Online Dating Is Shifting Educational Inequalities in Marriage Formation in Germany

Abstract

Digital technologies govern a large part of our social lives, including the pursuit of a romantic partner. Despite recent inquiries into the social consequences of meeting online, what remains unclear is how the link between education and union formation varies in online versus offline meeting contexts, particularly on the backdrop of growing educational gaps in marriage. Using 2008–2019 pairfam data from Germany (N = 3,561), this study ran a series of Fine-Gray competing risks models to assess how online dating shapes the transition to marriage for partnered adults with nontertiary and tertiary education. Results reveal that irrespective of education, men in online-formed couples had greater chances of marrying than men in couples established offline. Highly educated women who met their partner in nondigital ways were less prone to marry than lower-educated women; for women in couples initiated online, however, the pattern was reversed. The internet dating marriage advantage of well-educated women was partly related to better matching on marriage attitudes and gender ideology. Facing a scarcity of eligible partners offline, high-educated women draw on more abundant online options to select more egalitarian-minded men. This study overall suggests that internet dating fosters an uneven distribution of opportunities for marriage, highlighting the role of digital partner markets in the social demography of union formation.

Full text

Demography (2021) 58(5):1977–2007
DOI 10.1215/00703370-9420350 © 2021 The Author
This is an open access arti cle dis trib uted under the terms of a Creative Commons license (CC BY-NC-ND 4.0).
ELECTRONIC SUPPLEMENTARY MATERIAL The online ver sion of this arti cle (https: / /doi .org /10 .1215 /00703370
-9420350) con tains sup ple men tary mate rial.
Published online: 18 August 2021
Online Dating Is Shifting Educational Inequalities
in Marriage Formation in Germany
Gina Potarca
ABSTRACT Digital tech nol o gies gov ern a large part of our social lives, includ ing the
pur suit of a roman tic part ner. Despite recent inqui ries into the social con se quences of
meet ing online, what remains unclear is how the link between edu ca tion and union
formationvariesinonlineversusofinemeetingcontexts,particularlyontheback
drop of grow ing edu ca tional gaps in mar riage. Using 2008–2019 pairfam data from
Germany(N =3,561),thisstudyranaseriesofFineGraycompetingrisksmodelsto
assess how online dat ing shapes the tran si tion to mar riage for partnered adults with
nontertiaryandtertiaryeducation.Resultsrevealthatirrespectiveofeducation,menin
onlineformedcoupleshadgreaterchancesofmarryingthanmenincouplesestablished
ofine.Highly educatedwomen whomet theirpartnerinnondigitalwayswereless
pronetomarrythanlowereducatedwomen;forwomenin couplesinitiatedonline,
however,thepattern wasreversed.Theinternetdatingmarriage advantageofwell
educatedwomenwaspartlyrelatedtobettermatchingonmarriageattitudesandgen
derideology.Facingascarcityofeligiblepartnersofine,higheducatedwomendraw
onmoreabundantonlineoptionstoselectmoreegalitarianmindedmen.Thisstudy
over all sug gests that internet dat ing fos ters an uneven dis tri bu tion of oppor tu ni ties for
marriage,highlightingtheroleofdigitalpartnermarketsinthesocialdemographyof
union for ma tion.
KEYWORDS Marriage • Internetdating • Technology • Education • Gender
Introduction
Online dat ing through websites, phone apps, chat rooms, or social net works has
introduced new ways for individuals to meet and interact with potential partners
(Finkel et al. 2012). Research has already started noting some of the sociodemo
graphictransformations triggeredbyinternet dating.Forheterosexual couples,for
example,meetingonlineislinkedtomorecouplediversityintheUnitedStatesand
Germany(Potarca2017;Thomas2020) and to faster tran si tions to mar riage in the
UnitedStates(Rosenfeld2017).Still,howtheinternetasaprevalentsourceofmeet
ingpartners(Rosenfeldetal.2019) has affected mar i tal union for ma tion for dif fer ent
socialgroupsremainsunclear.Groupsaredenedherebyeducation,oneofthemost
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1978 G. Potarca
importantcommoditiesonthemarriagemarket(Becker1993;Oppenheimer1988)
and,althoughbynomeanstheironlyindicator,areliablemarkerofeconomicand
culturalresources(Blossfeld2009).
