Compliance in the public versus the private realm: Economic preferences, institutional trust and COVID‐19 health behaviors

S e nbe g, Hen ike; S eine , Janina Isabel; Bü he, Tim
A icle — Published Ve sion
Compliance in he public e sus he p i a e ealm:
Economic p e e ences, ins i u ional us and COVID‐19
heal h beha io s
Heal h Economics
P o ided in Coope a ion wi h:
John Wiley & Sons
Sugges ed Ci a ion: S e nbe g, Hen ike; S eine , Janina Isabel; Bü he, Tim (2024) : Compliance in
he public e sus he p i a e ealm: Economic p e e ences, ins i u ional us and COVID‐19 heal h
beha io s, Heal h Economics, ISSN 1099-1050, Wiley, Hoboken, NJ, Vol. 33, Iss. 5, pp. 1055-1119,
h ps://doi.o g/10.1002/hec.4807
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/293972
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 p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Recei ed: 4 Ap il 2022
-
Re ised: 1 Decembe 2023
-
Accep ed: 6 Janua y 2024
DOI: 10.1002/hec.4807
RESEARCH ARTICLE
Compliance in he public e sus he p i a e ealm:
Economic p e e ences, ins i u ional us and COVID‐19
heal h beha io s
Hen ike S e nbe g
1,2,3
|Janina Isabel S eine
1,3,4
|Tim Bü he
1,2,3,5
1
TUM School o Social Sciences and
Technology, Technical Uni e si y o
Munich, Munich, Ge many
2
TUM School o Managemen , Technical
Uni e si y o Munich, Munich, Ge many
3
Munich School o Poli ics and Public
Policy (H P), Technical Uni e si y o
Munich, Munich, Ge many
4
TUM School o Medicine and Heal h,
Technical Uni e si y o Munich, Munich,
Ge many
5
San o d School o Public Policy, Duke
Uni e si y, Du ham, No h Ca olina, USA
Co espondence
Hen ike S e nbe g, Depa men o
Go e nance, School o Social Sciences and
Technology, Technical Uni e si y o
Munich, Richa d‐Wagne ‐S asse 1,
Munich 80333, Ge many.
Email: [email p o ec ed]
Funding in o ma ion
Ho izon 2020 F amewo k P og am,
G an /Awa d Numbe : 101016233
PERISCOPE
Abs ac
To wha ex en do economic p e e ences and ins i u ional us p edic
compliance wi h physical dis ancing ules du ing he COVID‐19 pandemic?
We eexamine his ques ion by in oducing he heo e ical and empi ical
dis inc ion be ween indi idual heal h beha io s in he public and in he p i-
a e domain (e.g., keeping a dis ance om s ange s s. abs aining om p i-
a e ga he ings wi h iends). Using s uc u al equa ion modeling o analyze
su ey da a om Ge many's second wa e o he pandemic (N=3350), we
e eal he ollowing majo di e ences be ween compliance in bo h domains:
Social p e e ences, especially (posi i e) ecip oci y, play an essen ial ole in
p edic ing compliance in he public domain bu a e ba ely ele an in he
p i a e domain. Con e sely, indi iduals' deg ee o us in he na ional go -
e nmen ma e s p edominan ly o inc easing compliance in he p i a e
domain. The clea ly s onges p edic o in his domain is he pe cep ion
pandemic‐ ela ed h ea s. Ou indings encou age ailo ing communica ion
s a egies o ei he domain‐speci ic ci cums ances o ac o s common ac oss
domains. Tailo ed communica ion may also help p omo e compliance wi h
o he heal h‐ ela ed egula o y policies beyond COVID‐19.
KEYWORDS
compliance, COVID‐19, economic p e e ences, heal h beha io , ins i u ional us , physical
dis ancing
JEL CLASSIFICATION
D91, H12, H31, I12, I18
1
|
INTRODUCTION
Wha d i es indi idual compliance wi h no ms, s anda ds, and impe ec ly moni o ed laws and egula ions? The
impo ance o his ques ion has long been ecognized o gene al public policy con ex s such as ax o a e a oidance as
well as o heal h policy con ex s such as accina ion manda es. Mo e ecen ly, he ques ion has gained u he
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion‐NonComme cial‐NoDe i s License, which pe mi s use and dis ibu ion in any me-
dium, p o ided he o iginal wo k is p ope ly ci ed, he use is non‐comme cial and no modi ica ions o adap a ions a e made.
© 2024 The Au ho s. Heal h Economics published by John Wiley & Sons L d.
Heal h Economics. 2024;33:1055–1119. wileyonlinelib a y.com/jou nal/hec
-
1055
signi icance o COVID‐19‐ ela ed physical dis ancing—a con ex in which indi idual beha io has a e y high appa en
socie al ele ance, bu he indi idual and collec i e sho ‐and long‐ e m consequences o non‐compliance a e ela i ely
unce ain.
1
These cha ac e is ics pose a pa icula challenge o policy make s because hey imply ha a subs an ial
amoun o a iance in compliance beha io s may no be exclusi ely d i en by ully in o med, a ional cos ‐bene i
conside a ions. Ins ead, p e ious esea ch sugges s ha compliance beha io in such con ex s is shaped o a signi i-
can ex en by indi iduals' economic p e e ences (i.e., social, isk and ime p e e ences) and hei deg ee o us in he
ins i u ions endo sing he ules (e.g., Ba gain & Aminjono , 2020; Campos‐Me cade, Meie , Schneide , Meie ,
e al., 2021; Campos‐Me cade, Meie , Schneide , & Wengs öm, 2021; Chan, Skali, e al., 2020; Cucciniello e al., 2022;
Kese & Rau, 2023; Shim e al., 2012; Su inen & Kupe an, 1999).
This pape p oposes and empi ically in es iga es a he e o o e unadd essed implica ion eme ging om he p o-
nounced in luence o hese ac o s on compliance pa e ns: Compliance beha io may sys ema ically a y be ween he
public and he p i a e domain, induced by he di e en ial impac o indi iduals' economic p e e ences and ins i u ional
us on compliance in hese wo domains. This is highly ele an in he con ex o beha io al s ipula ions o con ain he
sp ead o COVID‐19, which ha e included ules go e ning people's beha io in public spaces, such as equi emen s o
wea masks and main ain physical dis ance om o he s, as well as ules go e ning ela i ely p i a e beha io s, such as
limi s o he numbe o iends wi h whom o mee a home. While compliance in bo h he public and p i a e domain is
c ucial o achie ing he o e a ching objec i e o hese ules, he wo domains di e ega ding he audiences who migh
obse e and en o ce compliance, sugges ing po en ially di e ing incen i e s uc u es.
Agains his backg ound, we examined whe he such a di e gence in compliance beha io exis s in he con ex o
COVID‐19‐ ela ed physical dis ancing ules. Compliance in he public domain he e comp ised ac ing in con o mi y
wi h heal h guidelines in ended o go e n beha io s ha a e easily obse able by membe s o he gene al public,
including public au ho i ies. Examples include wea ing a acemask o keeping a physical dis ance om people om
o he households in public spaces. Compliance in he p i a e domain comp ised beha io consis en wi h heal h
guidelines in ended o go e n mo e p i a e decisions abou es ic ing social con ac s and mobili y al oge he , which
is o a la ge ex en obse ed only by hose who also ail o comply. Recognizing and examining a po en ial di e -
gence in compliance be ween hese wo domains is impo an o ad ancing ou heo e ical unde s anding o
compliance in gene al and mo eo e highly ele an o public policy in he con ex o COVID‐19. Gi en he lowe
COVID‐19 accine access and co e age in coun ies o he Global Sou h and he p ospec o eme ging highly
con agious i us a ian s, lockdown and physical dis ancing manda es emain c ucial ools o con aining in ec ion
a es in such scena ios.
To assess he ex en o which economic p e e ences and ins i u ional us migh di e in hei abili y o p edic
heal h beha io s in he public e sus in he p i a e domain, we es ima ed sepa a e s uc u al equa ion models (SEMs) o
sel ‐ epo ed compliance wi h na ionwide issued physical dis ancing ules, using o iginal su ey da a om Ge many's
second wa e o he pandemic in he win e o 2020/21 (N=3350). As o economic p e e ences, we examined isk
a e sion, pa ience, ecip oci y, al uism and ci ic esponsibili y. As o ins i u ional us , we conside ed COVID‐19‐
ela ed us in he go e nmen and in scien i ic ins i u ions.
Ou esul s con i m ha compliance is signi ican ly co ela ed wi h indi iduals' social and isk p e e ences and hei
ins i u ional us . This inding holds when con olling o COVID‐19 h ea pe cep ion, which was e ealed as he
s onges p edic o o compliance in bo h domains.
Mo e impo an ly, ou su ey da a e ealed ha beha io al pa e ns di e signi ican ly ac oss he wo compliance
domains in h ee ways. Fi s , esponden s' le el o posi i e ecip oci y was o g ea impo ance o compliance in he
public domain bu ba ely ele an in he p i a e domain. The same (bu sligh ly weake ) domain‐speci ic di e ences
eme ged o nega i e ecip oci y. In e es ingly, co ela ions be ween ecip oci y and compliance we e posi i e in he
case o posi i e ecip oci y and nega i e in he case o nega i e ecip oci y. Second, we also ound domain‐speci ic
pa e ns o he deg ee o us in he na ional go e nmen and us in scien i ic ins i u ions: While us in he go -
e nmen ma e ed only o inc easing compliance in he p i a e domain, us in scien i ic ins i u ions was an impo an
ac o in bo h domains, bu signi ican ly mo e so in he public domain. Thi d and mo e gene ally, ou esul s sugges
di e ences ac oss domains in he ela i e impo ance o COVID‐19 h ea pe cep ions is‐à‐ is p e e ences and us .
Speci ically, he dominance o COVID‐19 h ea pe cep ions as he p ima y p edic o o compliance was signi ican ly
mo e p onounced in he p i a e domain. In con as , indi iduals' social p e e ences we e mo e s ongly associa ed wi h
compliance in he public domain.
This analysis con ibu es o he li e a u es on he p edic o s o indi idual‐le el ou side‐ he‐lab ule and no m
compliance in economics, law, poli ical science and psychology in a a ie y o con ex s, including heal h beha io s
1056
-
STERNBERG ET AL.
du ing epidemics o pandemics (e.g., Algan e al., 2021; Blai e al., 2017; Böhm e al., 2016; B odeu , G ay, e al., 2021;
Galizzi e al., 2022). Fu he mo e, we con ibu e o he mo e speci ic and ecen ly eme ging li e a u e on heal h be-
ha io s in imes o COVID‐19. In his li e a u e, social p e e ences, isk p e e ences, ( o a lesse ex en ) ime p e e -
ences, ins i u ional us , and pandemic‐ ela ed h ea pe cep ions ha e been iden i ied bo h heo e ically and
empi ically as impo an p edic o s o a ious ypes o compliance beha io s as well as mobili y pa e ns.
2
To he bes o
ou knowledge, his is he i s pape o in oduce—bo h in he gene al as well as in he mo e speci ic COVID‐19
compliance li e a u e— he concep ual dis inc ion be ween compliance in he public and he p i a e domain, iden i y
he implica ions o how compliance migh be linked o indi iduals' economic p e e ences and ins i u ional us in
dis inc ways ac oss he wo domains, and sys ema ically examine hese po en ial di e ences empi ically. Ou indings
sugges ha he same indi idual may exhibi di e en deg ees o compliance ac oss hese wo domains. They also imply
ha he e ec i eness o policies aimed a spu ing compliance will a y ac oss domains—o pu di e en ly: dis inc
policies migh be needed o spu compliance in each domain.
These policy implica ions o ou indings a e impo an o heal h policy well beyond beha io al s ipula ions du ing
a pandemic, which is he con ex ha allowed us o examine he issue. Fo ins ance, communica ion s a egies aimed a
imp o ing public heal h h ough en i onmen ally conscious consume beha io migh seek o encou age he pu chase
o p oduc s wi h a low ca bon oo p in a local s o es, which we would classi y as compliance in he public domain,
because i is easily obse ed by ellow ci izens. Con e sely, communica ion s a egies migh a ge consume beha io
wi h ega ds o online pu chases, which may be classi ied as compliance in he p i a e domain. Quan i ying he ex en
and analyzing he dynamics o a po en ial compliance di e gence in his con ex would be highly ele an o designing
e ec i e en i onmen al policies.
The emainde o his pape is o ganized as ollows. Sec ion 2cha ac e izes he wo compliance domains and a -
icula es expec a ions o (di e en ial) impac s o indi iduals' economic p e e ences and ins i u ional us . Sec ion 3
desc ibes he su ey design and he empi ical s a egy, ha is, he SEMs. Sec ion 4p esen s he main esul s and
obus ness checks. Sec ion 5discusses he b oade signi icance and policy implica ions.
2
|
THEORETICAL CONSIDERATIONS
2.1
|
Cha ac e iza ion o compliance domains
The exis ing li e a u e on (COVID‐19‐ ela ed) compliance does no dis inguish be ween he public and p i a e domain
as spu ing dis inc logics o compliance. We now u n o his dis inc ion.
Physical dis ancing ules du ing he COVID‐19 pandemic ha e in nume ous coun ies called o limi ing social
con ac s and mobili y in a a ie y o ways o educe he isk o sp eading he in ec ion. Some o hese ules p edom-
inan ly go e n beha io ha inhe en ly akes place in he public sphe e, such as equi emen s o no ms o, o example,
wea a mask o keep a ce ain dis ance o pe sons om o he households in public anspo , a es au an s, in a public
pa k, e c. Viola ions o hese ules a e easily obse ed (and hence en o ceable), including by go e nmen au ho i ies and
by complian ellow ci izens.
O he ules go e n beha io ha p edominan ly akes place in he p i a e sphe e, o example, ules asking ci izens
o es ic p i a e ga he ings o a maximum o wo households o o only lea e he house o necessa y daily e ands and
o he u gen easons. We e e o decisions abou complying o iola ing hese ules as compliance in he p i a e
domain. Non‐compliance wi h such ules lea ing he house o isi iends o un ins ead o lea ing he house only o
ge g oce ies, o a ending o hos ing a dinne pa y wi h iends om 10 di e en households is no easily obse able.
Mo eo e , i is mos eadily obse ed by indi iduals who ha e also chosen no o comply wi h he es ic ions ( he
iends who hemsel es a end he dinne pa y).
The wo domains hus a y in e ms o he obse abili y o compliance beha io s o ce ain audiences. This esul s
in di e ences wi h ega d o (i) he isk o o mal (i.e., s a e) punishmen o non‐complian beha io and (ii) he
likelihood o social punishmen by ellow ci izens. Fu he , he wo domains a y by (iii) he deg ee o social closeness
o he people ha seem mos immedia ely a ec ed by (non‐)compliance (in e ms o he medical isk o ge ing in ec ed
wi h COVID‐19). Impo an ly, no e ha al hough compliance decisions ac oss he wo domains may in p ac ice be
co ela ed, hey a e logically o hogonal in he sense ha compliance in any one ealm could be p ac iced ega dless o
whe he one complied wi h he ules o he espec i e o he domain.
STERNBERG ET AL.
-
1057
As in (i), he isk o o mal punishmen , o example, ge ing ined o non‐compliance, was inhe en ly highe in he
public han in he p i a e domain. Fo ins ance, mask‐wea ing was in many places moni o ed h ough an inc eased
police p esence in subways o c owded ci y cen e s, whe eas la ge ‐ han‐allowed ga he ings in he p i acy o a home
was subjec only o he much smalle isk o epo s by p oac i e neighbo s. Thus, while he amoun o ines a he ime
o da a collec ion was highe o non‐compliance in he p i a e domain, he isk o ac ually ge ing ined was highe in
he public domain (see Table A30). To ha end, a ecen s udy sugges s ha he impac o economic p e e ences may be
sensi i e o he exis ence o go e nmen en o cemen /punishmen in he o m o ines, which u he s eng hens he
a ionale o he sugges ed public‐p i a e dis inc ion (Papanas asiou e al., 2022).
As in (ii), in e ms o social punishmen by ellow ci izens, he ype o audience o po en ially execu e such a
punishmen di e s be ween bo h domains. While in he p i a e ealm, obse able non‐compliance is subjec ed o
disapp o al by one's close pee s, non‐compliance in he public ealm is widely obse ed by he gene al public. One may
a gue ha social incen i es o comply in p i a e se ings may o his eason be in p inciple e y s ong (see also (iii)
below). Howe e , as highligh ed abo e, in con as o he public domain, compliance in he p i a e domain is di ec ly
obse ed mos ly by o he s who a e also non‐complian . Consequen ly, he likelihood (no necessa ily he se e i y) o
social punishmen is also assumed o be lowe in he p i a e han in he public domain.
As in (iii), he deg ee o social closeness o he people ha seem mos immedia ely a ec ed by (non‐)complian
beha io (in e ms o he isk o an in ec ion) is highe in he p i a e han in he public domain. O cou se, a lack o
compliance wi h he ules in ei he domain can cause a close amily membe o iend o ge in ec ed h ough i us
ansmission. Howe e , his isk is much mo e salien in compliance beha io s in he p i a e domain, whe e one
di ec ly decides abou whe he o mee wi h amily and iends. Apa om ha , his hi d dis inc i e cha ac e is ic
also sugges s ha he pe sonal dilemma o whe he o comply o no is mo e subs an ial in he p i a e domain:
Complying means p o ec ing one's closes iends/ amily bu also no being able o main ain close social con ac and
suppo hem.
2.2
|
Logics o compliance: Dis inc i e decision‐making logics ac oss domains
In he ollowing, we i s p esen heo e ically and empi ically in o med expec a ions on how economic p e e ences and
ins i u ional us may a ec COVID‐19 compliance o e all be o e we hen elabo a e on how we expec dynamics o
di e ac oss he wo compliance domains.
2.2.1
|
Ci ic esponsibili y
Compliance is likely mo e p onounced among indi iduals wi h a highe sense o ci ic esponsibili y, as he ac o
complying wi h physical dis ancing egula ions du ing a pandemic esembles an ac o ci ic esponsibili y (e.g., Ba ios
e al., 2021). Rega ding di e en ial dynamics ac oss domains, expec a ions a e con lic ing: On he one hand, ci ic
esponsibili y may be a mo e ele an d i e o compliance in he public domain, which is he ealm ha social o ci ic
du ies a e mainly associa ed wi h. On he o he hand, ci ic esponsibili y can be iewed as an in e nalized, in insic
mo i a ion o compliance, which migh hus be mo e impo an in he p i a e domain, ha is, in he absence o o mal
en o cemen .
