)
Howe e , inc easing ideological disag eemen is no he only way in which
pola iza ion has been concep ualized and examined. Ra he han looking a
changing policy p e e ences, some schola s highligh inc easingly hos ile
a i udes owa ds poli ical opponen s and blind obedience o one’s p e e ed
pa y eli es. Banda and Clu e ius (2018) find ha when eli e posi ions di e ge,
pa isans de elop inc easing eelings o animus owa ds he opposing pa y.
And Johns on (2018) shows how posi ion- aking by eli es ele a es au ho i-
a ian eelings among pa isans. As pe sonal and poli ical iden i ies s a o
align, cons i uen ’s emo ions become a s onge mo i a ion o ac ual beha io
(see Achen & Ba els, 2016;Taj el, 1982;Tu ne e al., 1979). Vo e s engage
in poli ics no o see hei p e e ences pu in o legisla ion, bu o “exp ess”a
salien pa o hei iden i y (Huddy e al., 2015). S ong iden ifica ion wi h a
poli ical camp can also p o oke s ong animosi y agains poli ical opponen s.
A ange o s udies show ha his nega i e iew o poli ical compe i o s is
inc easing o e ime in a ange o es ablished democ acies (Boxell e al., 2022;
Iyenga e al., 2012) and shapes cons i uen beha io e en beyond poli ics
(McConnell e al., 2018).
I a gue ha bo h na a i es associa ed wi h pa y pola iza ion can ha e a
dis inc e ec on u nou , as illus a ed in Figu e 1. Pola iza ion can ei he
inc ease he dispe sion o poli ical o e ings ac oss he ideological ange (see
Dal on, 2008). Spa ially pola ized eli es wi h cohe en pa y p e e ences make
i easie o o e s o iden i y and suppo a pa y close o hei own ideological
iews (Le endusky, 2010;Lupu, 2015). Cons i uen s a e con on ed wi h
mo e dis inc elec o al op ions in a pola ized pa y sys em (Dal on, 2008).
Based on a a ional unde s anding o o ing beha io , his should inc ease he
u ili y o pa icipa ion (Hobol & Hoe ne , 2020;Mo al, 2017).
Bu o e mobiliza ion could also be a consequence o inc easingly salien
pa isan iden i ies (Huddy, 2001;Mason, 2018). In his scena io, beha io is
Figu e 1. Illus a ion o heo e ical a gumen .
Ellge 1313
no policy-, bu iden i y d i en (see Achen & Ba els, 2016). When he
ideological o e lap be ween i aling pa ies sh ink and hei ollowe s so
in o homogeneous g oups, app ecia ion o demands by poli ical compe i o s
declines (Iyenga & Wes wood, 2015). The e o e, indi idual pa icipa ion
could esul om “ange , p ejudice, and ac i ism on behal o ha p ejudice
and ange ”(Mason, 2018, p. 4). Fo example, Ab amowi z and S one (2006)
a gue ha he 2004 Ame ican p esiden ial elec ions expe ienced a spike in
u nou no because o e s we e di ided o e p esiden Bush’s p oposed
policies, bu because his pe sona igge ed emo ions among Republicans and
Democ a s alike. As he au ho s pu i , “Ame icans ei he lo ed him o ha ed
him and […] wen o he polls in eco d numbe s o exp ess hose eelings”
(Ab amowi z & S one, 2006, p. 141).
Fo concep ual cla i y, I ea spa ial and a ec i e pola iza ion as dis inc
mechanisms ha link pa y sys em pola iza ion o highe u nou . Howe e ,
bo h concep s a e in e wined and po en ially ein o cing. To my bes
knowledge, ew s udies explici ly model he ela ionship be ween a ec i e
and ideological pola iza ion. A ecen exemp ion is a seminal s udy by
D uckman e al. (2021). In he con ex o Ame ican public opinion o ma ion
du ing Co id, he au ho s find ha p io a ec i e pola iza ion p edic s highe
deg ees o ideological pola iza ion. Thei s udy sugges s ha when pa isan
animus is high al eady, subsequen policy p e e ences will be di ided. Pa s o
he li e a u e con e sely claim ha a ec i e pola iza ion has inc eased due o
g ea e ideological ex emi y (Boughe , 2017;Webs e & Ab amowi z, 2017).
Acco ding o he au ho s, adicalizing pa y eli es and he so ing in o cohesi e
pa isan camps ha e spa ked he g ow h o a ec i e ou -g oup hos ili y.
Confi ming he iew o policy-based a ec i e pola iza ion, Lelkes (2021)
finds ha pa y eli es wi h adical ideologies inc ease a ec i e pola iza ion
among he elec o a e.
One ca ea o hese esea ch endea o s is ha hey mos ly a e si ua ed
wi hin he Ame ican wo-pa y sys em. Unde plu ali y ule wi h single-
membe dis ic s, ba ie s o en y o new pa ies a e high. Hence, pa y
sys em pola iza ion is mos ly d i en by es ablished pa ies’mo ing o ex eme
posi ions and g owing animus ac oss exis ing poli ical camps. This s udy
howe e ocuses on he consequences o pa y sys em pola iza ion in Eu-
opean mul ipa y democ acies, many o which ely on a a ian o p o-
po ional ep esen a ion (PR). He e, pa y compe i ion is much mo e ola ile
and he o ma ion, discon inua ion, me ge , o sepa a ion o pa ies occu
equen ly. Mo eo e , elec o al ola ili y ac oss pa ies and pa y amilies is
mo e p onounced. Based on he concep ualiza ion o pa y sys em pola i-
za ion by Dal on (2008), one can in e h ee po en ial causes o ising po-
la iza ion in PR sys ems: Fi s , new adical pa ies can eme ge and gain
elec o al suppo . Second, exis ing pa ies can mo e owa ds mo e ex eme
posi ions and main ain o e en gain elec o al suppo . And hi d, e en when
1314 Compa a i e Poli ical S udies 57(8)
pa ies do no eposi ion, o e s can shi hei suppo om cen is owa ds
mo e adical pa ies.