InmostWesterncountries,currenteducationaldifferencesinunionformationdes
cribeapositivegradient,withmarriagemoreoftenassociatedwiththewelleducated
(McClendon2018;VanBaveletal.2018).ForGermany(thecountryoffocusinthis
study),pastresearchhasmostlyfoundtheopposite:individualsatthebottomofthe
educationaldistributionhavehadgreater chancesofmarryingthanthoseatthetop
(Baizánetal. 2003; Mulderetal.2006). Information based on more recent data is
neverthelesslacking.Regardlessofcurrentpatterns,severalpredictionscanbemade
about the internet’s role in shap ing union for ma tion across edu ca tional groups. On
the one hand, at a time when urban spaces and work set tings are becom ing more
socioeconomically segregated (Marcińczak et al. 2016; McClendon et al. 2014),
onlinemeetingpoolsprovidegreateropportunitiesforencounters,potentiallygener
at ing more chances for mar riage among peo ple of all edu ca tional lev els. On the other
hand,bettereducated adults, who are more skilled at navigating new technologies
(OllierMalaterreetal.2019),maybeengagingmoreeffectivelywithonlineresources
tomeetandndwellsuitedpartners.Facingashortageofeligiblecandidatesofine
(EckhardandStauder2019),universityeducatedwomenmay particularlyseize the
opportunities granted in a larger and less restrictive market to select more value
compatiblepartners(Finkeletal.2012)andthusformmoremarriageinducingunions
(Houtsetal.1996).
To test these sce nar ios and assess whether the edu ca tional gap in mar riage varies
acrossmeetingcontexts,thisstudyfocusesonhowpeoplewithlower(i.e.,nonter
tiary)andhigher(i.e.,tertiary)educationexperiencethetransitionintomarriage,as
opposed to cohabiting or break ing up, among partnered men and women who met
theirmatchonlineversuselsewhere.BasedontheGermanFamilyPanel(pairfam)
data, this research targets adults aged 18–48, a subpopulation most likely to be
using the internet as a romantic marketplace. By providing a wealth of couple
level infor ma tion, pairfam also allows for direct empir i cal tests of the o ret i cal
mechanisms,particularlytheexaminationofhowpartnermatchingonfamilyval
ues(Press2004), such asmarriageattitudes and genderideology,affectsmarital
chancesforcoupleswhometonlinecomparedwiththosewhometthroughconven
tionalchannels.Marriagefavorableattitudesandexpectationsatboththeindivid
ual(SasslerandSchoen1999)andthecouplelevel(WallerandMcLanahan2005)
represent acentral correlateof eventualtransitionstomarriage. Genderideology
embodies a set of beliefs regard ing men’s and women’s involve ment in sep a rate
spheresofactivity (Davis and Greenstein 2009).Traditionalideology imposes a
gendereddivisionoflabor(e.g.,menasbreadwinnersandwomenascaregivers),
whereasegalitarianprescriptspromotejointresponsibilitiesforbothgenders.How
(dis)similarromanticpartnersareingenderideologymattersnotonlyfortheactual
divisionofunpaidwork(NitscheandGrunow2016, 2018) but also for rela tion ship
progression;recentstudiessimilarlyusingpairfam data have high lighted the effect
ofgenderideologyonpartnershipstability(Hudde2020)andthetransitiontopar
enthood(HuddeandEngelhardt2020). I con tend that if internet dat ing encour ages
morecompatibilityinfamilyvalues(Finkeletal.2012),especiallyforthewelloff,
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1979Online Dating and Educational Inequalities in Marriage Formation
then indi vid u als who met their part ner online and who belong to a cer tain social
backgroundwillmarryfasterthanothers.
Germany is a compelling context in which to study the consequences of digi
tal dating on educational inequalities in marriage. Germany is not only a context
withwidelyadoptedonlinedatingpractices,basedontoolssuchasdatingplatforms
(Schulzetal.2008)or(morerecently)phone apps (Bitkom Research 2017; Suhr
2020), but it is also rep re sen ta tive of Western countries stuck in a male bread win ner
model(Bellani,EspingAndersen,andPessin2017;Blossfeld2009). The internet and
itspromiseofchallenginggenderedconceptsofcourtship(Hardey2002)mayhave
openeduppartneringpossibilitiesbyallowinghigheducatedwomentoselectpart
nerswillingtotakeamorenontraditionalroleinmarriage.Furthermore,despitehav
inganestablishedsystemofstratiededucation,knownforefcientlycoordinating
withthelabormarket(Dieckhoff2008),Germanyhasalsodisplayedarecentupturn
intheunemployment(Klein2015)andpovertygaps(Spannagel2016) between the
lowerandhighereducated.Inlightofthesegrowinginequalitiesandtheresulting
declineintheeconomicmatevalueofthelowereducated(ZagelandBreen2018),
thisstudyseekstoidentifywhetherthepreviouslyobservednegativemarriagegap
still holds and which German adults are more likely to marry in the digital age.