2.2.2
|
Posi i e and nega i e ecip oci y
Indi iduals' le el o posi i e and nega i e ecip oci y may also a ec compliance beha io : Speci ically, in an en i-
onmen in which compliance is gene ally high, a pe son wi h highe le els o posi i e ecip oci y (i.e., a s onge
willingness o e u n a a o ) should exhibi a highe deg ee o compliance because compliance by o he s also p o ec s
his pe son and hus may be pe cei ed as a a o o him o he (e.g., Nikolo e al., 2020). In con as , nega i e eci-
p oci y ( he willingness o punish an isocial beha io ) should in expec a ion no a ec one's own le el o compliance,
because non‐compliance as an a emp ed punishmen would, in he pandemic con ex , also punish hose indi iduals
who con ibu e o he public good (i.e., complian indi iduals). Howe e , one could a gue ha non‐complian in-
di iduals migh be punishable o a highe deg ee by one's own ac o non‐compliance because compliance also yields
1058
-
STERNBERG ET AL.

sel ‐p o ec ion om he i us. This would sugges ha indi iduals wi h highe le els o nega i e ecip oci y comply
ela i ely less wi h he imposed ules (e.g., Al a o e al., 2022). In e ms o di e en ial dynamics ac oss he wo domains,
a pe son's deg ee o (posi i e o nega i e) ecip oci y should ma e mo e o compliance in he public domain. He e,
compliance beha io s a e much mo e exposed o and obse ed by po en ial ecip oca o s han in he p i a e domain,
ha is, whe he a pe son wea s a mask on he ain o in he supe ma ke is obse ed by a highe numbe o indi iduals
han whe he a pe son s ays a home alone and e ains om mee ing wi h iends.
2.2.3
|
Al uism
We gene ally expec indi iduals wi h highe le els o al uism o exhibi highe compliance wi h COVID‐19 ules
because hose ules aim a educing he sp ead o ha m ul in ec ions among ellow ci izens (e.g., Nikolo e al., 2020;
Quaas e al., 2021). Pu e al uism should no ha e any di e en ial e ec s ac oss he wo compliance domains, since
pu e al uism e e s o in insic alues and does no include any ecip oca ed dynamics o incen i es. As long as
compliance in he public and p i a e domain mo e o less equally helps o educe he sp ead o he i us, highe le els
o pu e al uism should inc ease compliance ega dless o whe he i is ela i ely easily obse ed by o he s o no .
Howe e , gi en he pe sonal dilemma indi iduals ace in e ms o compliance in he p i a e domain, al uism could
also ha e opposing e ec s in his domain: Al uism migh no only call o p o ec ing o he s om he medical con-
sequences o he i us, bu also om he social consequences, ha is, social isola ion.
2.2.4
|
Risk p e e ences
Risk‐a e se indi iduals a e expec ed o comply o a la ge deg ee wi h physical dis ancing ules han indi iduals who
a e mo e isk‐accep ing o isk‐seeking because highe compliance lowe s he isk o ge ing in ec ed, as well as he isk
o ge ing ined o non‐compliance (e.g., Mülle & Rau, 2021; Papanas asiou e al., 2022). Rega ding di e en ial e ec s
ac oss domains, we do no ha e s ong expec a ions, gi en ha non‐compliance in bo h domains can be cha ac e ized
as isky beha io , only conce ning di e ing aspec s (e.g., he isk o punishmen o non‐compliance s. he isk o
passing on an in ec ion o close iends o amily membe s).
2.2.5
|
Time p e e ences
We migh expec mo e pa ien indi iduals (i.e., wi h lowe discoun a es) o exhibi highe le els o compliance as hey
a e mo e willing o sac i ice a ce ain immedia e ewa d (e.g., mee ing wi h iends) o a la e la ge ewa d (e.g., he
end o con ac es ic ions al oge he ) (e.g., Al a o e al., 2022; Papanas asiou e al., 2022). We do no ha e s ong
expec a ions ega ding di e en ial e ec s ac oss he wo domains.
2.2.6
|
Ins i u ional us
The go e nmen and scien i ic ins i u ions ac ed as key endo se s o he social‐dis ancing ules imposed du ing he
COVID‐19 pandemic. The e o e, we expec ha highe le els o COVID‐19‐ ela ed us in go e nmen al o scien i ic
ins i u ions spu compliance wi h physical dis ancing ules, which is in line wi h wha ecen empi ical e idence
sugges s (e.g., Ba gain & Aminjono , 2020; B odeu , G igo ye a, & Ka an, 2021). Fo us in he go e nmen , con-
lic ing logics make he di e ence be ween he public and p i a e ealms heo e ically inde e mina e. On he one hand,
us in he go e nmen migh ha e a mo e p onounced posi i e e ec in he p i a e domain, gi en ha lowe le els o
moni o ing and en o cemen by s a e au ho i ies make us in he go e nmen as an in insic mo i a o mo e
impo an . On he o he hand, an unde s anding o he p i a e domain as a ealm in which he go e nmen has
inhe en ly no legi ima e ole o play migh make us in he go e nmen less ele an as a p edic o . Fo us in
science, we do no ha e p onounced di e en ial expec a ions, hough one migh a gue ha i s ele ance should be
s onge in he public domain gi en he mo e echnical‐scien i ic na u e o he s ipula ions in his ealm, o example,
wea ing a mask o keeping a 1.50 m dis ance om ano he .
STERNBERG ET AL.
-
1059
3
|
MATERIAL AND METHODS
3.1
|
S udy se ing and sampling
The s udy was conduc ed as an online su ey be ween Feb ua y 3 and Ma ch 3, 2021, du ing he second
na ionwide COVID‐19 lockdown in Ge many, which had begun in No embe 2020. The only s o es ully ope a ing
a his ime we e hose o daily necessi ies and medical supplies, while es au an s, e ail s o es and he like
ope a ed a mos on a ake‐away o deli e y basis. Physical dis ancing ules o he second na ionwide lockdown
we e pu in place ea ly and we e epea edly enewed, ha is, hey emained unchanged du ing he en i e s udy
pe iod and we can expec he as majo i y o he popula ion o ha e been awa e o hei exis ence.
3
The na ional
and s a e go e nmen s me o discuss po en ial changes in he na ional lockdown s a egy on Ma ch 3, which
ma ks he end o he da a collec ion.
The sample consis ed o 3350 esponden s ec ui ed om a Ge man access panel main ained by he su ey company
Respondi. Indi iduals we e eligible o pa icipa e in he s udy i hey we e a leas 18 yea s old and epo ed ha hey
had spen he majo i y o he las 2 weeks in Ge many. Quo a sampling was used o ob ain a ep esen a i e sample o
he Ge man popula ion wi h ega d o (i) gende , (ii) age g oup, (iii) educa ion, and (i ) s a e. Responden s ecei ed
“mingle poin s” (wo h be ween h ee o i e Eu os) o pa icipa ing in he s udy, which hey could edeem in he o m
o cash, ouche s, o dona ions.
3.2
|
Su ey design and ou come a iables
The su ey was designed o collec in o ma ion on he main a iables o in e es o his s udy, namely e-
sponden s' compliance wi h na ional physical dis ancing ules in Ge many as well as hei economic p e e ences
and hei le el o ins i u ional us . We also collec ed in o ma ion abou an al e na i e highly ele an p edic o
o compliance, namely COVID‐19 h ea pe cep ion, which has been shown o a ec (COVID‐19) heal h beha io s
(e.g., Jø gensen e al., 2021; Kluwe‐Schia on e al., 2021; Papanas asiou e al., 2022; Plohl & Musil, 2021), and
may be co ela ed wi h p e e ences and us . The su ey mo eo e collec ed in o ma ion on a numbe o
addi ional explana o y a iables, including esponden s' demog aphic and socioeconomic cha ac e is ics, poli ical
and ideological ac o s, knowledge abou he e icacy o di e en p e en ion measu es o educe he sp ead o
COVID‐19, and a scale o assess possible social desi abili y bias (Kempe e al., 2012). Tables A1 and A2 in
Appendix A summa ize all su ey i ems employed in his pape (Table A21 e eals he exac o mula ion o he
su ey i ems as displayed o esponden s, ansla ed om he o iginal Ge man e sion). The ull ques ionnai e,
he p e‐analysis plan and he a ionale o de ia ions om he la e can be e ie ed om he supplemen a y
ma e ial.
3.2.1
|
Elici a ion o compliance beha io s
Compliance was elici ed by asking esponden s abou hei beha io in six si ua ions go e ned by a ious physical
dis ancing ules issued by he Ge man na ional go e nmen . In each case, esponden s we e asked o a e on a scale
om 1 o 5 he ex en o which hei own beha io in he pas 2 weeks e lec ed hese beha io s ( anging om ne e o
always).
Following he heo e ical conside a ions abo e, h ee ques ions we e in ended o p ima ily elici in o ma ion
abou compliance in he public domain, asking esponden s abou (i) wea ing a mask in public anspo o when
shopping, (ii) keeping he go e nmen ‐s ipula ed dis ance o app oxima ely 1.50 m in public spaces, and (iii) a oiding
handshakes when g ee ing o he people. Ano he h ee ques ions we e p ima ily in ended o elici in o ma ion abou
compliance in he p i a e domain, asking esponden s abou (i ) lea ing he home only when uly necessa y, ( )
es ic ing p i a e mee ings o he go e nmen ‐s ipula ed limi o one pe son om one o he household, and ( i)
minimizing in e ac ions wi h pe sons om ou side one's own households in gene al.
4
Impo an ly, he Ge man
ede al go e nmen 's s ipula ions—and hence he equi emen s o complian beha io —we e equally clea and
s able o he p i a e and o he public domain du ing he da a ga he ing phase.
1060
-
STERNBERG ET AL.
3.2.2
|
Elici a ion o economic p e e ences and ins i u ional us
Wi h ega ds o economic p e e ences, we elici ed esponden s' le el o al uism, posi i e and nega i e ecip oci y, isk
a e sion, pa ience and ci ic esponsibili y. To cap u e ins i u ional us , we elici ed hei COVID‐19‐ ela ed us in he
na ional go e nmen and in he Robe ‐Koch‐Ins i u e (RKI), he la e as a p oxy o us in science.
Fo he majo i y o hese ac o s (al uism, posi i e and nega i e ecip oci y, isk a e sion, pa ience), we adap ed he
i ems and measu emen p ocedu e om he Ge man e sion o he Global P e e ence Module (Falk e al., 2022,2018):
We used bo h (i) a i udinal measu es ha ask abou gene ally beha ing in a ce ain way, and (ii) ac ual incen i ized
choices (such as dona ion decisions in he case o al uism o lo e y pa icipa ion in he case o isk a e sion). These
su ey i ems we e s anda dized and hen used o cons uc one inal measu e o each p e e ence, based on he weigh s
o he su ey i ems ha eme ged om he expe imen al alida ion p ocedu e by Falk e al. (2018, p. 1653).
5
The su ey
i ems, weigh s, inal p e e ence measu es, and gene al p ocedu e a e summa ized in Tables A4–A6 and Figu e A5 in
Appendix A.
6
Ci ic esponsibili y was measu ed using (i) esponden s' epo ed o e u n‐ou in he las na ional elec ion, (ii)
hei sel ‐ epo ed endency (no ) o e ade a es in public anspo , and (iii) hei sel ‐ epo ed endency (no ) o li e .
Responses o hese h ee i ems we e used o es ima e a ac o sco e o esponden s' unde lying le el o ci ic e-
sponsibili y, which we assumed o be he p ima y common ac o among hese indica o s ( o a simila app oach, see,
e.g., Mülle and Rau (2021)).
Ins i u ional us was elici ed h ough ques ions abou esponden s' deg ee o us in he Ge man na ional go -
e nmen 's and he RKI's abili y o manage he pandemic si ua ion. These wo ins i u ions we e he p ima y endo se s o
physical dis ancing ules and he main sou ces o o icial public heal h communica ions du ing he pandemic in
Ge many. Th oughou he pandemic, he RKI has been he mos widely known and ecognized Ge man na ional‐le el
scien i ic body o conduc epidemiological and medical analyses o COVID‐19 and o issue policy ecommenda ions. I
hus se ed as a p oxy o us in science in he Ge man con ex (Be sch e al., 2021b).
3.2.3
|
Elici a ion o COVID‐19 h ea pe cep ion
COVID‐19 h ea pe cep ion was cap u ed using a ba e y o ques ions abou (i) how h ea ening esponden s pe cei ed
he COVID‐19 pandemic o be in gene al and (ii) how h ea ening hey pe cei ed i o be wi h ega d o speci ic aspec s
o hei li es, including hei own heal h o he heal h o hose close o hem, hei inancial si ua ion and hei social
li es. These i ems we e used o es ima e ac o sco es cap u ing esponden s' unde lying COVID‐19 h ea pe cep ions
o be hen employed in he subsequen analysis. The majo i y o he i ems we e adap ed om Be sch e al. (2021a).
3.3
|
Da a collec ion and p ocessing
The su ey was p og ammed in Ge man using Qual ics and pilo ed wi h 150 pa icipan s. The ec ui men o he inal
su ey was conduc ed by he su ey company Bilendi.
7
Analyses we e pe o med in R ( e sion 4.1.0) and STATA17.
In o med consen was ob ained om all esponden s be o e hey we e p esen ed wi h he ques ionnai e, which hey
could in e up o exi a any ime. As pa o he deb ie ing upon comple ion o he su ey, pa icipan s we e p o ided
wi h a subs an i e lis o esou ces o help and in o ma ion sou ces abou he COVID‐19 pandemic as well as men al
heal h suppo se ices.
3.4
|
Empi ical s a egy
The empi ical s a egy comp ised essen ially wo s eps. Fi s , we examined compliance ac oss he six di e en physical
dis ancing beha io s o asce ain o wha ex en he e is empi ical e idence o he exis ence o he concep ual
dis inc ion be ween compliance in he p i a e and in he public domain. In iew o his, we employed explo a o y ac o
analyses o iden i y he subse s o physical dis ancing beha io s ha e lec compliance in each domain and hen
de i ed ini ial es ima es o compliance in he public and p i a e domain, espec i ely.
STERNBERG ET AL.
-
1061
Second, we in es iga ed by means o SEM echniques o wha ex en compliance pa e ns a e co ela ed wi h in-
di iduals' economic p e e ences and ins i u ional us o each o he wo domains o compliance. SEMs help o educe
measu emen e o in he unde lying la en a iable(s) o in e es —he e compliance beha io —by combining pa h
analysis ( he s uc u al componen o he SEM) wi h con i ma o y ac o analysis ( he measu emen componen o he
SEM) (Acock, 2013). In ou case, he SEM simul aneously (i) i s a con i ma o y ac o analysis ha cap u es
compliance in he public/p i a e domain as a la en a iable and (ii) es ima es e ec s o p e e ences and us on his
la en measu e o compliance.
The con i ma o y ac o analysis (i.e., he measu emen componen o he SEM) is de ined as ollows o compliance
in each domain d, whe e d={public, p i a e}.
y0
d¼λ0
dCdþe0
dψd;ð1Þ
y0
din Equa ion (1) deno es a ec o o he subse o he six obse ed compliance i ems ha e lec he espec i e
compliance domain, using he esul s om he explo a o y ac o analyses in he i s s ep (see Sec ion 4.1). C
d
deno es
he iden i ied la en measu e o domain‐speci ic compliance (i.e., he common ac o wi hin each i em subse ), and λ0
dis
a ec o o he eg ession coe icien s o he model, ha is, he ac o pa e n coe icien s (loadings) o he obse ed
i ems o hei espec i e compliance domain. e0
dco espond o he unobse ed unique ac o s o he six compliance
i ems and ψ
d
a e he coe icien s ela ing he unique ac o s o he i ems. The a iables o in e es he e a e he ac o s
cap u ing compliance in bo h domains, C
d
, which a e assumed o induce obse ed esponses o he espec i e subse o
he six compliance i ems. The la e , y0
d, a e he e o e he dependen a iables in he measu emen model and cons i u e
he e lec i e indica o s o compliance in each domain (Acock, 2013).
The s uc u al componen o he SEM eg esses compliance beha io s in each domain (i.e., he la en a iables o
compliance) on economic p e e ences and ins i u ional us , and is de ined as ollows.
Cd¼P0αdþγdTþX0ηdþZ0ζdþυdð2Þ
P0deno es a ec o o he eg ession coe icien s o p e e ences and us on he la en measu e o domain‐speci ic
compliance (C
d
).γddeno es he eg ession coe icien o COVID‐19 h ea pe cep ion (T) as an al e na i e p edic o
o compliance in each domain. X0is a ec o o demog aphic and socioeconomic con ols, namely gende , age g oup,
s a e, educa ion, employmen in essen ial se ices, household size and income, and Z0is a ec o o speci ic compliance
con ols, namely knowledge abou COVID‐19 p e en i e measu es and he deg ee o social desi abili y bias. υddeno es
he e o e m. Indi idual subsc ip s a e supp essed o simplici y.
The measu emen componen and he s uc u al componen o he SEM a e connec ed h ough he la en a iable,
ha is, compliance in he public/p i a e domain, espec i ely, allowing us o simul aneously es ima e he abo e
equa ions. We es ima ed he SEM sepa a ely o each compliance domain using Diagonal Weigh ed Leas Squa es on a
polycho ic co ela ion ma ix while gene a ing obus s anda d e o s and a co ec ed es s a is ic o accoun o he
o dinal and no no mally dis ibu ed compliance i ems (e.g., Finney & DiS e ano, 2006; Li, 2016).
4
|
RESULTS
O e all, 3350 esponden s comple ed he online su ey, among which 49.85% we e emale, 49.91% we e male, and
0.24% indica ed hei gende o be di e se. Responden s we e on a e age 47.83 yea s old. In e ms o age, gende ,
educa ional a ainmen , and s a e o esidence ou sample was ep esen a i e o he Ge man popula ion aged 18–74
(see Table A2 in Appendix A). All six compliance i ems we e non‐no mally dis ibu ed and skewed o he le . This
means sel ‐ epo ed compliance was gene ally high, which could be indica i e o some social desi abili y bias in
epo ing bu is no necessa ily su p ising gi en ha da a was collec ed in he mids o he qui e in ense second wa e
o in ec ions in Ge many ( o a simila inding a ha ime, see Be sch e al., 2021b). His og ams and densi y plo s o
all six compliance i ems a e p esen ed in Figu e A1 in Appendix A. Desc ip i e s a is ics o all a iables employed in
he co e analysis a e p esen ed in Tables A1, A2
8
and A6 in Appendix A and he sc eeplo s and ac o sco es o
cons uc ing he indexes o COVID‐19 h ea pe cep ion and ci ic esponsibili y a e shown in Tables A7, A8 and
Figu es A2–A4 in Appendix A.
9
1062
-
STERNBERG ET AL.
4.3.2
|
Addi ional obus ness checks
We conduc ed a numbe o addi ional analyses o assess he obus ness o ou esul s by ha ing a close look a he
dependen a iable, ha is, he compliance measu es, as well as a he di e en hypo hesized p edic o s. Fo he sake o
b e i y, his subsec ion me ely summa izes he di e en app oaches and hei esul s b ie ly, while Appendix Bcon-
ains a mo e de ailed accoun o he a ionale behind he app oaches and o hei esul s.