Pa y sys em pola iza ion hus appea s mo e complex in a PR con ex , a
ac ha has ecei ed ela i ely li le a en ion in he ecen li e a u e on he
o igins o pola iza ion. No able s udies on his opic p ima ily pay a en ion o
he consequences o new pa y en y (see Bischo & Wagne , 2019;Leininge
& Meije s, 2021). Bu o he bes o my knowledge, he e exis ew sys ema ic
assessmen s o pa y adicaliza ion o shi ing o e suppo as de e minan s o
pola iza ion beyond he US.
The scope o his a icle howe e emains he consequence o pola iza ion
o o ing beha io . Tha is, I am somewha agnos ic abou he o igins o
pola iza ion wi hin a gi en coun y a a specific poin in ime. I is none heless
in e es ing o si ua e he le el o pa y sys em pola iza ion ac oss di e en
ypes o democ acies in a compa a i e pe spec i e. A simple analysis con-
duc ed in Appendix Sec ion A.4 compa es pola iza ion ac oss he h ee main
elec o al sys ems based on exis ing CSES da a. In acco dance wi h he
classical compa a i e pa y sys ems li e a u e, I find ha PR sys ems possess
he highes le els o poli ical pola iza ion (Downs, 1957;Sa o i, 1976). How
does his ela e o he dynamics a ound pola iza ion and mobiliza ion? I a gue
ha es ing his ela ionship unde ules o p opo ional ep esen a ion could
be mo e challenging. Fi s , he causes o pola iza ion a e mo e complex. In
o de o co ec ly assess changing pola iza ion in a mul ipa y sys em, o e s
need o be awa e o pa y en ies o exi s, he ideological pla o m o hese new
pa ies, as well as adjus men s in ideology by all exis ing pa ies. Mo eo e ,
hey need o accoun o elec o al shi s owa ds mo e o less ex eme pa ies
a ound elec ions. All in all, acking hese di e en aspec s o pola iza ion
unde PR- ule can be challenging and may lead o he ac ha ising po-
la iza ion has a mo e subdued e ec on o ing beha io . In con as , pa y
sys em pola iza ion in he US has p ima ily been d i en by g owing ideo-
logical di ides and animus ac oss wo exis ing poli ical pa ies wi h ela i ely
s able suppo bases. La ge elec o al swings o new pa y en y a e less
common unde plu ali y ule, hus po en ially simpli ying o e assessmen o
changing pola iza ion. The nex sec ion cla ifies and o malizes how bo h
spa ial and a ec i e pola iza ion a e expec ed o ope a e in a mul ipa y
con ex .
Spa ial Pola iza ion and he Decision o Vo e
Concep ualizing spa ial pola iza ion assumes ha poli ical p e e ences can be
o de ed along policy dimensions in which ac o s pu pose ully choose hei
posi ions. This incen i izes pa ies o di e si y hei p ofiles, sp ead ac oss
he ideological spec um, and ca e o o e s a mode a e and ex eme
posi ions. The s a egic posi ioning ac oss he policy space can be explained as
Ellge 1315
o e-maximizing beha io in which pa ies sea ch an elec o al niche (Cox,
1990). This unde s anding o pa y beha io o igina es om he spa ial
p oximi y model o o ing (Downs, 1957;S okes, 1963). I posi s ha o e s
delibe a ely choose he pa y ha is closes o hei pe sonal policy p e e -
ences. No e ha he spa ial p oximi y model is only one subb anch o exis ing
spa ial models on o ing beha io , which o ins ance also inco po a e heo ies
on di ec ional o ing (Rabinowi z & Macdonald, 1989). To he bes o my
knowledge, mos esea ch on pa y sys em pola iza ion and spa ial o ing
ei he implici ly o explici ly adop s he p oximi y model.
Gi en i s co e assump ions, he p oximi y model o o ing o esees a
sys em o spa ial ep esen a ion in which pa ies ake an ac i e ole and
posi ion hemsel es s a egically o dec ease he dis ance owa ds o e s
(S imson e al., 1995). This dynamic p ocess implies ha indi idual u ili y
om o ing is dynamic and depends on pa y ( e-)posi ioning a a gi en
elec ion. In o de o decide whe he o pa icipa e o abs ain, a ional o e s
engage in a simple cos -benefi analysis (Ald ich, 1993;Rike & O deshook,
1968). In his pape , I ocus on a ia ion in pe cei ed u ili y associa ed wi h
ising pa y sys em pola iza ion.