AlthoughtheGermancontextalsorequiresconsideringcohabitationasaviablelong
termunionform(Hiekeletal.2015),thisanalysisfocusesmainlyonmarriage,given
itsgreaterculturalandinstitutionaladvantages(LückandRuckdeschel2018).
Thecurrentresearchmakesseveralimportantcontributions.First,itisoneofthe
few stud ies seek ing to under stand the demo graphic con se quences of online dat ing,
anditisthersttoexamineeducationaldifferencesintheinternet’seffectonmar
riageformation.Withdigitaldatingbecomingoneofthemainwaystondapartner
(Rosenfeld and Thomas 2012; Rosenfeld et al. 2019), an increase in the partner
ing(dis)advantageofcertaingroupsamongonlineformedcouplescouldpotentially
redraw cur rent inequalities in mar riage.
Second,thisstudyusesrich,multiactorlongitudinaldatatomodelthecomplexity
ofunionformationandtotestfordyadic(i.e.,respondentpartner)matchingonfam
ilyvalues.Althoughfullyovercomingthelimitationsofpreviousworkfocusingon
associationsisunlikely,giventhatthedataarestillbasedonanonrandomsampleof
observations,thecurrentstudyaccountsformultiplesourcesofselectionbiasbycon
trollingfortheobservedheterogeneityofpeopledatingonlineandbyalsoaddressing
prepartneringpatterns.
Finally,thisresearchoffersanunprecedentedempiricalwindowintohowdigital
modesofinteractionchangemarriageformation.Macrolevelstudiesexaminingthe
dif fu sion of sta ble broad band con nec tions within house holds have con cluded that
theexpansionoftheinternetincreasedthenumberofpeoplegettingmarried(Bellou
2014).Thistypeofresearch,though,cannotdeterminewhethernewtechnologiesare
actualagentsofchangeoraresimplyreectingpreexistingshiftsinunionformation.
AsCesareetal.(2018) argued, under stand ing whether dig i tal tools for dat ing have
genuinelyalteredmaritalpatternsrequiresadirectcomparisonbetweenadultswho
usedonlinedatingandthosewhodidnot.Thisstudyhencereliesonamicrolevel,
eventcenteredapproachtoidentifythestratifyingconsequencesofonlinedatingon
mar i tal union for ma tion.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1980 G. Potarca
The Educational Marriage Gap
IntheUnitedStates,lesseducatedindividualshavelowerchancesofgettingmarried
aswell asstayingmarried(McClendon2018;ParkerandStepler2017).Apositive
educationalgradientinmarriage for bothmenand(morerecently) women isalso
evidentinEurope(Jalovaara2012;VanBaveletal.2018),althoughitsextentvaries
fromcountrytocountry,dependingonprevailinggenderrolesandlevelsofeconomic
inequality(Kalmijn2013).InGermany,Kalmijn(2013) found a pos i tive gra di ent
fortheprobabilityofbeinginaunionamong40to49yearoldmenbornbetween
1953and 1971.AmongWestGermanwomenborn duringapproximatelythesame
period, how ever, a high level of school ing was linked to a lower risk of transitioning
toarstunion(cohabitationordirectmarriage)oroftransitioningfromcohabitation
tomarriage(Baizánetal.2003).Mulderandcolleagues(2006) also showed that in
contrasttotheUnitedStates,highereducationisassociatedwithalowerlikelihood
ofrstunionformationforyoungwomeninGermany.Whereassomestudieshave
assertedthatthispatternmerelyreectsthedelayingeffectofeducationalexpansion
(BlossfeldandJaenichen 1992), oth ers have pos tu lated an over all neg a tive human
capitaleffectonmarriagepropensity,bothduring andafterschooling(Brüderl and
Diekmann 1994).