FIGURE 3 Ex e nal ele ance o compliance domains: S a e‐le el Google mobili y pa e ns. This igu e shows sca e plo s, linea
(wi h 95% con idence in e als) eg ession lines, Pea son co ela ion coe icien s, and p‐ alues o (i) compliance in he public and p i a e
domain, as p edic ed om he s uc u al equa ion model (SEM) by means o Equa ions (1) and (2) (on he x‐axis), and (ii) phone‐ acking‐
based changes in mobili y acco ding o Google's mobili y epo s in he a eas o Re ail and Rec ea ion, Wo kplace, T ansi , and Residen ial
(on he y‐axis). Highe alues o he SEM‐p edic ed compliance measu es indica e highe compliance. In he i s h ee g aphs, lowe alues
o mobili y changes (i.e., mo e nega i e alues) indica e s onge educ ions in mobili y ela i e o he 2020 baseline pe iod
(co espondingly o inc eases in mobili y o he case o Re ail and Rec ea ion, ou h g aph). The uni o analysis a e he 16 ede al s a es
in Ge many, gi en ha his is he lowes le el o Google mobili y epo s a ailable o Ge many.
STERNBERG ET AL.
-
1069

In e ms o he elici a ion o compliance, we e‐es ima ed he SEM wi h an al e na i e measu e o compliance in he
p i a e domain (i.e., al e ing he measu emen pa o he SEM), which we assume is less suscep ible o social desi -
abili y bias, bu s ill has he ad an age o being a ailable a he esponden le el (as opposed o he Google mobili y
pa e ns). Fo his al e na i e measu e o p i a e compliance, we u ilized h ee ques ions asking esponden s abou hei
willingness o pa icipa e in conc e e social ac i i ies ( o mo e de ails see Appendix B). The esul s o his exe cise
e eal es ima es ha a e simila o he p e ious ones o compliance in he p i a e domain (and ha di e om he
ones o public compliance in he same c ucial ins ances), hus, s eng hening he c edibili y o he main esul s. See
Tables A17 and A18 and Figu es A9 and A10 in Appendix A o he de ailed esul s o he adjus ed measu emen model
and he s uc u al model o he SEM.
In e ms o he di e en hypo hesized p edic o s, we conduc ed addi ional analyses o (i) accoun o po en ially
media ing e ec s o esponden s' COVID‐19 h ea pe cep ion (see Table A19 and Figu e A11), and (ii) con ol o wo
mo e possibly impo an compe ing p edic o s, namely esponden s' deg ee o gene alized in e pe sonal us and hei
esidence in u ban e sus u al a eas (see Table A22). As in (i), we a gue ha indi idual h ea pe cep ions o he
COVID‐19 pandemic may hemsel es be a ec ed by economic p e e ences and ins i u ional us (Ha pe e al., 2020;
Plohl & Musil, 2021), he e o e sugges ing no only di ec bu also indi ec e ec s o p e e ences and us h ough
h ea pe cep ion. In iew o his, we e‐es ima ed he SEM by adding ou measu e o h ea pe cep ion as a media o o
he s uc u al componen o he SEM. The esul s sugges ha he di e ences ac oss compliance domains p edomi-
nan ly s em om di ec e ec s— hus, ein o cing he co e a gumen and inding o his pape —while he es ima ed
indi ec e ec s we e e y simila ac oss domains and in mos cases also much smalle in magni ude.
As in (ii), we ind ha ou main esul s a e obus o including a esponden 's deg ee o gene alized us —as
measu ed by means o a su ey‐based i em adop ed om Falk e al. (2018)—o hei u al as opposed o u ban
place o esidence as addi ional p edic o s o public and p i a e compliance. Mo eo e , ou esul s do no sugges ha a
u al/u ban se ing plays a s a is ically signi ican ole o compliance beha io s in ei he domain (i any, he e is a
small nega i e es ima ed e ec o a u al se ing on compliance). In e es ingly, we ind ha gene alized us ( he deg ee
o which esponden s belie e o he people o gene ally ha e good in en ions) signi ican ly dec eases compliance only in
he public domain, while i has no s a is ically signi ican es ima ed e ec in he p i a e domain (see Appendix B o a
mo e de ailed discussion o his addi ional inding).
5
|
DISCUSSION AND CONCLUSION
Compliance wi h expe beha io al ecommenda ions and explici manda es is c ucial o a socie y's abili y o achie e a
wide ange o public heal h objec i es (and public policy objec i es in gene al), anging om sa egua ds o pa ien
p i acy o accina ion manda es o equi emen s o COVID‐19‐ ela ed physical dis ancing. I is especially c ucial when
he beha io al s ipula ions o manda es es ablished by such no ms, s anda ds o egula ions a e only impe ec ly
en o ceable, and compliance depends o a g ea e and subs an i e deg ee on indi idual choices and conside a ions
abou whe he o comply. Acco dingly, p e ious esea ch has iden i ied economic p e e ences and ins i u ional us as
impo an d i e s o indi idual‐le el compliance (e.g., Algan e al., 2021; Ba gain & Aminjono , 2020; Campos‐Me cade,
Meie , Schneide , Meie , e al., 2021; Campos‐Me cade, Meie , Schneide , & Wengs öm, 2021; Chan, Skali, e al., 2020;
Cucciniello e al., 2022; Kese & Rau, 2023; Papanas asiou e al., 2022; Shim e al., 2012; Su inen & Kupe an, 1999).
In his pape , we ha e in oduced he concep ual dis inc ion be ween compliance in he public and he p i a e
domain and ha e explo ed empi ically, in he con ex o compliance wi h COVID‐19‐ ela ed physical dis ancing ules,
o wha ex en i s co ela ions wi h economic p e e ences and ins i u ional us di e ac oss he wo domains. Un-
de s anding indi idual‐le el compliance and ecognizing po en ial di e ences be ween he p i a e and public domain
emains highly ele an in his con ex . E en hough he immedia e u gency o he cu en pandemic may seem o ha e
passed, an inc easing likelihood o no el epidemics and pandemics (e.g., Ma ani e al., 2021) combined wi h a sig-
ni ican deg ee o accine hesi ancy “especially owa ds newly de eloped accines” sugges s ha physical dis ancing
manda es will pe sis as a c ucial pa o go e nmen s' policy oolki .
Ou ine‐g ained analysis e ealed sys ema ic he e ogenei y ac oss he wo iden i ied domains, ad ancing ou un-
de s anding o compliance and hus p o iding mo e angible g ounds o policy in e en ions. Speci ically, while in-
di iduals' isk and ime p e e ences appea ed o be simila ly ele an o compliance ac oss domains, we ound
signi ican ly di e en co ela ions ac oss he wo compliance domains in he case o (i) ecip oci y (and o some ex en
also gene ally o social p e e ences as a whole), (ii) ins i u ional us and (iii) COVID‐19 h ea pe cep ion.
15
1070
-
STERNBERG ET AL.
Fi s , ou empi ical analyses sugges ha elying on, o appealing o, ecip ocal dynamics may only be a p omising
s a egy o compliance in he public domain. These esul s a e in line wi h ou heo e ical expec a ion laid ou in Sec ion 2:
in he public ealm, he ecip oca ion o compliance beha io s (e.g., in he o m o wea ing a mask in public o only
en e ing an ele a o sepa a ely) is mo e obse able and, hus, much mo e salien in people's minds han in he p i a e
domain. He e, compliance occu s in he o m o s aying a home and isola ing, bu one does no di ec ly pe cei e o he s
doing hesame—a leas no o heex en ha i is hecasein hepublic ealm.Thisin e p e a ionismo eo e suppo ed by
he ac ha we obse ed a subs an i ely and s a is ically much weake co ela ion be ween compliance in he public
domain and al uism, which, in i s pu e o m, we had hypo hesized and de ined wi hou any ecip ocal componen .
Second, he esul s e eal somewha opposing pa e ns o us in he na ional go e nmen and us in scien i ic
ins i u ions ( he RKI). While he o me only seems o ma e in he p i a e ealm, he la e plays a c ucial ole in bo h
ealms bu mo e dominan ly in he public domain. Fo us in he RKI, we had weak expec a ions o a s onge
ele ance in he public domain as a esul o he ela i ely mo e echnical and speci ic s ipula ions in his domain,
which may, hus, be mo e salien ly pe cei ed as scien i ically alida ed egula ions. We ound suppo o his
expec a ion in supplemen a y analyses, which e eal he exac same pa e n o o he p oxies o us in scien i ic
ins i u ions, namely, us in science in gene al and us in he Wo ld Heal h O ganiza ion (see Table A28 and
Appendix B). Fo us in he na ional go e nmen , we had inde e mina e heo e ical expec a ions. Addi ional sup-
plemen a y analyses exploi ing esponden s' us in go e nmen ‐ ela ed media channels poin o a possible explana ion
o he high ele ance o us in he na ional go e nmen in he p i a e domain (see Table A29 and Appendix B). The
obse ed dynamic could well be a esul o he communica ion s a egies employed by he na ional go e nmen o e he
cou se o he pandemic and chancello Angela Me kel's public add esses u ging ci izens o s ay a home and isola e (i.e.,
using a na a i e along he lines o “uni ed in sepa a ion”). This communica ion s a egy migh ha e wo ked agains he
pe cep ion o he p i a e ealm as a ealm in which he go e nmen has no p ominen ole o play, especially o in-
di iduals wi h high le els o us in he go e nmen .
Finally, ou indings sugges ha policies which succeed in adequa ely in o ming indi iduals abou he h ea o ( he
de imen al consequences o ) a COVID‐19 in ec ion a e likely o be highly e ec i e ac oss compliance domains bu o
an e en g ea e ex en in he p i a e ealm. Conside ing he di e en heo e ically ou lined cha ac e is ics o he wo
domains, his inding seems plausible. In he p i a e domain, he pe cei ed isk o passing on an in ec ion o a close
amily membe o iend h ough non‐compliance is much mo e salien han in he public domain. A he same ime,
he ea o losing a close amily membe o iend o COVID‐19 was one o he mos dominan indica o s o ou measu e
o COVID‐19 h ea pe cep ion.
This po en ial o a a ying e ec i eness o communica ion s a egies ac oss he compliance domains sugges s ha
policy make s should ei he ailo communica ions s a egies o he ci cums ances o each domain o ocus on he
de e minan s ha a e common ac oss domains.
We a e cau iously e e ing o ou indings in e ms o co ela ions a he han causal e ec s, gi en ha ga he ing
he da a h ough a c oss‐sec ional su ey in he mids o he pandemic did no allow o ensu ing exogenei y by design.
Howe e , ou esul s emained s able when con olling o a ious al e na i e in luences and when pe o ming a
numbe o addi ional obus ness checks (see Sec ion 4). Mo eo e , p e ious li e a u e seems o sugges ha economic
p e e ences and ins i u ional us a e likely exogenous o he analyses conduc ed in his pape .
16
In addi ion, we also
acknowledge ha ou pape only employs sel ‐ epo ed compliance measu es and u ilizes da a om an online su ey
panel. Howe e , gi en he a ious obus ness checks conduc ed o alida e he compliance measu es and he com-
pa isons o ou su ey da a wi h o he ep esen a i e non‐online su eys, we s ill belie e ha ou esea ch documen s
eal beha io al mechanisms ha can p o ide use ul insigh s o policy make s. Finally, ou pape , s ic ly speaking, only
cap u es a snapsho o beha io s a he one speci ic poin in he pandemic when he da a was collec ed. Ne e heless,
ou su ey was conduc ed in he mids o Ge many's hi d wa e, and, hus, in he mids o a phase o he pandemic,
du ing which accines we e no ye a ailable and physical dis ancing was c ucial—which is he exac phase ele an o
ou esea ch ques ion.
Al hough we ha e in es iga ed compliance only in he pandemic con ex o Ge many, he gene al dis inc ion be-
ween compliance in he public and p i a e domain is likely also ele an o o he coun ies ha in oduced a simila
ca alog o physical dis ancing ules. Mo eo e , he concep ual and empi ical con ibu ion o his pape may ex end
beyond he con ex o he cu en and possible u u e pandemics o o he aspec s o public policy mo e gene ally as well
as o heal h policy in pa icula . One ela ed example is easily moni o able e sus la gely unobse ed compliance wi h
di e en ypes o hygiene egula ions by heal h pe sonnel/p o essionals. Ano he , mo e gene al example o a
con empo a y and e y p essing egula o y challenge wi h simila cha ac e is ics a e policies encou aging
STERNBERG ET AL.
-
1071
en i onmen ally esponsible beha io . A public‐p i a e compliance di e gence in his con ex may o ins ance man-
i es i sel in he di e en ial beha io al p edic o s o en i onmen ally esponsible consume beha io s in supe ma ke s
e sus online shopping wi h a home deli e y op ion. Finally, ou app oach also has heo e ical ele ance o esea ch
and es ablished indings on egula o y compliance by challenging he way in which compliance is concep ualized.
ACKNOWLEDGMENTS
The au ho s would like o hank he edi o o hei guidance and suppo h oughou he e iew p ocess, and he wo
anonymous e iewe s o hei aluable inpu s. The au ho s mo eo e wish o hank he membe s o he In e na ional
Rela ions Resea ch G oup a he Hochschule ü Poli ik/TUM Depa men o Go e nance, especially Tobias Rommel and
Zla ina Geo gie a, as well as he o ganize s and he pa icipan s o he beha io al & empi ical wo k in p og ess Semina a
he TUM School o Managemen and he 3 d DGGÖ Wo kshop “Heal h Economics and De elopmen ” o hei aluable
commen s. We u he wish o hank he pa icipan s o he TUM PhD semina POL90002, Resea ch Design and Empi ical
Me hodsand he TUMResea chDesign Bachelo Semina o ui ul discussions o he ques ionnai e and esea chdesign.
Finally, we would like o hank lec u e s and pa icipan s o he S . Gallen Global School in Empi ical Resea ch Me hods
(GSERM) o impo an discussions on Fac o analyses and S uc u al Equa ion models. The au ho s g a e ully
acknowledge unding om he Eu opean Union's Ho izon 2020 esea ch and inno a ion p og am unde g an ag eemen
no. 101016233 PERISCOPE. The unding sou ce had no in ol emen in he s udy design, collec ion, analysis and in e -
p e a ion o da a, in he w i ing o he epo , and in he decision o submi he a icle o publica ion.
Open Access unding enabled and o ganized by P ojek DEAL.
CONFLICT OF INTEREST STATEMENT
The au ho s decla e no con lic s o in e es .
DATA AVAILABILITY STATEMENT
Da a a ailable on eques om he au ho s.
ETHICS STATEMENT
E hics app o al o his s udy was ob ained om he commi ee o human subjec s and esea ch e hics e iew o he
medical acul y a he Technical Uni e si y o Munich (TUM, 20/21 S‐SR).
ORCID
Hen ike S e nbe g h ps://o cid.o g/0000-0001-8539-6478
Janina Isabel S eine h ps://o cid.o g/0000-0001-7120-0075
Tim Bü he h ps://o cid.o g/0000-0002-4724-5000
ENDNOTES
1
By unce ain y we mean long ails in he p obabili y dis ibu ion, as in wha Knigh (1921) called “ isk”.
2
An ex ensi e, bu no comple e lis o conduc ed wo ks in his ega d include Ba ios e al. (2021), Campos‐Me cade, Meie , Schneide ,
and Wengs öm (2021), Du an e e al. (2021), Mülle and Rau (2021), Ba sche e al. (2021), Bo gono i and And ieu (2020), Nikolo
e al. (2020), Quaas e al. (2021), She h and W igh (2020), and an Hulsen e al. (2020) o a ious ypes o social p e e ences; Papa-
nas asiou e al. (2022), Ande sson e al. (2021), Mülle and Rau (2021), Schunk and Wagne (2021), Al a o e al. (2022), Chan, Skali,
e al. (2020), Nikolo e al. (2020), Pullano e al. (2020), Xie e al. (2020), and Xu and Cheng (2021) o isk p e e ences; Fang e al. (2022),
Papanas asiou e al. (2022), Ande sson e al. (2021), Mülle and Rau (2021), Schunk and Wagne (2021), Al a o e al. (2022), and Nikolo
e al. (2020) o ime p e e ences; B odeu , G igo ye a, and Ka an (2021), Fa zanegan and Ho mann (2022), Fazio e al. (2021), G anados
Samayoa e al. (2021), Kazemian e al. (2021), Koe ke e al. (2021), Plohl and Musil (2021), Ba gain and Aminjono (2020), Chan,
B ump on, e al. (2020), and Golds ein and Wiedemann (2022) o ins i u ional us ; and Papanas asiou e al. (2022), Algan e al. (2021),
Jø gensen e al. (2021), Kluwe‐Schia on e al. (2021), Plohl and Musil (2021), Ha pe e al. (2020), Vai e al. (2020), and Van Ba el
e al. (2020) o COVID‐19‐ ela ed h ea pe cep ions.
3
In addi ion o he na ionwide ules o in e es o his pape , he e we e some mino di e ences ac oss he Ge man s a es in he speci ic
ules and ecommenda ions ega ding o example, school/nu se y es ic ions, con ac es ic ions o young child en and disabled in-
di iduals, o he speci ics o qua an ining a e e u ning om a ip ou side Ge many (P ess and In o ma ion O ice o he Fede al
Go e nmen , 2021a,2021b).
4
The beha io s and he co esponding ques ions we e adap ed om Be sch e al. (2021a) and sligh ly adjus ed. See Table A3 in Appendix A
o he exac wo ding. As pa o ou obus ness checks, we conside an al e na i e way o dis inguishing be ween compliance in he
public and he p i a e ealm; see Sec ion 4.3.
1072
-
STERNBERG ET AL.
5
The expe imen al alida ion p ocedu e enabled Falk e al. (2018) o analyze which linea combina ion o he di e en su ey i ems
pe o med bes in p edic ing he co esponding beha io in an expe imen al se ing in he lab. We used hese same iden i ied weigh s o
o m ou p e e ence measu es. No e ha Falk e al. (2018) conduc ed he alida ion p ocedu e wi h a Ge man sample and hus, in he
same coun y con ex as his s udy.
6
Fo posi i e ecip oci y, we we e only able o collec one o he wo su ey i ems in ended o o m he inal measu e o posi i e eci-
p oci y. We he e o e p oceeded wi h his single i em and u he assessed he esul s o obus ness when using only a single su ey i em
o all he o he p e e ences as well (selec ing he i em ha had been assigned he highes weigh in he expe imen al alida ion p ocedu e
by Falk e al. (2018)). Ou co e indings emained obus , see Table A20 and Figu e A12 in Appendix A.
7
P io o he pilo launch in he ield, he su ey was mo eo e pilo ed and discussed in wo esea ch design semina s a he au ho s'
uni e si y.