U ¼jVPj¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð ile p ileÞ2
q(1)
Cons i uen s’u ili y o o ing U
depends on he posi ioning o pa ies
be o e elec ions. I is o malized as being he in e se o he absolu e dis ance
be ween o e s and pa ies along he gene al igh -le dimension. Spa ial
pola iza ion al e s he pe cei ed u ili y om o ing by changing he supply
side o poli ics. In a spa ially pola ized sys em, cons i uen s can choose om a
wide ange o elec o al al e na i es posi ioned ac oss he en i e ideological
spec um (Dal on, 2008;Mo al, 2017). I can be caused by o me ly mode a e
pa ies ha adop mo e adical posi ions o add ess new g oups in he
elec o a e, a s a egy known as di e en ia ion and p ofile building (Cox, 1990;
Spoon & Klü e , 2019). Bu spa ial pola iza ion can also esul om new
pa y en y (Ta i s, 2006). I pa s o he elec o a e a e no su ficien ly ep-
esen ed by es ablished pa ies, incen i es o new ac o s o appeal o un-
de ep esen ed o e g oups a e high. This pola izing e ec o eme ging
pa ies is eflec ed in he niche pa y li e a u e (Meguid, 2005,2008). Fo
example, o e s wi h en i onmen al p e e ences ha e been poo ly ep esen ed
by es ablished social democ a ic and conse a i e pa ies in he 1980s, which
changed a e he o ma ion o new g een pa ies ac oss Eu ope (Ki schel ,
1989). The inco po a ion o he G eens in o he adi ional pa y sys em has
aised he dispe sion o p e e ences, hus esul ing in highe le els o spa ial
pola iza ion. The adical igh , ano he g oup o newly o med poli ical
pa ies, has had an e en mo e subs an ial impac on ideological pola iza ion
1316 Compa a i e Poli ical S udies 57(8)
(Ki schel & McGann, 1997). Thei na ionalis and an i-immig a ion posi ions
significan ly b oadened he spec um o ideological iewpoin s ep esen ed in
many Eu opean pa y sys ems.
Spa ially pola ized sys ems wi h pa ies ca e ing o cons i uen s ac oss he
whole ideological ange— om libe a ian, o e en i onmen alis , o na-
ionalis p e e ences—on a e age minimize he dis ance be ween pa y and
cons i uen posi ions. This should inc ease he le els o pe cei ed u ili y o
o ing o indi iduals, who a e able o iden i y a pa y ha ep esen s hei
in e es s easonably well. Thus, I expec an inc ease o mobiliza ion.
Hypo hesis 1: Highe deg ees o spa ial pola iza ion mobilize o e s a a
gi en elec ion.
A ec i e Pola iza ion and he Decision o Vo e
When a ec i e pola iza ion igge s mobiliza ion, I do no assume ha o e s
a e ac ing pe ec ly a ional. To he con a y, I expec hem o make sense o
poli ics in e ms o social g oup ela ionships (Achen & Ba els, 2016;
Fes inge , 1954). This pa h is oo ed in social iden i y heo y and ela ed
heo ies o ela i e g oup wo h (see Huddy, 2001;Taj el, 1982;Tu ne e al.,
1979). Wi hin his amewo k, indi iduals a e p esumed o ca ego ize
hemsel es and hei pee s in o socially cons uc ed g oups. The mo e pa -
isanship becomes pa o his sel -de e mined iden i y, he likelie i is ha
emo ions a e a key d i e o poli ical beha io (Iyenga e al., 2019). Mason
(2018) defines a ec i e pola iza ion as he inc easing social dis ance be ween
pa isans based on h ee phenomena: “inc eased pa isan bias, inc eased e-
ac i i y [based on his bias], and inc eased ac i ism”(2018, p. 17), he la e
sugges ing ha g owing pa isanship is likely o igge poli ical ac ion.
Especially in he US, a ec i e pola iza ion has become a widely s udied
phenomenon (Iyenga e al., 2012;Rogowski & Su he land, 2016). The
hos ili y and p ejudice acing poli ical opponen s has g own (Mason, 2018).
Bu a ec i e pola iza ion is impe ec ly explained by di e ging pa y posi-
ions alone (Iyenga e al., 2012). Dynamics o in-g oup a o i ism and ou -
g oup animosi y eme ge (Taj el, 1982;Tu ne e al., 1979), which go beyond a
“ a ional”dis ega d o he opposing posi ions on ce ain policies (Lelkes,
2018).
U ¼Fing oup Fou g oup (2)
The simple equa ion o malizes he in ui ion behind he a ec i e pola i-
za ion p oposi ion. Vo e u ili y U
is now he esul o in-g oup eelings
F
ing oup
compa ed ou -g oup eelings F
ou g oup
. The g oups a e subjec i ely
classified by cons i uen s along pa y lines. Seeing you own pa y win, o he
Ellge 1317
Dependen Va iable and Es ima ion. Idefine u nou a he coun y elec ions as
he dependen a iable, which is again scaled om 0 o 100. The subsequen
es ima ion s a egy is a plain applica ion o a di e ence-in-di e ences
analysis wi h wo ime poin s and depic ed in he equa ion below (see
Ang is & Pischke, 2008, p. 227):
Tu nou u¼αsþβ1A dFoundeduþβ2A dCandida eu
þβ3A dFounded*Candida euþϵ
(4)
Acco ding o he model, he in e ac ion e m A dFounded *Candida e
u
deno es he change in u nou o coun ies uwhe e he A D was on he ballo
compa ed o he change whe e hey did no field a candida e. No e ha he
p e e ed specifica ion also includes α
s
s a e fixed e ec s o α
e
Eas fixed
e ec s o accoun o unobse ed di e ences ac oss egions. S anda d e o s
a e clus e ed on he coun y le el.