Aspreviously noted, welackdirectevidence regarding morerecentpatterns of
partneringormarriageacrosseducationalgroupsinGermany.Onecanexpectthatris
inglevelsofeconomicinequality(LippsandOesch2018) over the last two decades,
particularlyamongmen(ZagelandBreen2018),createdconditionsforapositivegra
dientinunionformation.Givenincreasedreturnstoeducation(Psacharopoulosand
Patrinos2018),onecouldalsoexpectthatforthegrowingnumbersofyoungwomen
investingintheireducation,thepreviouslyconventionaloptionofchoosingmarriage
overfulltimeemployment(Drobnicetal.1999)isnolongeroptimal.Researchhas
conrmedthatcomparedwiththelesseducated,bettereducatedGermanwomenare
moreoftenlinkedtolifelongsinglehood(Bellani,EspingAndersen,andNedoluzhko
2017),childlessness,andnonfamilylivingarrangements(Sobotka2011). Inacon
textinwhichthemalebreadwinnermodelisstillthriving(Bellani,EspingAndersen,
andPessin2017),itmayseemthateconomicallyindependentwomenaregradually
withdrawingfromtraditionalmarriage.Nevertheless,itisunclearwhetherthispattern
reectsadecreasingcentralityofmarriage(Oppenheimer1994) or scarce meet ing and
matingopportunities(EckhardandStauder2019). If the lat ter is the case, then dig i tal
datingmodalitiesandthelargesetofoptionstheyprovidecouldsignicantlyboost
marriageopportunitiesformarriageorienteduniversityeducatedGermanwomen.
Social Structure and Foci of Interaction
The the ory of social struc ture(Blau1978;Blauetal.1984) states that inter per sonal
choices,includingthoserelatedtopartners,arelargelyaffectedbythemacrosocial
structureoftheenvironment.Opportunitiesforcontactdeterminenotonlytheproba
bilityofndingamaritalpartnerbutalsothedegreeofsorting—thatis,howsimilar
partnersareonvarioussocialaspects(BlauandSchwartz1997). The con cept of foci
of activ ity (Feld 1984; Marsden1990)narrowsstructural inuence to the specic
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1981Online Dating and Educational Inequalities in Marriage Formation
contextswhereindividualsroutinelyinteract,suchastheworkplace,neighborhood,
orfamily.Theirdifferentsizeandlevelofsociodemographicsegregationmeanthat
eachsettingprovidesadistinctsetofpossibilitiesforinteractionandaspecicpool
ofpotentialpartners. Several studies have conrmed the relevanceofwherecou
plesmeet(KalmijnandFlap2001;McClendon2018), includ ing the role of meet ing
online(Potarca2017;Rosenfeld2017;Thomas2020),forthetimingofmarriagefor
ma tion and the mech a nism of sorting.
Despitetheparticularitiesofdifferentonlinedatingcontexts(e.g.,seesectionA
oftheonlineappendixforadiscussion),thisstudyconsidersthemasasinglepart
nershipmarketbyvirtueofthemsharingstructuralfeaturesthatuniquelydistinguish
themfromallofinepartnermarkets.Inadditiontoreducinguncertaintybymaking
partneringintentionsexplicitandunequivocal(Schmitz2017), internet dat ing grants
accesstoanunprecedentedlydiversepoolofcandidatesandprovidesuserswithvar
iousscreeningtoolstonetunetheirchoices. Inlightofthesefeatures,onecould
arguethatinternetdatingimprovesmaritalchancesacrossallgroups.Online,indi
vidualsofbothlowerandhighereducationndavarietyandabundanceofchoices
thatwouldnototherwisebeaccessible.Withoutexploringpatternsacrosspeoplewith
differentsocialbackgrounds,previousworkindeedfoundthatU.S.heterosexualcou
ples who met online experienced faster transitions to marriage (Rosenfeld 2017).
Certaingroups,however,mightbenetfromcertainsocialcontexts,includingonline
dat ing spaces, more than oth ers.