8
In o de o inc ease he c edibili y o and alida e he main a iables o in e es o ou empi ical analysis (i.e., compliance, economic
p e e ences and h ea pe cep ion), we compa e desc ip i e s a is ics o ou su ey da a wi h o he ep esen a i e su eys ha collec ed
da a on p esumably compa able i ems, speci ically da a om he COVID‐19 Snapsho Moni o ing, he Wo ld Value Su ey (WVS) and he
Global P e e ence Su ey. We gene ally ind a high simila i y be ween ou su ey da a and he o he da ase s (see Tables A24, A25 and
Figu e A13), while he simila i ies a e highes o ou measu es o compliance and h ea pe cep ion and sligh ly less so o ou measu es
o economic p e e ences. Speci ically, as in ui i ely o be expec ed, he simila i y in desc ip i e s a is ics is sligh ly lowe , he mo e a
measu e de ia es om hose employed by he Global P e e ence Su ey.
9
We examine, in an addi ional analysis, o wha ex en s a e‐le el a e aged COVID‐19 h ea pe cep ion is co ela ed wi h s a e‐le el
COVID‐19 case incidence a es. Fo his analysis, we employ da a abou he COVID‐19 incidence da e by egion (numbe o egis e ed
COVID‐19 in ec ions in a s a e wi hin he pas 7 days/100,000 inhabi an s) om he COVID‐19 da ase by he Fede al S a is ical O ice o
Ge many (“S a is isches Bundesam ”). We ind a posi i e (s a is ically insigni ican ) co ela ion be ween COVID‐19 case incidence du ing
he ime o he da a collec ion and a e age h ea pe cep ion a he s a e le el (Pea son co ela ion coe icien : þ0.2928). We, mo eo e ,
ind a nega i e (s a is ically insigni ican ) co ela ion be ween s a e‐le el h ea pe cep ion and he a e age case incidence in a s a e
h oughou he in ec ion wa es since he s a o he pandemic and he e o e also p io o he s a o ou da a collec ion (Pea son
co ela ion coe icien : −0.3847). See Table A23 in Appendix A as well as Appendix B o a de ailed summa y and in e p e a ion o esul s.
10
Deg ees o eedom o SEMs a e di e en ly calcula ed (see e.g., Rigdon, 1994).
11
See Table A16 in Appendix A o he Pea son's co ela ion ma ix o he co e explana o y a iables, which sugges s ha we we e no acing
a case o highly co ela ed explana o y a iables. Wi h he excep ion o us in go e nmen / us in he RKI (co . coe .: 0.71), all o he
pai wise co ela ions we e o a low/mode a e deg ee (co . coe .:0.01–0.38).
12
We conduc a numbe o addi ional analyses wi h po en ial p oxies o us in he RKI ( us in science, us in WHO) and po en ial
channels o us in he go e nmen ( us in s a e‐le el go e nmen , us in es ablished media channels) o be e unde s and he
di e en ial e ec s obse ed o he ins i u ional us a iables. The esul s e eal ha he es ima ed e ec s o us in he RKI a e indeed
obus , ha is, we obse e he same pa e n o he u ilized p oxies: a high, s a is ically signi ican ele ance in bo h domains, bu mo e so
in he public domain. Fo us in he go e nmen , he addi ional analyses seem o sugges ha he la ge, s a is ically signi ican es ima ed
e ec s only in he p i a e domain may be a esul o he communica ion s a egy o pa icula ly he na ional go e nmen , whose na a i e
ocused p edominan ly on he gene al message o s ay a home and isola e (i.e., closely ela ed o ou de ini ion o p i a e compliance). See
Tables A28 and A29 o de ails.
13
Ou indings o esponden s' age align well wi h hose o COVID‐19 h ea pe cep ions, which p e ious s udies ha e ound o be s onge
among olde ci izens: Bo h ac o s we e a s onge p edic o o compliance in he p i a e domain. Howe e , he ac ha bo h a iables
emain s a is ically signi ican when added simul aneously sugges s ha he age e ec cap u es a dynamic ha is somewha dis inc om
esponden s' pandemic‐ ela ed h ea pe cep ions (e.g., olde people being mo e ule‐complian in gene al, especially in p i a e se ings,
while younge people la gely comply only when subs an ially moni o ed).
14
Mobili y changes in going o he g oce y s o e/pha macy a e concep ually un ela ed o he compliance i ems in ou su ey. Mobili y
changes in pa ks would be ha d o in e p e gi en e y di e en wea he condi ions h oughou weeks o he yea , which p edominan ly
de e mine ou side ac i i ies in Ge many du ing hese mon hs.
15
Ou esul s a e pa icula ly in e es ing in ligh o a ecen con ibu ion by Papanas asiou e al. (2022): The au ho s ind ha economic
p e e ences, speci ically, isk and ime p e e ences, become less ele an as p edic o s o compliance beha io s i esponden s a e p e-
sen ed wi h he hypo he ical p ospec o being ined o non‐compliance. While hey conduc ed hei da a collec ion a a poin in ime
when ines had no ye been in oduced, ines had al eady come in o e ec when we conduc ed ou s udy. The compa ison wi h
Papanas asiou e al. (2022) hus seems o pe mi he ollowing wo addi ional in e p e a ions o ou esul s: Fi s , he ac ha we ind
economic p e e ences in gene al o be s a is ically ele an despi e he exis ence o ees sugges s ha he es ima ed e ec s a e a he a
lowe bound o hei ele ance in he absence o ees. Second, he inding by Papanas asiou e al. (2022), who examined isk and ime
p e e ences in pa icula , may also se e as a pa ial explana ion o why social p e e ences as opposed o isk and ime p e e ences a e
mo e impo an o compliance beha io s in ou s udy. This applies especially o compliance in he public ealm, which we had no only
assigned a la ge likelihood o o mal (s a e‐) punishmen h ough ines, bu also o social punishmen o which especially he impac s o
social p e e ences seem o be sensi i e.
STERNBERG ET AL.
-
1073
16
Be sch e al. (2021b), D ichou is and Nayga (2021), Shacha e al. (2021), Ang isani e al. (2020), Bu e al. (2020), Ikeda e al. (2023),
Lo i and Pe hiyagoda (2021), an de G oep e al. (2020), Habibpou e al. (2018), Chuang and Schech e (2015), Meie and
Sp enge (2015), Ca lsson e al. (2014), Volk e al. (2012), and Ande sen e al. (2008), see Appendix B o a s uc u ed and mo e de ailed
o e iew o hese a icles' ele ance o ou a gumen .
17
Da a was e ie ed om h ps://www.co ona‐da en‐deu schland.de/da ase /in ek ionen_bundeslaende .
REFERENCES
Acock, A. C. (2013). Disco e ing s uc u al equa ion modeling using S a a. S a a P ess Books.
Al a o, L., Faia, E., Lame sdo , N., & Saidi, F. (2022). Heal h ex e nali ies and policy: The ole o social p e e ences. Managemen Science,
68(9), 6751–6761. h ps://doi.o g/10.1287/mnsc.2022.4461
Algan, Y., Cohen, D., Da oine, E., Foucaul , M., & S an che a, S. (2021). T us in scien is s in imes o pandemic: Panel e idence om 12
coun ies. P oceedings o he Na ional Academy o Sciences,118(40), e2108576118. h ps://doi.o g/10.1073/pnas.2108576118
Ande sen, S., Ha ison, G. W., Lau, M. I., & Elisabe Ru s öm, E. (2008). Los in s a e space: A e p e e ences s able? In e na ional Economic
Re iew,49(3), 1091–1112. h ps://doi.o g/10.1111/j.1468‐2354.2008.00507.x
Ande sson, O., Campos‐Me cade, P., Meie , A. N., & Wengs öm, E. (2021). An icipa ion o COVID‐19 accines educes willingness o
socially dis ance. Jou nal o Heal h Economics,80, 102530. h ps://doi.o g/10.1016/j.jhealeco.2021.102530
Ang isani, M., Cip iani, M., Gua ino, A., Kendall, R., & O iz de Za a e, J. (2020). Risk p e e ences a he ime o COVID‐19: An expe imen
wi h p o essional ade s and s uden s (Fede al Rese e Bank o New Yo k S a Repo 927). h ps://ss n.com/abs ac =3609586
Ba gain, O., & Aminjono , U. (2020). T us and compliance o public heal h policies in imes o COVID‐19. Jou nal o Public Economics,192,
104316. h ps://doi.o g/10.1016/j.jpubeco.2020.104316
Ba ios, J. M., Benmelech, E., Hochbe g, Y. V., Sapienza, P., & Zingales, L. (2021). Ci ic capi al and social dis ancing du ing he COVID‐19
pandemic. Jou nal o Public Economics,193, 104310. h ps://doi.o g/10.1016/j.jpubeco.2020.104310
Ba sche , A. K., Sei z, S., Siegloch, S., Slo winski, M., & Weh hö e , N. (2021). Social capi al and he sp ead o COVID‐19: Insigh s om
Eu opean coun ies. Jou nal o Heal h Economics,80, 102531. h ps://doi.o g/10.1016/j.jhealeco.2021.102531
Be sch, C., Wiele , L., Bosnjak, M., Ramha e , M., S ollo z, V., Ome , S., Ko n, L., Sp engholz, P., Felgend e , L., Ei ze, S., & Schmid, P.
(2021a). Ge many COVID‐19 Snapsho MOni o ing (COSMO Ge many): Moni o ing knowledge, isk pe cep ions, p e en i e beha -
iou s, and public us in he cu en co ona i us ou b eak in Ge many – Wa e 27. Re ie ed om h ps://p ojek e.uni‐e u .de/
cosmo2020/web/summa y/27/
Be sch, C., Wiele , L., Bosnjak, M., Ramha e , M., S ollo z, V., Ome , S., Ko n, L., Sp engholz, P., Felgend e , L., Ei ze, S., & Schmid, P.
(2021b). Ge many COVID‐19 Snapsho MOni o ing (COSMO Ge many): Moni o ing knowledge, isk pe cep ions, p e en i e beha -
iou s, and public us in he cu en co ona i us ou b eak in Ge many – Wa e 38 summa y. Re ie ed om h ps://p ojek e.uni‐e u .
de/cosmo2020/web/summa y/38/
Blai , R. A., Mo se, B. S., & Tsai, L. L. (2017). Public heal h and public us : Su ey e idence om he Ebola Vi us Disease epidemic in
Libe ia. Social Science & Medicine,172, 89–97. h ps://doi.o g/10.1016/j.socscimed.2016.11.016
Böhm, R., Be sch, C., & Ko n, L. (2016). Sel ish‐ a ional non‐ accina ion: Expe imen al e idence om an in e ac i e accina ion game.
Jou nal o Economic Beha io & O ganiza ion,131, 183–195. h ps://doi.o g/10.1016/j.jebo.2015.11.008
Bo gono i, F., & And ieu, E. (2020). Bowling oge he by bowling alone: Social capi al and COVID‐19. Social Science & Medicine,265, 113501.
h ps://doi.o g/10.1016/j.socscimed.2020.113501
B odeu , A., G ay, D., Islam, A., & Bhuiyan, S. (2021). A li e a u e e iew o he economics o COVID‐19. Jou nal o Economic Su eys,35(4),
1007–1044. h ps://doi.o g/10.1111/joes.12423
B odeu , A., G igo ye a, I., & Ka an, L. (2021). S ay‐a ‐home o de s, social dis ancing, and us . Jou nal o Popula ion Economics,34(4),
1–34. h ps://doi.o g/10.1007/s00148‐021‐00848‐z
Bu, D., Hanspal, T., Liao, Y., & Liu, Y. (2020). Risk aking du ing a global c isis: E idence om Wuhan. Co id Economics,5, 106–146.
Campos‐Me cade, P., Meie , A. N., Schneide , F. H., Meie , S., Pope, D., & Wengs öm, E. (2021). Mone a y incen i es inc ease COVID‐19
accina ions. Science,374(6569), 879–882. h ps://doi.o g/10.1126/science.abm0475
Campos‐Me cade, P., Meie , A. N., Schneide , F. H., & Wengs öm, E. (2021). P osociali y p edic s heal h beha io s du ing he COVID‐19
pandemic. Jou nal o Public Economics,195, 104367. h ps://doi.o g/10.1016/j.jpubeco.2021.104367
Ca lsson, F., Johansson‐S enman, O., & Nam, P. K. (2014). Social p e e ences a e s able o e long pe iods o ime. Jou nal o Public Eco-
nomics,117, 104–114. h ps://doi.o g/10.1016/j.jpubeco.2014.05.009
Chan, H. F., B ump on, M., Macin y e, A., A apoc, J., Sa age, D. A., Skali, A., S adelmann, D., & To gle , B. (2020). How con idence in heal h
ca e sys ems a ec s mobili y and compliance du ing he COVID‐19 pandemic. PLoS One,15(10), e0240644. h ps://doi.o g/10.1371/
jou nal.pone.0240644
Chan, H. F., Skali, A., Sa age, D. A., S adelmann, D., & To gle , B. (2020). Risk a i udes and human mobili y du ing he COVID‐19
pandemic. Scien i ic Repo s,10(1), 1–13. h ps://doi.o g/10.1038/s41598‐020‐76763‐2
Chuang, Y., & Schech e , L. (2015). S abili y o expe imen al and su ey measu es o isk, ime, and social p e e ences: A e iew and some
new esul s. Jou nal o De elopmen Economics,117, 151–170. h ps://doi.o g/10.1016/j.jde eco.2015.07.008
Cucciniello, M., Pin, P., Im e, B., Po umbescu, G. A., & Melega o, A. (2022). Al uism and accina ion in en ions: E idence om beha io al
expe imen s. Social Science & Medicine,292, 114195. h ps://doi.o g/10.1016/j.socscimed.2021.114195
1074
-
STERNBERG ET AL.

Des a is S a is isches Bundesam . (2021). Be ölke ungss and: Am liche Einwohne zahl Deu schlands 2022. [Online]. Re ie ed Janua y, 29,
2021 om h ps://www.des a is.de/DE/Themen/Gesellscha ‐Umwel /Be oelke ung/Be oelke ungss and/_inhal .h ml
D ichou is, A. C., & Nayga, R. M. (2021). On he s abili y o isk and ime p e e ences amid he COVID‐19 pandemic. Expe imen al Eco-
nomics,25(3), 1–36. h ps://doi.o g/10.1007/s10683‐021‐09727‐6
Du an e, R., Guiso, L., & Gulino, G. (2021). Asocial capi al: Ci ic cul u e and social dis ancing du ing COVID‐19. Jou nal o Public Eco-
nomics,194, 104342. h ps://doi.o g/10.1016/j.jpubeco.2020.104342
Falk, A., Becke , A., Dohmen, T., Enke, B., Hu man, D., & Sunde, U. (2018). Global e idence on economic p e e ences. Qua e ly Jou nal o
Economics,133(4), 1645–1692. h ps://doi.o g/10.1093/qje/qjy013
Falk, A., Becke , A., Dohmen, T. J., Hu man, D., & Sunde, U. (2022). The p e e ence su ey Module: A alida ed ins umen o measu ing
isk, ime, and social p e e ences. Managemen Science,69(4), 1935–1950. h ps://doi.o g/10.1287/mnsc.2022.4455
Fang, X., F eye , T., Ho, C. Y., Chen, Z., & Goe e, L. (2022). P osociali y p edic s indi idual beha io and collec i e ou comes in he COVID‐
19 pandemic. Social Science & Medicine,308, 115192. h ps://doi.o g/10.1016/j.socscimed.2022.115192
Fa zanegan, M. R., & Ho mann, H. P. (2022). A ma e o us ? Poli ical us and he COVID‐19 pandemic. In e na ional Jou nal o So-
ciology,52(6), 476–499. h ps://doi.o g/10.1080/00207659.2022.2086729
Fazio, R. H., Ruisch, B. C., Moo e, C. A., G anados Samayoa, J. A., Boggs, S. T., & Ladanyi, J. T. (2021). Who is (no ) complying wi h he US
social dis ancing di ec i e and why? Tes ing a gene al amewo k o compliance wi h i ual measu es o social dis ancing. PLoS One,
16(2), e0247520. h ps://doi.o g/10.1371/jou nal.pone.0247520
Finney, S. J., & DiS e ano, C. (2006). Non‐no mal and ca ego ical da a in s uc u al equa ion modeling. In G. R. Hancock & R. O. Muelle
(Eds.), S uc u al equa ion modeling: A second cou se (Chap. 9, pp. 269–314). In o ma ion Age Publishing.
Galizzi, M. M., W. Lau, K., Mi aldo, M., & Hauck, K. (2022). Bandwagoning, ee‐ iding and he e ogenei y in in luenza accine decisions: An
online expe imen . Heal h Economics,31(4), 614–646. h ps://doi.o g/10.1002/hec.4467
Golds ein, D. A., & Wiedemann, J. (2022). Who do you us ? The consequences o poli ical and social us o public esponsi eness o
COVID‐19 o de s. The Consequences o Poli ical and Social T us o Public Responsi eness o COVID‐19 O de s. Pe spec i es on Poli ics,
20(2), 412–438. h ps://doi.o g10.1017/S1537592721000049
Google LLC. (2021). Google COVID‐19 communi y mobili y epo s [online]. Re ie ed July 9, 2021, om h ps://www.google.com/co id19/
mobili y/
G anados Samayoa, J. A., Ruisch, B. C., Moo e, C. A., Boggs, S. T., Ladanyi, J. T., & Fazio, R. H. (2021). When does knowing be e mean
doing be e ? T us in P esiden T ump and in scien is s mode a es he ela ion be ween COVID‐19 knowledge and social dis ancing.
Jou nal o Elec ions, Public Opinion, and Pa ies,31(sup1), 218–231. h ps://doi.o g/10.1080/17457289.2021.1924744
Habibpou , M. M., Pei e , M., Pepe mans, R., & Jege s, M. (2018). How gi ing a ec s gi ing: A long‐ e m analysis o dona ions. Applied
Economics,50(21), 2402–2413. h ps://doi.o g/10.1080/00036846.2017.1397853
Ha pe , C. A., Sa chell, L. P., Fido, D., & La zman, R. D. (2020). Func ional ea p edic s public heal h compliance in he COVID‐19
pandemic. In e na ional Jou nal o Men al Heal h and Addic ion,19(5), 1–14. h ps://doi.o g/10.1007/s11469‐020‐00281‐5
Ikeda, S., Yamamu a, E., & Tsu sui, Y. (2023). COVID‐19 Enhanced Diminishing Sensi i i y in P ospec ‐Theo y Risk P e e ences: A Panel
Analysis. Re iew o Beha io al Economics,10(4), 287–313.