Discussion o Resul s. Resul s o he analysis a e p esen ed in Table 2. I obse e
a nega i e e ec when mo ing om he p e-A D o pos -A D elec ion pe iod,
indica ed by he nega i e coe ficien o A D ounded. Tu nou dec eases by
oughly 3 pe cen age poin s. This confi ms ha u nou decline also pe sis s in
Ge man local poli ics. By looking a he es ima e o A D candida e, we see
ha he pa y appea s o field candida es in coun ies ha su e om lowe
le els o pa icipa ion. The subsample o coun ies in which he A D exis s on
Table 2. Tu nou A ound A D o ma ion in Ge man Local Elec ions (2008–2016).
Dependen Va iable
Tu nou
(1) (2) (3)
A D ounded 3.005*** (.280) 3.005*** (.282) 3.005*** (.280)
A D candida e 5.609*** (.732) 2.819*** (.850) 5.275*** (.685)
A D ounded*A D candida e 2.438*** (.393) 2.438*** (.396) 2.438*** (.393)
Cons an 56.332*** (.604)
Fixed e ec s —S a e Eas
Obse a ions 756 756 756
Coun ies 378 378 378
Adjus ed R
2
.103 .370 .204
No e. The able shows coe ficien s ob ained om mul i a ia e OLS- eg ession models. S anda d
e o s a e clus e ed by coun y. Model wo includes s a e fixed e ec s and model h ee includes
Eas fixed e ec s. *p< .1; **p< .05; ***p< .01.
1324 Compa a i e Poli ical S udies 57(8)
he ballo s a ou wi h abou 3–5 pe cen age poin s less u nou depending on
he model specifica ion.
Finally, he in e ac ion e m e eals he di e ence-in-di e ences, o di -
e ence in u nou change be ween ea men and con ol g oup. When
compa ing changes in u nou om coun y elec ions whe e he A D has
fielded candida es o he un ea ed coun e pa s, he o me expe ience an
inc ease o oughly 2.4 pe cen age poin s. Albei he ope a ionaliza ion o
pola iza ion, es ima ion s a egy, and elec o al con ex meaning ully di e s
om he fi s analysis, he e ec magni ude is su p isingly simila . Local pa y
sys em pola iza ion, induced by A D p esence a coun y elec ions, indeed
inc eases o e pa icipa ion by 2.4 pe cen age poin s. The esul s hold when
including s a e o Eas fixed e ec s. Appendix Sec ion A.9 u he explo es he
e ec o he o me Eas e n and Wes e n coun ies, showing ha he di ec ion
o he e ec emains simila in bo h subsamples, bu he o al e ec is la gely
d i en by coun ies in Wes e n Ge many, which cons i u e 80% o he o al
sample. This di e en ial e ec could be caused by a lack o obse a ions in he
Eas , bu also by highe p eexis ing pola iza ion due o he success o he
adical le pa y.
Validi y Checks and Limi a ions. The di e ence-in-di e ences es ima ion elies
on a ange o assump ions o ensu e causal in e ence. Among he mos
ele an is he assump ion o pa allel ends. I posi s ha he ea ed and
un ea ed coun ies a e on simila ajec o ies wi h ega d o he ou come p io
o being ea ed. Figu e 3 shows ha ends in u nou a e i ually iden ical o
coun ies in which he A D fielded o did no field candida es in coun y
elec ions. Bo h plo s show local u nou s a ing in 1990. The le pane as-
sumes a linea end—and bo h slopes a e i ually he same. The igh pane
allows o local weigh ing and o e s a mo e nuanced pic u e o u nou ends.
Up un il he ea ly 2000s, u nou declined s eeply, be o e he cu e simul-
aneously fla ens o bo h ea ed and un ea ed coun ies.
In addi ion, I conduc a se ies o alidi y checks in he appendix. Fi s , I es
i he occu ence o A D candidacy can be explained by ends in he ou come.
This is o ule ou he suspicion ha he A D field candida es in coun ies whe e
u nou significan ly ose o declined e en p io o i s o ma ion, pe haps o
s a egic easons. Table A.13 shows ha ea ed coun ies ha e been on simila
ajec o ies wi h ega d o pa icipa ion and ha changes in u nou do no
p edic A D candidacies. In a second alidi y check, I es i ea ed and
un ea ed coun ies di e on o he impo an dimensions. Figu e A.1 shows
ha o he co a ia e ends do no p edic he ea men s a us. Tha is, he A D
is no mo e likely o field candida es in coun ies ha expe ience popula ion
decline, economic decline (measu ed as coun y-le el GDP pe capi a change)
o unemploymen change. Acco dingly, Table A.14 shows he balance ac oss
said co a ia es o he ea ed and un ea ed coun ies. In o al, he alidi y es s
Ellge 1325
a e in line wi h he assump ions p esupposed by he di e ence-in-di e ences
design. Coun ies wi h he A D on he ballo a e on no di e en ajec o y wi h
ega d o u nou p io o ea men and do no di e on a ange o ele an
social and economic de elopmen s.