Jø gensen, F., Bo , A., & Pe e sen, M. B. (2021). Compliance wi hou ea : Indi idual‐le el p o ec i e beha iou du ing he i s wa e o he
COVID‐19 pandemic. B i ish Jou nal o Heal h Psychology,26(2), 679–696. h ps://doi.o g/10.1111/bjhp.12519
Kazemian, S., Fulle , S., & Alga a, C. (2021). The ole o ace and scien i ic us on suppo o COVID‐19 social dis ancing measu es in he
Uni ed S a es. PLoS One,16(7), e0254127. h ps://doi.o g/10.1371/jou nal.pone.0254127
Kempe , C. J., Beie lein, C., Bensch, D., Ko ale a, A., & Ramms ed , B. (2012). Eine Ku zskala zu E assung des Gamma‐Fak o s sozial
e wünsch en An wo e hal ens: Die Ku zskala Soziale E wünsch hei ‐Gamma (KSE‐G) (GESIS‐Leibniz‐Ins i u ü Sozialwissen-
scha en Wo king Pape s, 2012/25). GESIS ‐ Leibniz‐Ins i u ü Sozialwissenscha en.
Kese , C., & Rau, H. A. (2023). De e minan s o people’s mo i a ions o app oach COVID‐19 accina ion cen e s. Scien i ic Repo s,13(1),
5282. h ps://doi.o g/10.1038/s41598‐023‐30244‐4
Klopp, E. (2022). A u o ial on es ing he equali y o s anda dized eg ession coe icien s in s uc u al equa ion models using Wald es s wi h
la aan. The Quan i a i e Me hods o Psychology,16(4), 315–333.
Kluwe‐Schia on, B., Viola, T. W., Bandinelli, L. P., Cas o, S. C. C., K is ensen, C. H., Cos a da Cos a, J., & G assi‐Oli ei a, R. (2021). A
beha io al economic isk a e sion expe imen in he con ex o he COVID‐19 pandemic. PLoS One,16(1), e0245261. h ps://doi.o g/10.
1371/jou nal.pone.0245261
Knigh , F. H. (1921). Risk, unce ain y and p o i (Vol. 31). Hough on Mi lin.
Koe ke, J., Schumann, K., & Po e , T. (2021). T us in science inc eases conse a i e suppo o social dis ancing. G oup P ocesses &
In e g oup Rela ions,24(4), 680–697. h ps://doi.o g/10.1177/1368430220985918
Li, C.‐H. (2016). Con i ma o y ac o analysis wi h o dinal da a: Compa ing obus maximum likelihood and diagonally weigh ed leas
squa es. Beha io Resea ch Me hods,48(3), 936–949. h ps://doi.o g/10.3758/s13428‐015‐0619‐7
Lo i, L., & Pe hiyagoda, S. (2021). Gene osi y du ing COIVID‐19: In es iga ing socioeconomic shocks and game aming. Humani ies and
Social Sciences Communica ions,9(1). h ps://doi.o g/10.1057/s41599‐022‐01200‐w
Ma ani, M., Ka ul, G. G., Pan, W. K., & Pa ola i, A. J. (2021). In ensi y and equency o ex eme no el epidemics. P oceedings o he Na ional
Academy o Sciences,118(35), e2105482118. h ps://doi.o g/10.1073/pnas.2105482118
STERNBERG ET AL.
-
1075
Meie , S., & Sp enge , C. D. (2015). Tempo al s abili y o ime p e e ences. The Re iew o Economics and S a is ics,97(2), 273–286. h ps://doi.
o g/10.1162/ es _a_00433
Me kel, A. (Tele ised speech). (2020). Fe nsehansp ache on Angela Me kel (P essemi eilung 100). [Online]. Re ie ed May, 10, 2023, om
h ps://www.bundes egie ung.de/b eg‐de/ak uelles/ e nsehansp ache‐ on‐bundeskanzle in‐angela‐me kel‐1732134
Mülle , S., & Rau, H. A. (2021). Economic p e e ences and compliance in he social s ess es o he COVID‐19 c isis. Jou nal o Public
Economics,194, 104322. h ps://doi.o g/10.1016/j.jpubeco.2020.104322
Nikolo , P., Pape, A., Tonguc, O., & Williams, C. (2020). P edic o s o social dis ancing and mask‐wea ing beha io : Panel su ey in se en US
s a es. Human Capi al and Economic Oppo uni y Wo king G oup.
Papanas asiou, A., Ru le, B. J., & Zheng, A. (2022). Compliance wi h social dis ancing: Theo y and empi ical e idence om On a io du ing
COVID‐19. Canadian Jou nal o Economics/Re ue canadienne d’économique,55(S1), 705–734. h ps://doi.o g/10.1111/caje.12565
Plohl, N., & Musil, B. (2021). Modeling compliance wi h COVID‐19 p e en ion guidelines: The c i ical ole o us in science. Psychology
Heal h & Medicine,26(1), 1–12. h ps://doi.o g/10.1080/13548506.2020.1772988
P ess and In o ma ion O ice o he Fede al Go e nmen . (2021a). Co ona i us in Deu schland [online]. Re ie ed Sep embe 19, 2021, om
h ps://www.bundes egie ung.de/b eg‐de/ hemen/co ona i us/co ona‐diese‐ egeln‐und‐einsch aenkung‐gel en‐1734724
P ess and In o ma ion O ice o he Fede al Go e nmen . (2021b). Regeln in den Bundeslände n [online]. Re ie ed Sep embe 19, 2021,
om h ps://www.bundes egie ung.de/b eg‐de/ hemen/co ona i us/co ona‐bundeslaende ‐1745198
Pullano, G., Valdano, E., Sca pa, N., Rub ichi, S., & Colizza, V. (2020). E alua ing he e ec o demog aphic ac o s, socioeconomic ac o s,
and isk a e sion on mobili y du ing he COVID‐19 epidemic in F ance unde lockdown: A popula ion‐based s udy. The Lance Digi al
Heal h,2(12), e638–e649. h ps://doi.o g/10.1016/s2589‐7500(20)30243‐0
Quaas, M., Meya, J., Schenk, H., Bos, B., D upp, M. A., & Requa e, T. (2021). The social cos o con ac s: Theo y and e idence o he COVID‐
19 pandemic in Ge many. PLoS One,16(3), e0248288. h ps://doi.o g/10.1371/jou nal.pone.0248288
Rigdon, E. E. (1994). Calcula ing deg ees o eedom o a s uc u al equa ion model. S uc u al Equa ion Modeling: A Mul idisciplina y
Jou nal,1(3), 274–278. h ps://doi.o g/10.1080/10705519409539979
Schunk, D., & Wagne , V. (2021). Wha de e mines he willingness o sanc ion iola ions o newly in oduced social no ms: Pe sonali y ai s
o economic p e e ences? E idence om he COVID‐19 c isis. Jou nal o Beha io al and Expe imen al Economics,93, 101716. h ps://
doi.o g/10.1016/j.socec.2021.101716
Shacha , J., Walke , M., & Wei, L. (2021). How he onse o he COVID‐19 pandemic impac ed p o‐social beha iou and indi idual p e -
e ences: Expe imen al e idence om China. Jou nal o Economic Beha io & O ganiza ion,190, 480–494. h ps://doi.o g/10.1016/j.jebo.
2021.08.001
She h, K., & W igh , G. C. (2020). The usual suspec s: Do isk ole ance, al uism, and heal h p edic he esponse o COVID‐19? Re iew o
Economics o he Household,18(4), 1041–1052. h ps://doi.o g/10.1007/s11150‐020‐09515‐w
Shim, E., Chapman, G. B., Townsend, J. P., & Gal ani, A. P. (2012). The in luence o al uism on in luenza accina ion decisions. Jou nal o
The Royal Socie y In e ace,9(74), 2234–2243. h ps://doi.o g/10.1098/ si .2012.0115
Su inen, J. G., & Kupe an, K. (1999). A socio‐economic heo y o egula o y compliance. In e na ional Jou nal o Social Economics,26(1–2),
174–193. h ps://doi.o g/10.1108/03068299910229569
Vai, B., Cazze a, S., Ghiglino, D., Pa en i, L., Saibene, G., To i, M., Ve ga, C., Wykowska, A., & Benede i, F. (2020). Risk pe cep ion and
media in shaping p o ec i e beha io s: Insigh s om he ea ly phase o COVID‐19 I alian ou b eak. F on ie s in Psychology,11.h ps://
doi.o g/10.3389/ psyg.2020.563426
Van Ba el, J. J., Baicke , K., Boggio, P. S., Cap a o, V., Cichocka, A., Cika a, M., C ocke , M. J., C um, A. J., Douglas, K. M., D uckman, J. N.,
D u y, J., Dube, O., Elleme s, N., Finkel, E. J., Fowle , J. H., Gel and, M., Han, S., Haslam, S. A., Je en, J., … Wille , R. (2020). Using
social and beha iou al science o suppo COVID‐19 pandemic esponse. Na u e Human Beha iou ,4(5), 460–471. h ps://doi.o g/10.
1038/s41562‐020‐0884‐z
an de G oep, S., Zanolie, K., G een, K. H., Sweijen, S. W., & C one, E. A. (2020). A daily dia y s udy on adolescen s’ mood, empa hy, and
p osocial beha io du ing he COVID‐19 pandemic. PLoS One,15(10), e0240349. h ps://doi.o g/10.1371/jou nal.pone.0240349
an Hulsen, M., Rohde, K. I., & an Exel, J. (2020). In e ‐ empo al and social p e e ences p edic compliance in a social dilemma: An
applica ion in he con ex o COVID‐19 (Tinbe gen Ins i u e Discussion Pape 2020‐047/I). Tinbe gen Ins i u e. h p://hdl.handle.ne /
10419/229667
Volk, S., Thöni, C., & Ruig ok, W. (2012). Tempo al s abili y and psychological ounda ions o coope a ion p e e ences. Jou nal o Economic
Beha io & O ganiza ion,81(2), 664–676. h ps://doi.o g/10.1016/j.jebo.2011.10.006
Xie, K., Liang, B., Dulebene s, M. A., & Mei, Y. (2020). The impac o isk pe cep ion on social dis ancing du ing he COVID‐19 pandemic in
China. In e na ional Jou nal o En i onmen al Resea ch and Public Heal h,17(17), 6256. h ps://doi.o g/10.3390/ije ph17176256
Xu, P., & Cheng, J. (2021). Indi idual di e ences in social dis ancing and mask‐wea ing in he pandemic o COVID‐19: The ole o need o
cogni ion, sel ‐con ol and isk a i ude. Pe sonali y and Indi idual Di e ences,175, 110706. h ps://doi.o g/10.1016/j.paid.2021.110706
How o ci e his a icle: S e nbe g, H., S eine , J. I., & Bü he, T. (2024). Compliance in he public e sus he
p i a e ealm: Economic p e e ences, ins i u ional us and COVID‐19 heal h beha io s. Heal h Economics,
33(5), 1055–1119. h ps://doi.o g/10.1002/hec.4807
1076
-
STERNBERG ET AL.
APPENDIX A
TABLE A1 O e iew and desc ip i e s a is ics o su ey i ems.
Va iable Values/desc ip ion Mean (SD) Min Max N
Female 0: Male (50.03%) 0.499 (0.50) 0 1 3342
1: Female (49.97%)
Age g oup 1: 18–29 (18.99%) 3.12 (1.44) 1 5 3350
2: 30–39 (17.85%)
3: 40–49 (18.06%)
4: 50–59 (22.09%)
5: 60þ(23.01%)
Educa ion 1: Low (33.50%) 2.02 (0.83) 1 3 3349
2: Medium (31.11%)
3: High (35.38%)
Employed in essen ial se ices 0: No (78.81%) 0.21 (0.41) 0 1 3350
1: Yes (21.19%)
Household income (mon hly) 1: Less han 900€ (9.30%) 4.54 (2.06) 1 10 3333
2: 900–1299€ (10.68%)
3: 1300–1699€ (10.56%)
4: 1700–2299€ (16.17%)
5: 2300–3199€ (20.70%)
6: 3200–3999€ (15.39%)
7: 4000–4999€ (10.23%)
8: 5000–5999€ (3.96%)
9: 6000–9999€ (2.61%)
10: Mo e han 10,000€ (0.39%)
Household size 1: Jus me (28.90%) 2.06 (0.85) 1 4 3350
2: 2–3 people (40.15%)
3: 3–4 people (26.81%)
4: Mo e han 4 people (4.15%)
Compliance i ems
Keeping dis ance Scale om 1 (applies ne e ) o 5 (applies always in
las 2 weeks)
4.44 (0.87) 1 5 3346
Wea ing masks Scale om 1 (applies ne e ) o 5 (applies always in
las 2 weeks)
4.78 (0.71) 1 5 3347
A oiding handshakes Scale om 1 (applies ne e ) o 5 (applies always in
las 2 weeks)
4.70 (0.79) 1 5 3347
Only lea ing home when necessa y Scale om 1 (applies ne e ) o 5 (applies always in
las 2 weeks)
4.21 (1.11) 1 5 3346
A oiding o he households Scale om 1 (applies ne e ) o 5 (applies always in
las 2 weeks)
4.15 (1.11) 1 5 3345
(Con inues)
STERNBERG ET AL.
-
1077
TABLE A1 (Con inued)
Va iable Values/desc ip ion Mean (SD) Min Max N
Res ic ing p i a e mee ings Scale om 1 (applies ne e ) o 5 (applies always in
las 2 weeks)
4.14 (1.16) 1 5 3347
Al e na i e compliance i ems
Willingness o join iends’ sledding ip nex
weekend ( e . coded)
Scale om 1 (no a all willing) o 7 ( e y willing) 2.33 (1.91) 1 7 2742
Willingness o join iends’ su p ise bi hday
ca oling ( e . coded)
Scale om 1 (no a all willing) o 7 ( e y willing) 3.32 (2.18) 1 7 3002
Willingness o join iends’ dinne /mo ie nigh
( e . coded)
Scale om 1 (no a all willing) o 7 ( e y willing) 2.40 (1.96) 1 7 2944
Economic p e e ence i ems
Al uism
Incen i ized dona ion decision 0–10 (dona ion om 10€ wind all endowmen ) 3.88 (3.46) 0 10 3320
Willingness o gi e o good causes Scale om 1 (no a all willing) o 11 ( e y willing) 7.82 (2.82) 1 11 3349
Pos. ecip oci y
Willingness o e u n a a o Scale om 1 (no a all willing) o 11 ( e y willing) 9.44 (1.98) 1 11 3349
Neg. ecip oci y
Willingness o ake e enge Scale om 1 (no a all willing) o 11 ( e y willing) 3.80 (2.78) 1 11 3348
Willingness o punish un ai beha io owa d
sel
Scale om 1 (no a all willing) o 11 ( e y willing) 4.88 (2.87) 1 11 3349
Willingness o punish un ai beha io owa d
o he s
Scale om 1 (no a all willing) o 11 ( e y willing) 5.28 (2.83) 1 11 3349
Risk a e sion
Willingness o ake isks ( e . coded) Scale om 1 (no a all willing) o 11 ( e y willing) 6.45 (2.63) 1 11 3350
Incen i ized lo e y choice 0: Lo e y wi h 5% chance o 10 Eu o paymen
(54.45%)
0.46 (0.50) 0 1 3339
1: Gua an eed paymen o 0.50 Eu o (45.55%)
Pa ience
In e empo al choice sequence using s ai case
me hod
1–32 (see Figu e A5 o de ails) 17.20 (12.11) 1 32 3111
Willingness o wai Scale om 1 (no a all willing) o 11 ( e y willing) 7.53 (2.48) 1 11 3349
Ci ic esponsibili y
Vo e u nou 0: Did no o e in he las na ional elec ion 2017
(14.94%)
0.85 (0.36) 0 1 3200
1: Vo ed in he las na ional elec ion 2017 (85.06%)
F equency o e ade a es ( e . coded) Scale om 1 (ne e ) o 6 (always) 1.71 (0.98) 1 6 3350
F equency o li e ing ( e . coded) Scale om 1 (applies no a all) o 4 (applies
always)
1.78 (0.93) 1 4 3350
Ins i u ional us i ems
T us in RKI Scale om 1 ( e y low us ) o 7 ( e y high us in
his ins i u ion)
4.91 (1.91) 1 7 3350
T us in go e nmen Scale om 1 ( e y low us ) o 7 ( e y high us in
his ins i u ion)
4.10 (1.90) 1 7 3349
1078
-
STERNBERG ET AL.
TABLE A12 S uc u al equa ion model (SEM) measu emen componen (con i ma o y ac o analysis).
La en a iable
Compliance in he public
domain
Compliance in he p i a e
domain
Wea ing a mask in public anspo /when shopping 0.892***
(0.035)
Keeping a 1.5 m dis ance in public spaces (whene e possible) 0.833***
(0.027)
A oiding handshake g ee ings 0.886
(cons .)
Lea ing he home only when absolu ely necessa y 0.775***
(0.028)
Gene ally a oiding o he households 0.895***
(0.038)
Res ic ing p i a e mee ings o 1 pe son om ano he
household
0.743
(cons .)
Obse a ions =2922
No e: The able shows he esul s pu ely o he measu emen componen o he SEM o each compliance domain, as in Equa ion (1) (bu esul ing om
es ima ing bo h equa ions simul aneously). Coe icien es ima es a e s anda dized and es ima ed using diagonal weigh ed leas squa es and a polycho ic
co ela ion ma ix. See Table 2 o he i s a is ics o he ull SEM.
*p<0.05, **p<0.01, ***p<0.001.
TABLE A11 Compliance: Summa ed a ing scale eliabili y analysis.
I em C onbach's α α I i em is dele ed
Scale: Compliance in he public domain 0.797
Wea ing a mask in c owded places 0.765
Keeping a 1.5 m dis ance in public spaces (whene e possible) 0.712
A oiding handshakes 0.693
Scale: Compliance in he p i a e domain 0.780
Lea ing he home only when absolu ely necessa y 0.709
Gene ally a oiding o he households 0.642
Res ic ing p i a e mee ings o 1 pe son om ano he household 0.753
Co ela ion be ween scales: 0.643
Obse a ions: 3340
No e: The able shows he esul s o a eliabili y analysis (i.e., o e all C onbach's αand C onbach's αwhen d opping a ce ain i em) o wo summa ed a ing
scales using he wo i em iple s as indica ed in he able and iden i ied in he explo a o y ac o analysis in Table 1.
TABLE A13 Decision‐making ac oss compliance domains—ex ended s uc u al equa ion model (SEM) esul s (also showing
coe icien s o con ol a iables).
Ou come Compliance in he public domain Compliance in he p i a e domain
COVID‐19 h ea pe cep ion 0.297*** 0.365***
(0.026) (0.017)
Al uism 0.037 0.012
(0.031) (0.020)
(Con inues)
STERNBERG ET AL.
-
1085

TABLE A13 (Con inued)
Ou come Compliance in he public domain Compliance in he p i a e domain
Ci ic esponsibili y 0.127*** 0.108***
(0.024) (0.016)
Pos. ecip oci y 0.180*** 0.030
(0.024) (0.017)
Neg. ecip oci y −0.149*** −0.086***
(0.029) (0.019)
Risk a e sion 0.065** 0.055**
(0.034) (0.022)
Pa ience −0.001 0.028
(0.031) (0.020)
T us in RKI 0.204*** 0.068**
(0.032) (0.021)
T us in go e nmen 0.012 0.138***
(0.033) (0.021)
Female 0.117*** 0.039*
(0.025) (0.016)
Age g oup 0.057* 0.135***
(0.027) (0.018)
High educa ion −0.002 0.019
(0.030) (0.020)
Medium educa ion 0.027 0.016
(0.028) 0.018
Employed essen ial se ices 0.020 −0.021
(0.024) (0.016)
Household income 0.003 −0.034
(0.030) (0.019)
Household size −0.044 −0.013
(0.028) (0.018)
Knowledge p e en i e measu es 0.117*** 0.087***
(0.025) (0.016)
Social desi abili y index 0.092*** 0.102***
(0.024) (0.016)
R
2
0.492 0.411
Obse a ions 2918 2918
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o he esul s o he SEMs es ima ed by means o Equa ions (1) and
(2) (shown a e only he s uc u al componen esul s, see Table A12 o he esul s o he measu emen componen ). The SEM was es ima ed sepa a ely o
each compliance domain, using Diagonal Weigh ed Leas Squa es, a polycho ic co ela ion ma ix, obus s anda d e o s and a co ec ed es s a is ic. All
es ima ions mo eo e con ol o he esponden s' s a e o esidence. Table A14 epea s he same analyses wi h he inal‐model sample o N=2922
h oughou . See Tables A1, A15 and A21 and Figu e A8 o he sou ce su ey i ems and ac o sco es o he social desi abili y bias index.
*p<0.05, **p<0.01, ***p<0.001.
1086
-
STERNBERG ET AL.
TABLE A14 Decision‐making ac oss compliance domains—s uc u al equa ion model (SEM) esul s (same obse a ions in all submodels).
Ou come
Compliance in he public domain Compliance in he p i a e domain
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
COVID‐19 h ea pe cep ion 0.464*** 0.321*** 0.304*** 0.314*** 0.297*** 0.502*** 0.387*** 0.372*** 0.381*** 0.365***
(0.022) (0.025) (0.026) (0.025) (0.026) (0.016) (0.017) (0.017) (0.017) (0.017)
Al uism 0.164*** 0.060** 0.044 0.053* 0.037 0.127*** 0.030 0.017 0.024 0.012
(0.028) (0.030) (0.031) (0.030) (0.031) (0.019) (0.019) (0.020) (0.019) (0.020)
Ci ic esponsibili y 0.196*** 0.172*** 0.147*** 0.151*** 0.127*** 0.202*** 0.169*** 0.127*** 0.148*** 0.108***
(0.023) (0.023) (0.024) (0.023) (0.024) (0.016) (0.016) (0.016) (0.016) (0.016)
Pos. ecip oci y 0.248*** 0.218*** 0.205*** 0.193*** 0.180*** 0.115*** 0.066** 0.056** 0.039* 0.030
(0.023) (0.023) (0.024) (0.024) (0.024) (0.018) (0.017) (0.017) (0.017) (0.017)
Neg. ecip oci y −0.193*** −0.184*** −0.170*** −0.165*** −0.149*** −0.125*** −0.109*** −0.103*** −0.092*** −0.086***
(0.027) (0.028) (0.029) (0.028) (0.029) (0.019) (0.019) (0.019) (0.019) (0.019)
Risk a e sion 0.169*** 0.085*** 0.066** 0.085*** 0.065** 0.168*** 0.080*** 0.055** 0.081*** 0.055**
(0.031) (0.033) (0.034) (0.033) (0.034) (0.022) (0.021) (0.022) (0.021) (0.022)
Pa ience 0.141*** −0.014 0.004 −0.019 −0.001 0.115*** 0.010 0.032 0.006 0.028
(0.027) (0.029) (0.031) (0.029) (0.031) (0.019) (0.019) (0.020) (0.019) (0.020)
T us in RKI 0.385*** 0.223*** 0.216*** 0.210*** 0.204*** 0.239*** 0.087** 0.075** 0.079** 0.068**
(0.030) (0.031) (0.032) (0.031) (0.032) (0.021) (0.021) (0.021) (0.021) (0.021)
T us in go e nmen 0.042 0.014 0.017 0.008 0.012 0.186*** 0.137*** 0.145*** 0.129*** 0.138***
(0.031) (0.031) (0.033) (0.031) (0.033) (0.022) (0.020) (0.021) (0.020) (0.021)
Demog . and socioecon. con ols No No No No No Yes No Yes No No No No No Yes No Yes
Compliance‐speci ic con ols No No No No No No Yes Yes No No No No No No Yes Yes
R
2
0.215 0.222 0.043 0.173 0.448 0.477 0.464 0.492 0.252 0.116 0.037 0.155 0.366 0.396 0.382 0.411
Obse a ions 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918 2918
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o he esul s o he SEMs es ima ed by means o Equa ions (1) and (2) (shown a e only he s uc u al componen esul s, see
Table A12 o he esul s o he measu emen componen ), eplica ing Table 2bu using he same obse a ions o all es ima ions. The SEM was es ima ed sepa a ely o each compliance domain, using Diagonal
Weigh ed Leas Squa es, a polycho ic co ela ion ma ix, obus s anda d e o s and a co ec ed es s a is ic. Demog aphic con ols con ain esponden s' gende , age g oup and s a e. Socioeconomic con ols include
educa ion, employmen in essen ial se ices, household size and income. Compliance‐speci ic con ols con ain knowledge abou COVID‐19 p e en i e measu es and he deg ee o social desi abili y bias ( ac o sco es
as in Figu e A8 and Table A15).
*p<0.05, **p<0.01, ***p<0.001.
STERNBERG ET AL.
-
1087
TABLE A15 Social desi abili y:
Fac o loadings.
Fac o 1
Social desi abili y i ems (posi i e)
Uncondi ional objec i i y in dispu es 0.658
Uncondi ional kindness unde s ess 0.772
Uncondi ional a en ion in con e sa ions 0.564
Eigen alue 1.347
Pe cen sha ed a iance accoun ed o 44.90
Mul iple R
2
o sco es wi h ac o s 0.730
Obse a ions: 3339
No e: The able shows s anda dized ac o pa e n coe icien s om a 1‐ ac o solu ion o he su ey
i ems aimed a cap u ing social desi abili y bias (exagge a ing posi i e cha ac e is ics). Es ima ed using
i e a ed p incipal axis ac o ing, no o a ion and polycho ic co ela ions.
TABLE A16 Co ela ion ma ix o co e explana o y a iables.
CV19‐ h ea Pos. Reci. Neg. Reci. Al uism Ci . Resp. Risk a e sion Pa ience T us (RKI) T us (Go .)
CV19‐ h ea 1.00
Pos. Reci. 0.16*** 1.00
Neg. Reci. 0.02 −0.03 1.00
Al uism 0.18*** 0.24*** −0.03 1.00
Ci . Resp. 0.08*** 0.13*** −0.19*** 0.12*** 1.00
Risk a e sion 0.08*** −0.05** −0.21*** −0.12*** 0.12*** 1.00
Pa ience 0.06** 0.14*** 0.02 0.22*** 0.05** −0.11*** 1.00
T us (RKI) 0.38*** 0.14*** −0.04* 0.23*** 0.10*** −0.00 0.20*** 1.00
T us (Go .) 0.33*** 0.06*** −0.03 0.21*** 0.06** −0.02 0.20*** 0.71*** 1.00
Sou ce: Shown a e Pea son co ela ions be ween he main explana o y a iables o he SEM, ha is, he inal p e e ence measu es as well as he COVID‐19
h ea pe cep ion measu e.
*p<0.05, **p<0.01, ***p<0.001.
TABLE A17 Al e na i e
compliance measu e: S uc u al
equa ion model (SEM) measu emen
componen (con i ma o y ac o
analysis).
La en a iable Al e na i e compliance measu e
Sledding ip ( e e se coded) 0.896
(cons .)
Su p ise bi hday ca oling ( e e se coded) 0.830***
(0.018)
Dinne o mo ie nigh ( e e se coded) 0.932***
(0.024)
Obse a ions =2141
No e: The able pu ely shows he esul s o he measu emen componen o he SEM, as in Equa ion (1)
(bu esul ing om es ima ing bo h equa ions simul aneously). Howe e , ins ead o he ini ial
compliance i ems, we used h ee al e na i e su ey i ems as e lec i e indica o s, as displayed in he
able (see Sec ion 4.3 and Tables A1 and A21 o de ailed desc ip ions). Coe icien es ima es a e
s anda dized and es ima ed using diagonal weigh ed leas squa es and a polycho ic co ela ion ma ix.
*p<0.05, **p<0.01, ***p<0.001.
1088
-
STERNBERG ET AL.
TABLE A18 Robus ness: Decision‐making ac oss compliance domains—s uc u al equa ion model (SEM) esul s wi h al e na i e
compliance measu e.
Ou come: Compliance (al e na i e measu e)
COVID‐19 h ea pe cep ion 0.319***
(0.025)
Al uism 0.025
(0.030)
Ci ic esponsibili y 0.173***
(0.024)
Pos. ecip oci y −0.053*
(0.026)
Neg. ecip oci y −0.164***
(0.028)
Risk a e sion 0.099***
(0.031)
Pa ience 0.010
(0.030)
T us in RKI 0.156***
(0.033)
T us in go e nmen 0.079**
(0.034)
Female −0.034
(0.023)
Age g oup 0.207***
(0.026)
High educa ion 0.013
(0.029)
Medium educa ion 0.009
(0.027)
Employed essen ial se ices −0.040*
(0.022)
Household income 0.014
(0.029)
Household size 0.008
(0.026)
Knowledge p e en i e measu es 0.073***
(0.023)
Social desi abili y index −0.014
(0.023)
(Con inues)
STERNBERG ET AL.
-
1089
TABLE A18 (Con inued)
Ou come: Compliance (al e na i e measu e)
R
2
0.469
Fi s a is ics ( ull SEM)
Robus χ
2
(66) =126.907; p<0.001
Robus RMSEA =0.021
Robus TLI =0.999
Robus CFI =0.987
SRMR =0.007
R
2
=0.469
Obse a ions =2141
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o he esul s o he SEM es ima ed by means o Equa ions (1) and
(2) (shown a e only he s uc u al componen esul s, see Table A17 o he esul s o he measu emen componen ) and i s a is ics o he ull SEM.
Howe e , ins ead o he ini ial compliance i ems, we used h ee al e na i e su ey i ems as e lec i e indica o s (see Sec ion 4.3 and Tables A1 and A21 o
de ailed desc ip ions). The SEM was es ima ed using Diagonal Weigh ed Leas Squa es, a polycho ic co ela ion ma ix, obus s anda d e o s and a
co ec ed es s a is ic. All es ima ions mo eo e con ol o he esponden s' s a e o esidence. See Tables A1, A15 and A21 and Figu e A8 o he sou ce
su ey i ems and ac o sco es o he social desi abili y bias index.
Abb e ia ions: CFI, compa a i e i index; RMSEA, oo mean squa e e o o app oxima ion; SRMR, s anda dized oo mean squa e esidual; TLI,
Tucke ‐Lewis index.
*p<0.05, **p<0.01, ***p<0.001.
TABLE A19 Robus ness: Decision‐making ac oss compliance domains—s uc u al equa ion model (SEM) esul s wi h media ion.
Model
Compliance in he public domain Compliance in he p i a e domain
Di ec e ec Indi ec e ec To al e ec Di ec e ec Indi ec e ec To al e ec
COVID‐19 h ea pe cep ion 0.286*** ‐ 0.286*** 0.359*** ‐ 0.359***
(0.021) (0.021) (0.017) (0.017)
Al uism 0.044 0.019*** 0.063** 0.013 0.024*** 0.037
(0.031) (0.007) (0.031) (0.024) (0.008) (0.026)
Ci ic esponsibili y 0.121*** −0.004 0.105*** 0.110*** −0.005 0.100***
(0.024) (0.006) (0.024) (0.019) (0.007) (0.021)
Pos. ecip oci y 0.172*** 0.021*** 0.193*** 0.027 0.026*** 0.053*
(0.024) (0.006) (0.025) (0.021) (0.007) (0.022)
Neg. ecip oci y −0.142*** 0.020*** −0.122*** −0.085*** 0.025*** −0.060**
(0.028) (0.007) (0.029) (0.023) (0.008) (0.025)
Risk a e sion 0.069** 0.028*** 0.097*** 0.056** 0.035*** 0.092***
(0.033) (0.008) (0.034) (0.026) (0.009) (0.028)
Pa ience 0.003 −0.003 0.000 0.029 −0.004 0.025
(0.030) (0.007) (0.031) (0.024) (0.008) (0.026)
T us in RKI 0.201*** 0.068*** 0.270*** 0.069** 0.086*** 0.156***
(0.031) (0.009) (0.032) (0.026) (0.009) (0.027)
T us in go e nmen 0.031 0.038*** 0.069* 0.144*** 0.047*** 0.192***
(0.032) (0.008) (0.033) (0.026) (0.009) (0.027)
1090
-
STERNBERG ET AL.

TABLE A19 (Con inued)
Model
Compliance in he public domain Compliance in he p i a e domain
Di ec e ec Indi ec e ec To al e ec Di ec e ec Indi ec e ec To al e ec
Female 0.116*** 0.031*** 0.147*** 0.039* 0.039*** 0.078***
(0.024) (0.006) (0.025) (0.020) (0.007) (0.021)
Age g oup 0.057* 0.020*** 0.078** 0.138*** 0.026*** 0.163***
(0.027) (0.006) (0.027) (0.021) (0.007) (0.023)
High educa ion −0.003 −0.011 −0.014 0.020 −0.013 0.007
(0.030) (0.007) (0.030) (0.024) (0.008) (0.026)
Medium educa ion 0.023 −0.002 0.021 0.016 −0.003 0.013
(0.027) (0.006) (0.028) (0.022) (0.007) (0.024)
Employed essen ial se ices 0.019 −0.016** 0.003 −0.020 −0.020** −0.040*
(0.023) (0.006) (0.024) (0.020) (0.007) (0.021)
Household income 0.009 −0.005 0.004 −0.033 −0.007 −0.039
(0.029) (0.007) (0.030) (0.024) (0.008) (0.025)
Household size −0.046 0.015** −0.030 −0.015 0.019 0.004
(0.027) (0.006) (0.028) (0.022) (0.008) (0.024)
Knowledge measu es 0.130*** ‐ 0.130*** 0.102*** ‐ 0.102***
(0.025) ‐ (0.025) (0.021) ‐ (0.021)
Social desi abili y index 0.099*** 0.099*** 0.110*** ‐ 0.110***
(0.024) ‐ (0.024) (0.021) ‐ (0.021)
Fi s a is ics ( ull SEM) Public P i a e
Robus χ
2
(68) =167.704; p<0.001 Robus χ
2
(68) =116.578; p<0.001
Robus RMSEA =0.022 Robus RMSEA =0.016
Robus TLI =0.997 Robus TLI =0.999
Robus CFI =0.965 Robus CFI =0.988
SRMR =0.025 SRMR =0.013
R
2
=0.487 R
2
=0.412
Obse a ions =2918 Obse a ions =2918
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o es ima ing he SEMs by means o Equa ions (1) and (2), and
addi ionally allowing o a media ing e ec o COVID‐19 h ea pe cep ions. Shown a e he esul ing es ima ed di ec , indi ec and indi ec e ec s (i.e., he
esul s o he s uc u al componen ) as well as i s a is ics o he ull SEM. The SEM was es ima ed sepa a ely o each compliance domain, using Diagonal
Weigh ed Leas Squa es, a polycho ic co ela ion ma ix, obus s anda d e o s and a co ec ed es s a is ic. All es ima ions mo eo e con ol o he
esponden s' s a e o esidence. See Tables A1, A15 and A21 and Figu e A8 o he sou ce su ey i ems and ac o sco es o he social desi abili y bias index.
*p<0.05, **p<0.01, ***p<0.001.
STERNBERG ET AL.
-
1091
TABLE A20 Robus ness: Decision‐making ac oss compliance domains—s uc u al equa ion model (SEM) esul s using single‐i em
p e e ence measu es.
Ou come Compliance in he public domain Compliance in he p i a e domain
COVID‐19 h ea pe cep ion 0.290*** 0.361***
(0.026) (0.018)
Al uism 0.026 0.006
(0.025) (0.016)
Ci ic esponsibili y 0.122*** 0.104***
(0.024) (0.016)
Pos. ecip oci y 0.179*** 0.032
(0.024) (0.017)
Neg. ecip oci y −0.155*** −0.088***
(0.024) (0.016)
Risk a e sion 0.094*** 0.081***
(0.026) (0.017)
Pa ience −0.033 −0.010
(0.025) (0.016)
T us in RKI 0.200*** 0.066**
(0.032) (0.021)
T us in go e nmen 0.016 0.144***
(0.033) (0.021)
Female 0.110*** 0.032
(0.025) (0.016)
Age g oup 0.050* 0.128***
(0.027) (0.018)
High educa ion 0.003 0.028
(0.030) (0.020)
Medium educa ion 0.023 0.017
(0.028) (0.018)
Employed essen ial se ices 0.020 −0.019
(0.024) (0.016)
Household income 0.016 −0.024
(0.029) (0.019)
Household size −0.048 −0.014
(0.028) (0.018)
Knowledge p e en i e measu es 0.114*** 0.084***
(0.025) (0.016)
Social desi abili y index 0.086*** 0.098***
(0.024) (0.016)
1092
-
STERNBERG ET AL.
TABLE A20 (Con inued)
Ou come Compliance in he public domain Compliance in he p i a e domain
R
2
0.500 0.414
Obse a ions 2918 2918
Fi s a is ics ( ull SEM) Public P i a e
Robus χ
2
(66) =160.35; p<0.001 Robus χ
2
(66) =114.982; p<0.001
Robus RMSEA =0.022 Robus RMSEA =0.016
Robus TLI =0.998 Robus TLI =0.999
Robus CFI =0.956 Robus CFI =0.983
SRMR =0.019 SRMR =0.009
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o he esul s o he SEMs es ima ed by means o Equa ions (1) and
(2) (shown a e only he s uc u al componen esul s) and i s a is ics o each SEM. Howe e , ins ead o he ini ially cons uc ed p e e ence measu es, we
used only a single su ey i em o each ype o p e e ence o assess he obus ness o he singula su ey i em o posi i e ecip oci y. Fo each p e e ence ype,
we used he su ey i em wi h he highes weigh in he expe imen al alida ion p ocedu e de eloped by Falk e al. (2018). See Table A4 o hese i ems, he
espec i e i ems a e ma ked in i alic on . The SEM was es ima ed sepa a ely o each compliance domain, using Diagonal Weigh ed Leas Squa es, a
polycho ic co ela ion ma ix, obus s anda d e o s and a co ec ed es s a is ic. All es ima ions mo eo e con ol o he esponden s s a e o esidence. See
Tables A1, A15 and A21 and Figu e A8 o he sou ce su ey i ems and ac o sco es o he social desi abili y bias index.