The local le el analysis s ill comes wi h impo an limi a ions, mos o
which a e discussed in mo e de ail in he appendix as well. Fi s , he fielding o
adical igh candida es ac s as a simple p oxy o a ec i e pola iza ion. In
Appendix Sec ion A.7, I show ha coun ies wi h A D candida es indeed
became mo e pola ized, bu I canno ule ou he e ec o o he ac o s ela ed
o A D en y. This could be inc eased media epo ing o s onge mobiliza ion
e o s by es ablishmen pa ies whe e e he A D fielded candida es. Thus,
my ea men should be conside ed as a p oxy measu e o pa y sys em
pola iza ion and causal es ima es ha e o be in e p e ed acco dingly. Nex , he
ques ion a ises i dynamics on he local le el a e gene alizable o na ional
poli ics. As highligh ed ea lie and in Appendix Sec ion A.8, Ge man local
and na ional poli ics sha e many o mal and in o mal ea u es. Bo h a e
s uc u ed along simila ideological lines o conflic and possess he same
poli ical ac o s. Mo eo e , o e anonymi y is highly ega ded e en in local
elec ions, making in o mal p essu es o kinship ne wo ks less po en . Em-
pi ically, he appendix sec ion also shows ha elec o al beha io on he local
le el explains na ional elec ion esul s. While u nou is mo e a enua ed on
he local le el, pa e ns be ween local and ede al elec ions a e ex emely
Figu e 3. Pa allel ends in u nou o Ge man local elec ions.
No e. The plo s show ends in u nou o he Ge man coun ies in he sample
s a ing in 1990. The le plo shows linea ends, whe eas he igh plo shows ends
based on local non-pa ame ic weigh ing. Red lines indica e ends o un ea ed
coun ies and g een lines show ends o ea ed coun ies. The e ical black line
indica es he o ma ion o he A D in 2013.
1326 Compa a i e Poli ical S udies 57(8)
simila wi h ega d o ex eme pa y o ing. Ex eme pa y o e sha es in local
elec ions a e highly p edic i e o ex eme pa y o e sha es du ing na ional
elec ions, indica ing ha he na u e o poli ical compe i ion is compa able in
bo h a enas.
Howe e , based on he second analysis alone, one canno assume simila
mobiliza ion e ec s a ound new pa y en y o he en i e c oss-coun y
sample. As discussed in he heo y sec ion, pa y sys em pola iza ion is
complex and mul i- ace ed. Pola iza ion ac oss Eu ope is no only d i en by
he eme gence o new ex eme pa ies, bu also shaped by pa y eposi ioning
and o e fluc ua ions. A final limi a ion is ha none o he esul s p esen ed so
a can di e en ia e be ween a ec i e and spa ial pola iza ion. The nex
sec ion he e o e explo es bo h mechanisms on he indi idual le el.
Analysis 3: Disen angling A ec i e and Spa ial Pola iza ion
So a , he esul s seem o co obo a e he na a i e o “bene olen ”pola i-
za ion which os e s mo e engagemen . Bu elec ion esul s alone e eal li le
abou he unde lying mechanism and mo i a ion o u n ou . Why do mo e
people go o he ballo in a pola ized pa y sys em? I discussed wo plausible
mechanisms ha could link pola iza ion o u nou . Pa y sys em pola iza ion
is associa ed wi h a mo e p onounced di e en ia ion o elec o al al e na i es.
I pa ies o e a wide ange o policy posi ions, o e s a e be e able o
dis inguish hei p ofiles and also ha e a highe chance o finding a pa y ha
ep esen s hei own p e e ences easonably well. Fo a a ional o e , his
de elopmen is likely o inc ease he u ili y o pa icipa ion.
Howe e , pola iza ion is also associa ed wi h shi ing eelings owa ds
poli ical ac o s. When pa ies ake on undamen ally di e en posi ions on
salien issues and pa y eli es figh wi h hei compe i o s, cons i uen s end o
c ea e an inc easing emo ional bond wi h he pa y hey iden i y wi h (Lelkes,
2021;Rogowski & Su he land, 2016). Con e sely, pa ies on he o he end o
he poli ical spec um can become he a ge o disgus and ange (Mason,
2018). This inc ease in a ec i e pola iza ion has p o en o be an equally
ele an de e minan o poli ical ac i ism. In he end, o e s migh no be
in e es ed in an accu a e ep esen a ion o hei policy p e e ences, bu me ely
in seeing hei poli ical opponen s loose.
In he final sec ion o his a icle I he e o e analyze su ey da a o he
same 22 Eu opean pa liamen a y sys ems included in analysis one.
6
The
su ey da a has been ha monized and cu a ed by he compa a i e s udy o
elec o al sys ems (CSES, 2020) and co e s elec ions om 1996 o 2020.
7
Fo
each coun y and elec ion, a ep esen a i e sample o he o ing age pop-
ula ion is su eyed. The analysis elies on esponses o 460505 o
480681 indi iduals, depending on he inclusion o co a ia es and missingness
o he independen a iables.
Ellge 1327
Measu ing Spa ial and A ec i e Pola iza ion in Su ey Da a. Acco ding o he
spa ial pola iza ion p oposi ion, elec o al choice is he main d i e o mo-
biliza ion. Thus, indi idual-le el measu es o spa ial pola iza ion ha e o
include some so o posi ional in o ma ion based on o e pe cep ions. A
commonly used me hod o ope a ionaliza ion is he Dal on index o pola -
iza ion, which can be adap ed o cap u e he pe cei ed ideological dispe sion
o a pa y sys em. In his case, i is calcula ed based on esponden s subjec i e
classifica ion o pa y posi ions along he le - igh dimension. Ins ead o
weigh ing posi ions by pa y o e sha es, one simply weigh s posi ions by he
sha e o esponden s indica ing o o e o each pa y. Howe e , some au ho s
deba e i pola iza ion, especially when measu ed on he indi idual le el,
should acknowledge pa y size (see Wagne , 2021). A mo e s aigh o wa d,
al e na i e measu e o pa y posi ion dispe sion is o simply calcula e he
unweigh ed s anda d de ia ion o pe cei ed pa y posi ions. I label his he
measu e o spa ial de ia ion. Bo h he spa ial de ia ion and he Dal on index
a iable cap u e o e pe cep ions abou he dispe sion o pa y p e e ences
ac oss he ideological spec um.
While he measu emen o ideological dispe sion is ela i ely uncon o-
e sial, a a ie y o ecen s udies and wo king pape s make di e ging claims
o e he ope a ionaliza ion o a ec i e pola iza ion in a mul ipa y en i-
onmen (Gid on e al., 2019;Reiljan, 2019;Wagne , 2021). Boxell e al.
(2022), o example, adop a simple measu e and me ely calcula e he ange
be ween highes and lowes eeling owa ds a pa y. While being easy o
implemen based on he so called eeling he mome e sco es, his measu e has
wo majo d awbacks. Fi s , i is e y sensi i e o he inclusion o ou lie s,
such as adical inge pa ies. In a mul ipa y con ex , he inclusion o omission
o such pa ies can se e ely al e he esul s, which could unde mine mea-
su emen alidi y. Tha aside, i also does no allow o di e en ia e posi i e
om nega i e emo ions. High sco es can be he p oduc o s ong a ec ion
owa ds one pa y, s ong dis as e owa ds a compe ing pa y, o bo h.
To ci cum en hese p oblems, I come up wi h wo measu es ha also ely
on he likabili y a ings (o eeling he mome e sco es) p o ided by CSES
(2020).Posi i e a ec is defined as he di e ence be ween he highes eeling
sco e owa ds one pa y and he a e age eeling sco e owa ds all emaining
pa ies. Thus, cons i uen s ha exp ess ex emely s ong admi a ion o one
pa y ela i e he o he s sco e high on his me ic. Nega i e a ec is con-
s uc ed in he same way, bu his ime ocuses on he ange be ween he
poo es a ed pa y and he a e age eeling owa ds all emainde s. Disce ning
among posi i e and nega i e a ec also allows me o assess whe he a ac ion
o a e sion owa ds a specific pa y d i es mobiliza ion. In o de o acili a e
meaning ul compa isons ac oss he ou di e en measu es o pola iza ion, I
decided o s anda dize hem on a scale om 0 o 10.
1328 Compa a i e Poli ical S udies 57(8)
Dependen Va iables and Fu he Con ols. Two measu es o u nou a e used as
dependen a iables. Fi s I show he e ec o pola iza ion on sel - epo ed
pa icipa ion (Pe sonal Vo e). Such sel epo s in su eys ypically su e
om social desi abili y bias, leading o unde epo ing o abs en ion. Albei
being a dicho omous ou come, I escale he a iable so ha abs en ion is
coded 0 and sel - epo ed pa icipa ion is coded 100. This is done o make
es ima es compa able o models which include he second measu e o pa -
icipa ion, which is he o ficial u nou a he elec ion (Na ional Vo e) mea-
su ed on a con inuous scale.
Due o he s uc u e o his c oss-sec ional da a, coun y and elec ion yea
fixed e ec s a e added o all analyses. Those accoun o unobse ed he -
e ogenei y ac oss uni s and o e ime, bu I also con ol o a ange o
indi idual-le el cha ac e is ics ha a e associa ed wi h poli ical p e e ences
and pa icipa ion. Fi s , I con ol o each indi idual’s spa ial cong uence,
which measu es he dis ance be ween a esponden ’s sel -placemen on he
le - igh scale and he posi ion she assigned o he closes pa y. In addi ion,
all models include sociodemog aphic cha ac e is ics such as gende , age, le el
o educa ion, and income b acke . Table A.4 in he appendix con ains de-
sc ip i e in o ma ion o all a iables o analysis h ee.
Discussion o Indi idual-Le el Resul s. Figu e 4 displays coe ficien s o each
measu e o pola iza ion de i ed om he mul i a ia e eg ession models. The
fi s ou models inside he le pane o he figu e use sel - epo ed u nou as a
dependen a iable. All o hem a e posi i e and p ecisely es ima ed, anging
om 1.5 pe cen age poin o 2.8 pe cen age poin changes in p obabili y o
pa icipa ion pe uni inc ease in pola iza ion.
8
These models confi m ha all
ou ypes o pola iza ion a e associa ed wi h highe deg ees o sel - epo ed
u nou . We also obse e ha bo h measu es o a ec i e pola iza ion ob ain
la ge es ima es han he measu es o spa ial pola iza ion.
The igh side o he plo illus a es he influence o pola iza ion on he
o ficial elec ion u nou , measu ed on a con inuous scale om 0 o 100. The
measu es o spa ial pola iza ion do no p oduce significan es ima es a all.
Now, o e pe cep ions o elec o al choice a e no associa ed wi h an inc ease
in ac ual u nou . The same is ue o posi i e a ec , o which I also find a
null es ima e. Nega i e a ec , depic ed in iole , emains a subs an ially
significan de e minan o pa icipa ion.
Wha mo e can we lea n om he su ey da a abou nega i e a ec ? In
Appendix Sec ion A.12, I explo e indi idual-le el cha ac e is ics ha co -
ela e wi h nega i e a ec . P ima ily, I find ha poli ical ex emism is as-
socia ed wi h highe deg ees o ou pa y animosi y. This is mainly d i en by
cons i uen s ha loca e hemsel es on he ex eme igh . Addi ionally, o e s
wi h high poli ical knowledge epo significan ly highe deg ees o animosi y,
while hose wi h low poli ical knowledge show less animus. This aligns wi h
Ellge 1329
p io e idence on he US (Rogowski & Su he land, 2016) and sugges s ha
hose who a e pa icula ly in e es ed and in es ed in poli ics de elop s onge
emo ional a achmen s.