Abb e ia ions: CFI, compa a i e i index; RMSEA, oo mean squa e e o o app oxima ion; SRMR, s anda dized oo mean squa e esidual; TLI,
Tucke ‐Lewis index.
*p<0.05, **p<0.01, ***p<0.001.
TABLE A21 Ques ionnai e i ems ( ansla ed om o iginal Ge man e sion).
Ques ion (Va iable name as in Table A1) Response op ions
Compliance
Below you ind a lis o beha io al pa e ns. How well does
each e lec you beha io in he pas 2 weeks? All
in o ma ion you p o ide in he ques ionnai e will be ea ed
anonymously and canno be linked o you iden i y.
I ha e kep a dis ance o a leas 1.50 m o o he people in
public, whene e possible. (Keeping dis ance)
1: Applied ne e ; 2: Applied a ely; 3: Applied some imes; 4: Applied
o en; 5: Applied always
I ha e wo n a mask in public anspo o when shopping.
(Wea ing masks)
1: Applied ne e ; 2: Applied a ely; 3: Applied some imes; 4: Applied
o en; 5: Applied always
I ha e only le he house when absolu ely necessa y (such as
g oce ies, medical easons o spo s). (Only lea ing home
when necessa y)
1: Applied ne e ; 2: Applied a ely; 3: Applied some imes; 4: Applied
o en; 5: Applied always
I ha e delibe a ely a oided physical con ac wi h people om
o he households. (A oiding o he households)
1: Applied ne e ; 2: Applied a ely; 3: Applied some imes; 4: Applied
o en; 5: Applied always
I ha e es ic ed p i a e ga he ings o only one o he pe son
om ano he household. (Res ic ing p i a e mee ings)
1: Applied ne e ; 2: Applied a ely; 3: Applied some imes; 4: Applied
o en; 5: Applied always
I ha e a oided shaking o he people's hands when g ee ing
hem. (A oiding handshakes)
1: Applied ne e ; 2: Applied a ely; 3: Applied some imes; 4: Applied
o en; 5: Applied always
Compliance (al e na i e measu es)
When answe ing he nex ques ions, please emembe ha all
he in o ma ion you p o ide in he ques ionnai e will be
ea ed anonymously by us and canno be linked o you
iden i y. How would you ac in he ollowing si ua ions?
I i snows hea ily nex weekend and some iends o you s
o ganize a sledding ip ( he e is s ill oom o you on hei
Scale om 1 (De ini ely no ) o 7 (Ce ainly); 99 (I am gene ally no e y
en husias ic abou his ype o ac i i y.)
(Con inues)
STERNBERG ET AL.
-
1093
TABLE A21 (Con inued)
Ques ion (Va iable name as in Table A1) Response op ions
sleds), would you join hem? (Willingness o join sledding
ip wi h iends nex weekend)
I a iend in i ed you o ha e dinne wi h h ee o he iends
nex weekend and hen wa ch socce o a mo ie, would you
accep he in i a ion? (Willingness o join su p ise bi hday
ca oling o a iend)
Scale om 1 (De ini ely no ) o 7 (Ce ainly); 99 (I am gene ally no e y
en husias ic abou his ype o ac i i y.)
Suppose one o mo e o you bes iends has a bi hday nex
week. A g oup o iends is planning o ge oge he in he
mo ning o se enade he bi hday gi l o boy a he doo o
unde he window. Will you join hem? (Willingness o join a
dinne /mo ie nigh a iends' house)
Scale om 1 (De ini ely no ) o 7 (Ce ainly); 99 (I am gene ally no e y
en husias ic abou his ype o ac i i y.)
Economic p e e ences
We now ask o you willingness o ac in a ce ain way in
di e en a eas. Please indica e you answe on a scale om
0 o 10, whe e 0 means you a e “comple ely unwilling o do
so” and a 10 means you a e “ e y willing o do so”.
How willing a e you o gi e up some hing ha is bene icial o
you oday in o de o bene i mo e om ha in he u u e?
(Willingness o wai )
Scale om 0 (comple ely unwilling o do so) o 10 ( e y willing o do so)
How willing a e you o punish someone who ea s you
un ai ly, e en i he e may be cos s o you? (Willingness o
punish un ai beha io owa d sel )
Scale om 0 (comple ely unwilling o do so) o 10 ( e y willing o do so)
How willing a e you o punish someone who ea s o he s
un ai ly, e en i he e may be cos s o you? (Willingness o
punish un ai beha io owa d o he s)
Scale om 0 (comple ely unwilling o do so) o 10 ( e y willing o do so)
How willing a e you o gi e o good causes wi hou expec ing
any hing in e u n? (Willingness o gi e o good causes)
Scale om 0 (comple ely unwilling o do so) o 10 ( e y willing o do so)
How well do he ollowing s a emen s desc ibe you as a pe son?
Please indica e you answe on a scale om 0 o 10. A
0 means “does no desc ibe me a all” and a 10 means
“desc ibes me pe ec ly”.
When someone does me a a o , I am willing o e u n i .
(Willingness o e u n a a o )
Scale om 0 (does no desc ibe me a all) o 10 (desc ibes me pe ec ly)
I I am ea ed e y unjus ly, I will ake e enge a he e y i s
occasion, e en i he e is a cos o do so. (Willingness o ake
e enge)
Scale om 0 (does no desc ibe me a all) o 10 (desc ibes me pe ec ly)
In gene al, how willing o unwilling you a e o ake isks?
Please use a scale om 0 o 10, whe e 0 means you a e
“comple ely unwilling o ake isks” and a 10 means you a e
“ e y willing o ake isks”. (Willingness o ake isks)
Scale om 0 (comple ely unwilling o ake isks) o 10 ( e y willing o
ake isks)
By answe ing he ollowing ques ions, you ha e he chance o
win a bonus paymen o 10 eu os. A e comple ing he
su ey, 5% o he pa icipan s (i.e., 1 ou o 20) will be
andomly selec ed o ecei e his bonus paymen o 10
eu os.
You ha e he op ion o dona e any po ion o he bonus
paymen o a cha i able o ganiza ion, while you would
ecei e he emaining amoun om Respondi. I you decide
o make a dona ion, we will do so a e he su ey is
1094
-
STERNBERG ET AL.
TABLE A26 Robus ness check Google mobili y pa e ns: Re ail and ec ea ion and ansi .
Ou come: Mobili y changes in
Re ail and ec ea ion a eas T ansi a eas
(1) (2) (3) (4) (5) (6) (7) (8)
P i a e compliance −16.058** −14.030* −10.410 −31.044** −24.835 −21.600
(4.772) (6.348) (5.331) (9.540) (12.523) (12.965)
Public compliance −6.807* −1.797 −2.769 −14.370* −5.502 −7.957
(3.118) (3.571) (3.095) (5.986) (7.044) (7.527)
Popula ion densi y (inhabi an s/km) −0.003 −0.004
(0.001) (0.003)
S ad s aa 3.193 8.649
(3.817) (9.283)
Cons an −46.774*** −45.191*** −44.924*** −44.375*** −25.619** −20.430 −19.958 −16.525
(3.165) (5.545) (4.908) (4.45349) (6.328) (10.644) (9.681) (10.832)
Adj. R
2
0.408 0.201 0.374 0.584 0.584 0.241 0.373 0.366
Obse a ions 16 16 16 16 16 16 16 16
No e: Shown a e he esul s o eg essing (i) s a e‐le el phone‐ acking‐based changes in mobili y acco ding o Google's mobili y epo s in he a eas o Re ail
and Rec ea ion (columns 2–5) and T ansi (columns 6–9) and on (ii) s a e‐le el a e ages o ou measu es o compliance in he public and p i a e domain, as
p edic ed om he SEM by means o Equa ions (1) and (2) (con olling o a s a e's popula ion densi y and whe he he s a e is a ci y s a e, i.e., Hambu g,
Be lin and B emen). Mobili y da a e ie ed om Google LLC (2021). Changes in mobili y e e o he 2020 baseline pe iod be o e he s a o he pandemic.
*p<0.05, **p<0.01, ***p<0.001.
TABLE A27 Robus ness check Google mobili y pa e ns: Wo kplace and esiden ial.
Ou come: Mobili y changes in
Wo kplace a eas Residen ial a eas
(1) (2) (3) (4) (5) (6) (7) (8)
P i a e compliance −14.731* −15.859 −9.420** 5.756** 7.613** 6.109***
(5.726) (7.675) (1.723) (2.167) (1.374)
Public compliance −4.664 0.999 0.083 1.073 −1.645 −1.134
(3.705) (4.318) (1.570) (1.267) (1.219) (0.798)
Popula ion densi y (inhabi an s/km) −0.004*** 0.001**
(0.001) (0.000)
S ad s aa 2.711 −1.721
(1.936) (0.984)
Cons an −18.931* −20.261** −19.959** −20.490*** 8.366*** 10.204*** 10.059*** 9.631***
(3.798) (6.588) (5.934) (2.259) (1.143) (2.253) (1.675) (1.148)
Adj. R
2
0.273 0.038 0.220 0.909 0.404 −0.019 0.437 0.787
Obse a ions 16 16 16 16 16 16 16 16
No e: Shown a e he esul s o eg essing (i) s a e‐le el phone‐ acking‐based changes in mobili y acco ding o Google's mobili y epo s in he a eas o
Wo kplace (columns 2–5) and Residen ial (columns 6–9) and on (ii) s a e‐le el a e ages o ou measu es o compliance in he public and p i a e domain, as
p edic ed om he SEM by means o Equa ions (1) and (2) (con olling o a s a e's popula ion densi y and whe he he s a e is a ci y s a e, i.e., Hambu g,
Be lin and B emen). Mobili y da a e ie ed om Google LLC (2021). Changes in mobili y e e o he 2020 baseline pe iod be o e he s a o he pandemic.
*p<0.05, **p<0.01, ***p<0.001.
STERNBERG ET AL.
-
1101

TABLE A29 Ex ension ins i u ional us : Unde s anding e ec s o us in go e nmen .
Ou come
Public compliance P i a e compliance
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
T us in go e nmen 0.012 −0.007 0.030 0.138*** 0.145*** 0.118***
(0.033) (0.021) (0.019) (0.021) (0.013) (0.012)
T us in s a e go . 0.031 0.034 0.070** −0.010
(0.017) (0.022) (0.011) (0.013)
T us in es ab. media −0.017 −0.026 0.081*** 0.045*
(0.017) (0.019) (0.010) (0.011)
T us in RKI 0.204*** 0.190*** 0.194*** 0.219*** 0.207*** 0.068** 0.112*** 0.124*** 0.070** 0.062*
(0.032) (0.017) (0.018) (0.015) (0.017) (0.021) (0.011) (0.010) (0.012) (0.011)
O he main beha io al p edic o s Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Demog . and socioecon. con ols Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Compliance‐speci ic con ols Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
R
2
0.492 0.490 0.491 0.492 0.494 0.411 0.406 0.413 0.407 0.413
Obse a ions 2918 2917 2917 2918 2918 2918 2917 2917 2918 2918
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o he ins i u ional us a iables o he SEM, as es ima ed by
means o Equa ions (1) and (2) (shown a e only he s uc u al componen esul s). Columns 1 and 6 he e o e me ely epea he esul s om Table 2. The
SEM was es ima ed sepa a ely o each compliance domain, using Diagonal Weigh ed Leas Squa es, a polycho ic co ela ion ma ix, obus s anda d e o s
and a co ec ed es s a is ic. O he main beha io al p edic o s con ain COVID‐19 h ea pe cep ion, pos. and neg. ecip oci y, al uism, ci ic esponsibili y,
isk a e sion and pa ience. Demog aphic con ols con ain esponden s' gende , age g oup and s a e. Socioeconomic con ols include educa ion, employmen
in essen ial se ices, household size and income. Compliance‐speci ic con ols con ain knowledge abou COVID‐19 p e en i e measu es and he deg ee o
social desi abili y bias ( ac o sco es as in Figu e A8 and Table A15).
*p<0.05, **p<0.01, ***p<0.001.
TABLE A28 Ex ension ins i u ional us : P oxies o us in Robe Koch Ins i u e (RKI).
Ou come
Public compliance P i a e compliance
(1) (2) (3) (4) (5) (6)
T us in RKI 0.204*** 0.068**
(0.032) (0.021)
T us in science 0.182*** 0.053*
(0.018) 0.012
T us in WHO 0.165*** 0.055*
(0.018) (0.011)
T us in go e nmen 0.012 0.050 0.044 0.138*** 0.153*** 0.147***
(0.033) (0.016) (0.017) (0.021) (0.010) (0.011)
O he main beha io al p edic o s Yes Yes Yes Yes Yes Yes
Demog . and socioecon. con ols Yes Yes Yes Yes Yes Yes
Compliance‐speci ic con ols Yes Yes Yes Yes Yes Yes
R
2
0.492 0.493 0.490 0.411 0.411 0.410
Obse a ions 2918 2918 2918 2918 2918 2918
No e: Displayed a e s anda dized coe icien es ima es and s anda d e o s in pa en heses o he ins i u ional us a iables o he SEM, as es ima ed by
means o Equa ions (1) and (2) (shown a e only he s uc u al componen esul s). Columns 1 and 4 he e o e me ely epea he esul s om Table 2. The
SEM was es ima ed sepa a ely o each compliance domain, using Diagonal Weigh ed Leas Squa es, a polycho ic co ela ion ma ix, obus s anda d e o s
and a co ec ed es s a is ic. O he main beha io al p edic o s con ain COVID19 h ea pe cep ion, pos. and neg. ecip oci y, al uism, ci ic esponsibili y,
isk a e sion and pa ience. Demog aphic con ols con ain esponden s' gende , age g oup and s a e. Socioeconomic con ols include educa ion, employmen
in essen ial se ices, household size and income. Compliance‐speci ic con ols con ain knowledge abou COVID‐19 p e en i e measu es and he deg ee o
social desi abili y bias ( ac o sco es as in Figu e A8 and Table A15).
*p<0.05, **p<0.01, ***p<0.001.
1102
-
STERNBERG ET AL.
TABLE A30 Fines o non‐compliance ac oss Ge man s a es.
S a e
Beha io
Da e o
egula ions Sou ce
No wea ing a
ace‐mask
No keeping dis ance
o 1.50 m
Mee ing wi h mo e han allowed
numbe o people
Hos ing an illegal
ga he ing
Baden‐
Wü embe g
50–250 Eu o n.a. n.a. 100–1000 Eu o Sep embe 15,
2021
h ps://www.bussgeldka alog.o g/co ona‐
baden‐wue embe g/
Baye n 250 Eu o 150 Eu o 150 Eu o 5000 Eu o No embe 20,
2020
h ps://www.bussgeldka alog.o g/co ona‐
baye n/
Be lin 50–500 Eu o 100–500 Eu o 25–500 Eu o n.a. No embe 20,
2020
h ps://www.bussgeldka alog.o g/co ona‐
be lin/
B andenbu g 50–250 Eu o n.a. 50–250 Eu o 1000–5000 Eu o Augus 25,
2020
h ps://www.bussgeldka alog.o g/co ona‐
b andenbu g/
B emen n.a. n.a. 50–250 Eu o 250–2500 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
b emen/
Hambu g 150 Eu o 150 Eu o 150–500 Eu o 1000 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
hambu g/
Hessen 200 Eu o n.a. 200 Eu o 500–1000 Eu o Ap il 20, 2020 h ps://www.bussgeldka alog.o g/co ona‐
hessen/
Mecklenbu g‐
Vo pomme n
50–150 Eu o n.a. 50–500 Eu o n.a. No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
mecklenbu g‐ o pomme n/
Niede sachsen 100–150 Eu o 100–400 Eu o 150–400 Eu o 300–3000 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
niede sachsen/
No d hein‐
Wes alen
50–150 Eu o n.a. 250 Eu o 1000–5000 Eu o Sep embe 6,
2020
h ps://www.bussgeldka alog.o g/
co ona‐n w/
Rheinland‐P alz 50 Eu o 50 Eu o 100 Eu o 500 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
heinland‐p alz/
(Con inues)
STERNBERG ET AL.
-
1103
TABLE A30 (Con inued)
S a e
Beha io
Da e o
egula ions Sou ce
No wea ing a
ace‐mask
No keeping dis ance
o 1.50 m
Mee ing wi h mo e han allowed
numbe o people
Hos ing an illegal
ga he ing
Saa land 50–100 Eu o n.a. 200 Eu o 1000–4000 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
saa land/
Sachsen 60 Eu o 150 Eu o 150 Eu o 5000 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
sachsen/
Sachsen‐Anhal 50–75 Eu o n.a. 50 Eu o 1000 Eu o Sep embe 14,
2021
h ps://www.bussgeldka alog.o g/co ona‐
sachsen‐anhal /
Schleswig‐
Hols ein
150 Eu o 150 Eu o 150 Eu o 1000–2000 Eu o No embe 23,
2020
h ps://www.bussgeldka alog.o g/co ona‐
schleswig‐hols ein/
Thü ingen 60 Eu o 100 Eu o n.a. 1000–3000 Eu o No embe 20,
2020
h ps://www.bussgeldka alog.o g/co ona‐
hue ingen/
No e: The able shows ines o non‐compliance wi h COVID‐19 egula ions ac oss ede al s a es in Ge many. No e ha he ime a which egula ions came in o place di e ac oss s a es, o example, in
Baden‐Wü embe g and Sachsen‐Anhal , we we e only able o ob ain in o ma ion abou he egula ions a he la e s ages in he pandemic, such ha hey did no anymo e include ines o mos ules ha we e
ele an du ing ou su ey da a collec ion. “n.a.” indica es ha we we e no able o ob ain in o ma ion abou he amoun o he speci ic ine in his s a e, gi en he dynamically changing condi ions du ing he
pandemic and di icul ies o ind in o ma ion abou egula ions ha a e no longe in place. Cohe en ly, an “n.a.” en y does no mean ha he e was no ine o he espec i e beha io .
1104
-
STERNBERG ET AL.
FIGURE A2 COVID‐19 h ea pe cep ion: sc eeplo ac o analysis. This igu e shows a sc eeplo o an explo a o y ac o analysis o
he 8 su ey i ems aimed a cap u ing COVID‐19 h ea pe cep ion (see Tables A1, A6 and A21 o he i ems and p ocedu e), wi h
Eigen alues o ac o s on he y‐axis and he iden i ied ac o s on he x‐axis. The ac o analysis was pe o med using i e a ed p incipal axis
ac o ing, no o a ion, and a mixed co ela ion ma ix (con inuous and polycho ic).