Also, he esul s could be influenced by endogenous con ounde s like he
closeness o an elec ion. In Appendix Sec ion A.13, I hus show ha my
esul s hold when con olling o his elec ion-specific co a ia e ha could
ha e an influence on bo h o e a i udes and u nou .
O e all, he esul s in Figu e 4 e eal a dispa i y be ween cons i uen sel -
epo s and hei ac ual beha io on elec ion day. Wi hin he sample o
22 coun ies, u nou a na ional elec ions has only inc eased when o e s held
in ense nega i e eeling owa ds pa icula pa ies. The no ion o a a ional
o e who is s a egically choosing om an ex ended menu o poli ical fla o s
does no seem o be suppo ed by indi idual-le el da a. I animosi y becomes
he uel o engagemen , as sugges ed by he es ima e o nega i e a ec ,
pola iza ion migh no be an easy solu ion o es o e democ a ic engagemen .
Figu e 4. Reg essing u nou on pola iza ion in c oss sec ional su ey da a.
No e. The plo shows he es ima ed e ec o ou di e en measu es o pola iza ion
on wo ou come a iables: pe sonal o e, as well as elec ion u nou . Pe sonal o e is
a bina y indica o o sel - epo ed u nou , whe eas elec ion u nou measu es he
u nou a he espec i e elec ion. Bo h DVs a e scaled om 0 o 100. All ou
measu es o pola iza ion a e s anda dized on a scale om 0 o 10. Resul s a e de i ed
om models ha con ol o spa ial cong uence, age, gende , le el o educa ion, and
income b acke o he esponden . All specifica ions include coun y and yea fixed
e ec s. The plo shows s anda d e o s clus e ed by coun y a he 95% C.I.
1330 Compa a i e Poli ical S udies 57(8)
Conclusion
All esul s p esen ed in his pape poin o he ac ha pa y sys em pola -
iza ion is inc easing o e mobiliza ion in es ablished mul ipa y democ acies.
When pa ies become pola ized and pi o away om he ideological cen e ,
mo e o e s pa ake in elec ions. This pa e n pe sis s o he c oss-coun y
sample o 22 na ional elec ions o when ocusing on Ge man local elec ions.
Thus, inc easing le els o pa y sys em pola iza ion obse ed in a ious
es ablished democ acies ac oss Eu ope (see Boxell e al., 2022;Wagne , 2021)
could be conside ed an e ec i e emedy agains p e alen ends o declining
pa icipa ion. Howe e , his does no mean ha pola iza ion is ul ima ely
beneficial o democ acy. A close examina ion o su ey da a o 22 Eu opean
democ acies e eals ha nega i e a ec , he in ensi y o animosi y exp essed
owa ds he leas liked opposi ion pa y, is he mos eliable p edic o o
u nou .
Tes ing he elec o al implica ions o g owing pola iza ion in a mul ipa y
con ex is ele an since mos s udies ocus on he US unde majo i a ian ule.
As I ha e highligh ed be o e, co ec ly acing pola iza ion unde PR ules can
be mo e challenging. Gene al le els o pola iza ion a e highe (Downs, 1957;
Sa o i, 1976) and a ia ion wi hin one coun y can o igina e om a mul i ude
o ac o s (see Dal on, 2008).
Wha do hese findings imply o democ a ic poli ics mo e gene ally?
Rega ding u nou , he conclusions a e sobe ing. Pa icipa ion in elec ions has
only ecen ly eco e ed om decades o pe sis en decline (see Figu e 2). Bu
my analysis sugges s ha his ally could ha e been ueled by ising animosi y
and nega i e a ec . Looking back a ecen yea s, e en s like he e ugee c isis
o he Co id pandemic ha e beyond doub inc eased social di isions. I pa y
eli es success ully mobilize based on esen men , mo e pa y sys em pola -
iza ion and u nou a ise. This end alone can be seen as a challenge o
democ acy which impedes coope a ion and consensus-seeking. I he e o e
doub ha nega i e a ec can lead o pe sis en eco e ies o elec o al u nou .
Animus ac oss poli ical camps should u he unde mine he abili y o find
majo i ies in o de o enac policies. In PR sys ems ha ely on coope a ion
and coali ion-building, g owing nega i e a ec should be especially de i-
men al o he unc ioning o democ a ic p ocesses and a cause o g idlock. I is
ques ionable o assume ha nega i e a ec will endu e o ele a e elec o al
in ol emen i o e s become disillusioned by a di ided and dys unc ional
poli ical sys em. When esen men owa ds he democ a ic sys em g ows,
cons i uen s’animus can also spill o e in o c ime. E en in he ealm o
es ablished Eu opean democ acies, iolen c ime such as he mu de o
Labou MP Jo Cox in 2016 o he Ge man CDU poli ician Wal e Lübcke in
2019 appea o become mo e equen . While hese a e ex eme examples, less
se e e o ms o iolence, ha assmen , and in imida ion agains ep esen a i es
Ellge 1331
gained ac ion whe e e esen men is high. In addi ion, inc easing nega i e
a ec as de e minan o o e mobiliza ion could pu mode a e eli es in o an
elec o al dilemma. Fo long, hey ha e seen a public disengagemen om
poli ics as one o he mos se e e h ea s o democ a ic go e nance and
legi imacy. Bu g owing pola iza ion should mo i a e o me non- o e s wi h
ex eme ideology o engage in poli ics based on nega i e a ec owa ds he
es ablishmen (see Ez ow e al., 2014). While his would ensu e ha unhea d
oices ha e a say in poli ics, i migh also ein o ce de ision and u he
exace ba e pola iza ion.