FIGURE A1 Dis ibu ion o sel ‐ epo ed compliance i ems. This igu e shows he scaled densi y dis ibu ion o each o he six su ey
i ems aimed o quan i y compliance beha io . The bold line indica es he mean and he non‐bold line he median in each case. Pa icipan s
we e asked he ollowing ques ion: “Below you ind a lis o beha io al pa e ns. How well does each e lec you beha io in he pas
2 weeks?” Response op ions we e 1 =“Applied ne e ”; 2 =“Applied a ely”; 3 =“Applied some imes”; 4 =“Applied o en”; 5 =“Applied
always”.
STERNBERG ET AL.
-
1105
FIGURE A4 Ci ic esponsibili y: sc eeplo ac o analysis. This igu e shows a sc eeplo o an explo a o y ac o analysis o he 3
su ey i ems aimed a cap u ing ci ic esponsibili y (see Tables A1, A6 and A21 o he i ems and p ocedu e), wi h Eigen alues o ac o s
on he y‐axis and he iden i ied ac o s on he x‐axis. The ac o analysis was pe o med using i e a ed p incipal axis ac o ing, no o a ion,
and a polycho ic co ela ion ma ix.
FIGURE A3 Dis ibu ion o COVID‐19
h ea pe cep ion ac o sco es. This igu e
shows a scaled densi y and his og am o he
ac o sco es cap u ing COVID‐19 h ea
pe cep ion, as in Table A7.
1106
-
STERNBERG ET AL.

FIGURE A5 T ee o he s ai case ime
ask (ex ac ed om Falk e al. (2022), p. 66).
This igu e shows he in e empo al choice
sequence o he s ai case me hod in de ail, as
de eloped by Falk e al. (2022), and employed
in his pape as pa o he p e e ence
measu e o pa ience. The numbe s indica e
he paymen in 12 mon hs, “A” deno es he
esponden 's choice o “100 eu os oday,”
while “B” deno es he esponden 's choice o
“x eu os in 12 mon hs”. All esponden s s a
wi h he le ‐mos decision, he emaining
in e empo al choice sequence is e ealed by
going om le o igh h ough he ee.
STERNBERG ET AL.
-
1107
FIGURE A6 Compliance: sc eeplo explo a o y un o a ed ac o analysis. This igu e shows a sc eeplo o an explo a o y ac o
analysis o he 6 su ey i ems aimed a cap u ing compliance beha io (see Tables A1, A3 and A21 o he i ems), wi h Eigen alues o
ac o s on he y‐axis and he iden i ied ac o s on he x‐axis. The ac o analysis was pe o med using i e a ed p incipal axis ac o ing, no
o a ion, and a polycho ic co ela ion ma ix.
1108
-
STERNBERG ET AL.
FIGURE A7 Compliance: summa ed a ing scale mono one homogenei y assump ion. This igu e illus a es an assessmen o he
mono one homogenei y assump ion o he wo summa ed a ing scales o compliance in he public ( i s ow o g aphs) and p i a e
(second ow o g aphs) domain, as in Table A11. Plo ed a e es sco es on he x‐axis (i.e., he public/p i a e scale sco e i d opping one o
he i ems) and he alue o he espec i e d opped i em on he y‐axis. The mono one homogenei y assump ion he eby equi es expec ed
esponses o each o he h ee i ems o compliance in he public (p i a e) domain o be inc easing/dec easing as he ue dimension o
compliance in he public (p i a e) domain also inc eases/dec eases (i.e., he es sco es). This inc ease/dec ease has o be o a mono one
na u e, which he e seems o be he case (all lines a e mono onously inc easing). The ed line shows a linea i , he blue line a smoo hed i .
[Colou igu e can be iewed a wileyonlinelib a y.com]
STERNBERG ET AL.
-
1109
FIGURE A8 Social desi abili y: sc eeplo ac o analysis. This igu e shows a sc eeplo o an explo a o y ac o analysis o he 3 su ey
i ems aimed a cap u ing social desi abili y bias (exagge a ing posi i e cha ac e is ics) (see Tables A1, A6 and A21 o he i ems and
p ocedu e), wi h Eigen alues o ac o s on he y‐axis and he iden i ied ac o s on he x‐axis. The ac o analysis was pe o med using
i e a ed p incipal axis ac o ing, no o a ion, and a polycho ic co ela ion ma ix.
FIGURE A9 Al e na i e compliance measu e: sc eeplo ac o analysis. This igu e shows a sc eeplo o an explo a o y ac o analysis
o 3 al e na i e su ey i ems aimed a cap u ing compliance beha io (see Tables A1 and A21 o he i ems), wi h Eigen alues o
ac o s on he y‐axis and he iden i ied ac o s on he x‐axis. The ac o analysis was pe o med using i e a ed p incipal axis ac o ing,
no o a ion, and a polycho ic co ela ion ma ix.
1110
-
STERNBERG ET AL.
highe due o he highe densi y and obse abili y o he beha io o o he s, his line o a gumen may be e en mo e
ele an o he p i a e domain gi en he con as be ween c owded li ing si ua ions in he ci y and he much mo e
spaced, u al housing. On he o he hand, one could a gue ha esiden s in smalle owns know each o he (and
especially hei neighbo s) subs an ially be e and a e less anonymous han hose li ing in a la ge ci y. In ha case,
ou expec a ions would sugges he opposi e, ha is, (p i a e) compliance being highe in he u al con ex .
Weexamine hepo en ial impo ance o esponden s' esidencein u al e sus u ban se ings as an addi ionalp edic o
o compliance in he public and p i a e domain by using a su ey‐based measu e o he size o he communi y a
esponden li es in. Speci ically, as pa o ou su ey, we asked esponden s how many inhabi an s li e in hei com-
muni y/municipali y, wi h esponse op ions being “below 5000,” “5001–20,000,” “20,001–100,000,” “100,001–500,000,”
“abo e 500,000.” Based on his ques ion, we cons uc ed an indica o a iable ha akes on he alue o 1 i a esponden
epo ed li ing in a municipali y wi h less han o equal o 20,000 inhabi an s and ha akes on he alue o 0 i (s)he
epo ed li ing in a municipali y wi h mo e han 20,000 inhabi an s. We chose his de ini ion o he a iable because,
in Ge many, municipali ies wi h less han 20,000 inhabi an s a e coun ed as “Kleins äd e,” ha is, small owns/ illages.
Table A22 (Columns 3 and 6) p esen s he esul s o including his measu e o u al/u ban esidence se ing as an
addi ional p edic o in he ull model o compliance in he public and p i a e domain, espec i ely. We ind a nega i e,
bu no s a is ically signi ican associa ion be ween a esponden li ing in a u al as opposed o a mo e u ban se ing and
hei deg ee o compliance in he public/p i a e domain. Howe e , in e ms o magni ude, he e ec size o he co-
e icien is much la ge o compliance in he p i a e han in he public domain ( hough his di e ence is also no
s a is ically signi ican acco ding o a Wald es es ing o he equali y o coe icien s). Thus, he di ec ion and pa e n o
his inding is gene ally in line wi h he i s heo e ical expec a ion ou lined abo e, e en hough hey lack s a is ical
signi icance. Finally and impo an ly, he main esul s in e ms o ou p e iously examined co e p edic o s (economic
p e e ences, ins i u ional us , h ea pe cep ion) emain s able when including esponden s' u ban/ u al esidence
se ing as an addi ional p edic o .
Co ela ion be ween egional CV‐19 h ea pe cep ion and case incidence a e
We examine he ela ion be ween ou measu e o esponden s' COVID‐19 h ea pe cep ion and he incidence a e o
COVID‐19 cases, using da a om he Fede al S a is ical O ice o Ge many (“S a is isches Bundesam ”) abou he
numbe o egis e ed COVID‐19 cases in he pas 7 days pe 100 people.
17
To do so, we employ ede al s a e le el
a e ages o ou h ea pe cep ion measu e and he s a e le el a e aged COVID‐19 case incidence (i) ac oss he ime
pe iod o he da a collec ion (Feb ua y 2021) and (ii) ac oss all p e ious in ec ion wa es (i.e., om he beginning o he
pandemic un il he s a o he da a collec ion). We chose o examine he case incidence le el du ing bo h o hese ime
pe iods because i seems likely ha , i indi iduals' h ea pe cep ions a e indeed ela ed o he ac ual numbe o cases,
bo h, he ajec o y and in ensi y o p e ious in ec ion wa es as well as he se e i y o he in ec ion wa e in he mids o
ou da a collec ion, play a ole (in po en ially opposing ways). In addi ion o epo ing Pea son co ela ion coe icien s,
we also show he esul s o eg essing s a e le el h ea pe cep ion on s a e le el case incidence a es while con olling
o possible con ounde s a he s a e le el (popula ion densi y, sha e o he Ge man popula ion li ing in his s a e, he
s a e being a S ad s aa , i.e., one o he h ee Ge man ci ies ha a e also a ede al s a e).
The esul s o his exe cise a e shown in Table A23 in Appendix A. We ind ha nei he he a e age COVID‐19 case
incidence du ing he da a collec ion no he case incidence du ing he p io in ec ion wa es a e signi ican ly ela ed o
esponden s' COVID‐19 h ea pe cep ion a he s a e le el. Howe e , he indings do e eal a pa e n wi h espec o he
di ec ion o he co ela ions: While he e is a small, posi i e co ela ion be ween esponden s' h ea pe cep ion and he
case incidence a e du ing ou da a collec ion pe iod, i s co ela ion wi h he case incidence a e du ing p io in ec ion
wa es is sligh ly nega i e.
We, i s , in e p e he o me esul as p o iding some o m o alida ion o ou h ea pe cep ion measu e. Second,
in combina ion, hese indings in e es ingly also sugges ha indi iduals' h ea pe cep ion may be, on he one hand,
highly dependen on he immedia e pandemic si ua ion, and, on he o he hand, pa icula ly sensi i e o e ospec i e
adjus men s. Speci ically, we obse e ha esponden s who expe ienced highe case numbe s in he i s wa es o he
pandemic ac ually epo ed a sligh ly lowe ea o COVID‐19 du ing ou su ey da a collec ion, while hose ha
expe ienced highe case incidence le els du ing he da a collec ion also e ealed a sligh ly highe pe cei ed h ea . To
ha end, hese indings mo eo e con i m he concep ualiza ion o ou measu e o h ea pe cep ion as a no necessa ily
a ionally o objec i ely in o med concep , bu a he as a subjec i e accoun o how h ea ening esponden s pe cei ed
he COVID‐19 pandemic o be in gene al and how h ea ening hey pe cei ed i o be wi h ega d o speci ic aspec s o
hei li es (heal h, inancial, social; see also he ac o sco es we used o cons uc he h ea pe cep ion measu e). This
STERNBERG ET AL.
-
1117

concep ualiza ion o h ea pe cep ion may also con ibu e o explaining why i appea s o play such a c ucial ole o
indi iduals' compliance beha io s du ing he pandemic, as iden i ied in he main analysis.
Finally, we emain, o he ollowing easons, cau ious no o o e s a e o o e ‐in e p e he ou lined co ela ions
be ween ou measu e o h ea pe cep ion and COVID‐19 case incidence a es—nei he in e ms o he absence o
s a is ical signi icance no in e ms o he me e di ec ions o he co ela ions. Fi s , he analysis was conduc ed a he
s a e‐le el and is, hus, based on N=16. I is he e o e s a is ically unde powe ed. Second, du ing he pandemic he e
was a lo o a ia ion wi hin ede al s a es hemsel es and i is likely ha li ing in a la ge ci y wi hin a ce ain s a e as
opposed o li ing in a u al a ea wi hin a ce ain s a e plays a c ucial ole o he co ela ion be ween case numbe s and
h ea pe cep ion. A mo e ho ough analysis would he e o e be o go e en lowe han s a e‐le el. Un o una ely, we did
no ask esponden s abou he speci ic dis ic hey li e in and, hus, could no conduc such an analysis. Thi d, and
ela edly, pa e ns a indi idual le el and he agg ega e le el can di e .
Addi ional analyses o he ins i u ional us a iables
To be e unde s and he di e en ial obse ed dynamics o us in he RKI and us in he go e nmen ac oss he wo
domains, we conduc ed a numbe o supplemen a y analyses ha exploi addi ional in o ma ion collec ed in ou su ey.
In e ms o us in he RKI, we had, in ou heo y sec ion, a gued ha we would, i any, expec a la ge impac on
compliance in he public domain gi en he mo e echnical s ipula ions in his ealm, o example, wea ing a mask o
keeping a 1.50 m dis ance om ano he as opposed o he gene al ecommenda ion o “s aying a home” in he p i a e
domain. Recall ha we had chosen us in he RKI as a measu e o us in scien i ic ins i u ions in he Ge man case,
because in Ge many he RKI was he dominan scien i ic ins i u ion ha was hea ily in ol ed in policy consul a ion
and communica ion wi h ega ds o he COVID‐19 pandemic. In line wi h ou heo e ical expec a ions, ou co e esul s
sugges ha us in he RKI, indeed, plays a s onge ole o public compliance han o p i a e compliance, hough i s
es ima ed e ec s a e s a is ically signi ican in bo h ealms. Thus, o assess whe he he abo e explana ion is plausible,
we d aw on wo al e na i e p oxies o us in scien i ic ins i u ions, namely (i) us in science (in gene al) and (ii)
us in he WHO and examine whe he he obse ed pa e n holds. Table A28 in Appendix A epo s he esul s o his
exe cise. Es ima ing he SEM wi h us in science and us in he WHO, espec i ely, ins ead o us in he RKI as a
p edic o o public and p i a e compliance (all o he a iables emaining as in he main speci ica ion) e eals he same
pa e n, which we also obse ed o us in he RKI: Bo h p oxies e eal s a is ically signi ican co ela ions wi h
compliance in he wo domains, bu much la ge es ima ed e ec s in he public domain. The magni udes and le els o
s a is ical signi icance a e s ikingly simila o hose o us in he RKI ( us in science: β
pub
=0.182, p‐ alue <0.001;
β
p i
=0.053, p‐ alue <0.05; us in he WHO: β
pub
=0.165, p‐ alue <0.001; β
p i
=0.055, p‐ alue <0.05). Mo eo e ,
including us in science and us in he WHO did no change he pa e n iden i ied o us in he go e nmen (o any
o he o he co e p edic o s) subs an ially.
In e ms o us in he na ional go e nmen , we had, in ou heo y sec ion hypo hesized con lic ing logics: On he
one hand, he absence o o mal moni o ing by s a e au ho i ies in he p i a e ealm migh make us as an in insic
mo i a o mo e impo an . On he o he hand, he p i a e ealm may be unde s ood as a ealm in which he go -
e nmen has inhe en ly no legi ima e ole o play. Ou esul s e ealed ha us in he go e nmen is only signi ican ly
co ela ed wi h compliance in he p i a e domain, no wi h compliance in he public domain. To examine his inding
u he and unde s and how i ela es o ou ini ial expec a ions, we u ilize wo addi ional ins i u ional us a iables,
which aim a quan i ying us in ins i u ions o ac o s ha a e ela ed o he Ge man go e nmen , bu in di e en
ways: Fi s , we epea he main SEM speci ica ion, bu eplace us in he go e nmen wi h us in he s a e‐le el
go e nmen . We use his a iable since hey allow us o in es iga e whe he he obse ed pa e n is a gene al
pa e n o go e nmen ‐ ela ed ins i u ions o whe he i is a speci ic pa e n o ci izens' us in he na ional go -
e nmen in pa icula . The esul s o his exe cise a e epo ed in Table A29 (Columns 2, 3, 7 and 8) in Appendix A. Fo
us in he s a e‐le el go e nmen , we ini ially ind a simila pa e n as o us in he (na ional) go e nmen , ha is,
he e is only a signi ican co ela ion wi h compliance in he p i a e, bu no in he public domain. Howe e , he
es ima ed e ec s a e smalle and also sligh ly less s a is ically meaning ul (β
pub
=0.031, p‐ alue =0.267; β
p i
=0.070, p‐
alue <0.01). Mo eo e , when adding bo h, us in he s a e‐le el go e nmen and us in he na ional go e nmen ,
simul aneously, he s a is ical signi icance o he es ima ed e ec s o us in he s a e‐le el go e nmen disappea s,
while hose o us in he na ional go e nmen emains (see Columns 3 and 8). These indings sugges ha he pa e n
obse ed in ou main analysis may be speci ic o and d i en by us in he na ional go e nmen (“Bundes egie ung”).
Gi en hese indings, we p opose an al e na i e explana ion o he obse ed dynamics, which we aim o in es iga e
by epea ing he abo e exe cise bu ins ead including us in Ge many's es ablished media channels as a p edic o in
1118
-
STERNBERG ET AL.
he model. Speci ically, we a gue ha he high ele ance o us in he na ional go e nmen o ci izens' compliance in
he p i a e ealm could be a esul o he speci ic policy communica ion and messaging campaigns issued du ing he
pandemic in Ge many. Fo ins ance, he na a i es o he egula public add esses by Chancello Angela Me kel o he
Ge man popula ion we e much mo e ocused on he gene al aspec o s aying a home (i.e., along he lines o being
uni ed in isola ion), which is closely ela ed o ou de ini ion o p i a e compliance, and less so on he mo e echnical
ules o mask‐wea ing o a oiding handshakes (i.e., ou de ini ion o public compliance) (see e.g., Angela Me kel,
ele ised speech, 2020). Thus, i may be ha , as a esul o such communica ion s a egies h oughou he en i e
pandemic, he iew o he p i a e ealm as a ealm in which he go e nmen has no p ominen ole o play was
somewha loosened, especially o indi iduals wi h a high le el o us in he na ional go e nmen . We y o assess he
plausibili y o his al e na i e in e p e a ion by d awing on ano he a iable, namely us in es ablished media
channels, which a e widely b oadcas ed and suppo ed he men ioned public add esses by Angela Me kel. Thus, i he
obse ed pa e ns a e a esul o policy communica ion du ing he pandemic, we should obse e a e y simila pa e n
o ci izens' us in he said go e nmen ‐ ela ed media channels—which is wha we indeed obse e: The es ima ed
coe icien s o ci izens' us in es ablished media channels show he same pa e n as hose o us in he na ional
go e nmen , ha is, hey a e s a is ically signi ican o compliance he p i a e ealm, bu no o compliance in he
public ealm (β
pub
= −0.017, p‐ alue =0.499; β
p i
=0.081, p‐ alue <0.001). Mo eo e , while he coe icien s a e smalle
in magni ude, hey a e, unlike he coe icien s o us in he s a e‐le el go e nmen , obus o a simul aneous es i-
ma ion oge he wi h us in he na ional go e nmen (see Columns 4, 5, 9 and 10 o Table A29). We he e o e in e p e
hese indings as suppo o he al e na i e explana ion ou lined abo e. Finally, i is no ewo hy ha including he
men ioned addi ional a iables did no change he pa e ns iden i ied o us in he RKI (o any o he o he co e
p edic o s) subs an ially.
STERNBERG ET AL.
-
1119