Looking beyond socie al implica ions, his a icle also engages wi h ecen
con ibu ions o he compa a i e li e a u e on pola iza ion. Fo ins ance, i
highligh s ha schola s ha e mos ly ocused on he elec o al implica ions o
pola iza ion h ough he lens o spa ial p oximi y o ing. Pola iza ion leads o
easie dis inguishable pa y p ofiles (Lupu, 2015) and mo e choice on he
poli ical menu (C epaz, 1990;Mo al, 2017), bu hese ac o s do no ha e o be
he sole cause o mobiliza ion. We need o acknowledge ha a pola ized pa y
sys em also leads o mo e en enched poli ical iden i ies, endencies o so ing
in o insula pa isan g oups, and heigh ened le els o emo ional in ol emen
(Huddy, 2001;Iyenga e al., 2019). Such in ol emen can easily backfi e
once ac ions a e d i en by emo ions ins ead o delibe a ion (Mason, 2018).
Going o wa ds, he a ec i e componen o o e mobiliza ion has o
ecei e mo e a en ion in compa a i e esea ch. One challenge is he lack o
consensus a ound a clea concep ualiza ion and measu emen o a ec i e
pola iza ion in mul ipa y democ acies (see Boxell e al., 2022;Wagne , 2021,
o aluable con ibu ions in his ega d). Due o he di e en ways in which
pola iza ion e ol es in PR sys ems, simple adap a ions o concep s and
measu emen s om he US con ex a e limi ed. Building on exis ing wo k, I
hus p opose a no el s a egy o di e en ia e posi i e om nega i e a ec in a
compa a i e analysis. Fu u e wo k should pu mo e e o in o disen angling
di e en aspec s o pola iza ion and mo e di ec ly es he ela ionship be-
ween spa ial and ideological pola iza ion. So a , his li e a u e is domina ed
by s udies on he US and inconclusi e in answe ing whe he spa ial pola -
iza ion p ecedes a ec i e pola iza ion o ice e sa (see D uckman e al.,
2021;Lelkes, 2021;Webs e & Ab amowi z, 2017). Gi en he sys emic
di e ences in mul ipa y democ acies, compa a i e esea ch should con-
ibu e o his deba e bo h wi h new heo e ical insigh s and a b oade em-
pi ical scope.
Acknowledgmen s
I hank Heike Klü e and Ma k Kayse o supe ision and guidance. In addi ion, I am
hank ul o help ul commen s by An hea Albe o, Jeyhun Alizade, Daniel Bischo ,
Anselm Hage , Hanno Hilbig, Jochen Mülle , Sascha Riaz, Lukas S ¨
o ze , Denise
1332 Compa a i e Poli ical S udies 57(8)
T abe , Tim Wappenhans, Daniel Zibla , as well as audiences a he 2019 Be lin
Poli ical Beha io and he 2020 F ee Uni e si y-WZB Wo kshops.
Decla a ion o Conflic ing In e es s
The au ho decla ed no po en ial conflic s o in e es wi h espec o he esea ch,
au ho ship, and/o publica ion o his a icle.
Funding
The au ho ecei ed no financial suppo o he esea ch, au ho ship, and/o publi-
ca ion o his a icle.
ORCID iD
Fabio Ellge h ps://o cid.o g/0000-0002-7179-7139
Da a A ailabili y S a emen
Replica ion code and da a a e a ailable on he Compa a i e Poli ical S udies Da a-
e se: h ps://doi.o g/10.7910/DVN/9Z1QK6 (see Ellge , 2023).
Supplemen al Ma e ial
Supplemen al ma e ial o his a icle is a ailable online.
No es
1. Replica ion ma e ials o all empi ical analyses can be ound a Ellge (2023).
2. Coun ies wi h compulso y o ing laws, such as Belgium, G eece and Luxembou g,
a e omi ed om he analysis. The same is ue o Swi ze land, whe e ci izens
adi ionally pa icipa e in poli ics by means o di ec democ acy and elec o al
pa icipa ion is no o iously low.
3. No e ha Bulga ia and Romania joined in 2007, while C oa ia only joined in 2013. I
s ill conside elec ions s a ing in 2005 o hese coun ies.
4. 1s qua ile: 9,9 and 3 d qua ile: 19,1. See Appendix Table A.1 o mo e desc ip i e
s a is ics.
5. Wo ms (RP), Guenzbu g (BY), Ros ock (MP), and Schwe in (MP).
6. Aus ia, Bulga ia, C oa ia, Czech Republic, Denma k, Es onia, Finland, Ge many,
Hunga y, Iceland, I eland, La ia, Li huania, Ne he lands, No way, Poland, Po -
ugal, Romania, Slo akia, Slo enia, Spain, Sweden.
7. The sample is based on he CSES ‘in eg a ed module da ase ’(1996–2016), o
which I ha e manually me ged he mos ecen module 5 (2016–2020).
8. Reg ession ables o Analysis 3 can be ound in Appendix Table A.15.
Ellge 1333