Cultural Capital and Digital Skills
InBourdieu’s(1989, 1998, 2013)conceptualizationofsocialspaceandsymbolic
goods,socialstructureisviewedasbothexternalandinternalized in the form of
preferences. Bourdieu (2008), cited by Schmitz (2017), also showed that a wider
structureofopportunitydoesnotnecessarilypromptanallencompassingincrease
inmaritalprospects:beforetheinternetage,anenlargementofthemarriagemarket
in1960sruralFrance(asaresultofmodernizedinfrastructureandeconomy)ledto
moresegregationratherthantoanopeningofmaritalchoices.Thetheoryofsocial
space (Bourdieu 1989, 1998, 2013) states that social agents position and classify
themselvesinrelationtootheragentsbylevelofculturalcapital.Socializationpro
cessesandeducationensurethatindividualsexhibitclassspecictastes,dispositions,
andhabits(i.e.,embodiedcapital),ownsymbolicpossessions(i.e.,objectiedcapi
tal),andhavespecicqualicationsandskills(i.e.,institutionalizedcapital).Social
groupswith greaterculturalcapitalconvertsuchassetsinto moreorothertypesof
capitaltopreservedominanceandexcludeothersfromgaininghighstatuspositions,
hencereinforcingsocialinequality.ApplyingaBourdieusianviewondigitalmating
inGermany,andshowingthatsocialclassreproductionstronglypermeatespeople’s
onlinepreferencesandinteractions,Schmitz(2017)alreadynotedthatinternetdating
benetswellpositionedgroupsmorethanothers.
Generallymoreendowedwithculturalcapitalandspecicallymoreequippedin
knowinghowtoengagewithtechnology(OllierMalaterreetal.2019),highlyedu
catedadultsarelikelymoreskilledatnavigatingandultimatelyseizingwhatonline
partnermarkets have to offer.Comparedwiththose who have less education,the
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1982 G. Potarca
welleducatedarebetteratselfpresentation,canlocate digitalresourcesmoreef
ciently,andareabletomoreeasilysiftthroughunwantedoptions(OllierMalaterre
et al. 2019).Intheonlinequestforapartner,digitalskillsandstrategiesofefciency
(DrögeandVoirol2011)mayresultinbettermatchedunions,withagreaterlikeli
hood of advanc ing to mar riage.
Thegreaterinternetrelatedaccelerationinmarriageformationforindividualswith
tertiaryeducationrelativetoothersmightparticularlyapplytowomen.Facinglim
itedmaritalprospectsinofinespaces,higheducatedwomenmaymorepurposely
engagewithandbenetfrominternetdatingthanhigheducatedmen.Onlinedating
poolsoftenincludeasurplusofmen(Felicianoetal.2009;Skopeketal.2011),the
oretically providing ademographicadvantage to women acrosstheboard.Never
theless,by meansofgreatereconomicresources,highlytrainedwomenhavemore
bargainingpowerandcouldthusmoreeasilyupholdtheirpreferencesthanthelower
educated(Meeussenetal.2019).Becauseindividualswithhighermatestandardsare
moredeterminedtondacompatiblepartner(Sprecheretal.2019),higheducated
womenmightthusmakethebestuseofdigitaldatingpossibilitiesformorerened,
marriagepromotingpartnershipchoices.
Online Matching on Marital and Gender Values
Universityeducatedwomenmayalsoaccessonlinepartner markets to seek com
patibility along noneconomiclines,suchasshared values orinterests.Inlight of
enhancedparticipationinthelabormarketandincreasesinnancialindependence,
highlyeducatedwomenarefocusing less on the economic prospects of a partner
andmoreonnonmaterialtraits(Press2004). The recent increase in the prev a lence
andstability of hypogamous unions (Grow et al. 2017; Schwartz and Han2014),
inwhichthefemalepartner is bettereducatedandoftentheprimary breadwinner
ofthefamily(KlesmentandVanBavel2017;Qian2018;VanBavelandKlesment
2017), is also believed to her ald a shift in the qual i ties that women pre fer in a part ner
(BouchetValatandDutreuilh2015).Less reliant onmen’seconomic resourcesor
highsocialstatus,welleducatedwomenmayinsteadevaluateprospectivematesin
termsofpotentialcontributionstodomesticandfamilywork,personality,physical
attractiveness,orsociability(Press2004;ZentnerandEagly2015).Recentevidence
conrmedthateconomicallywelloffwomenpreferfamilyorientedmenwhopriori
tizetimewiththeirfamilyandtakeanontraditionalroleincaringforchildren(Croft
et al. 2020;Meeussenetal.2019;ThomaeandHouston2016).
The realization of such preferences, however, is often contingent on a coun
try’s advancement toward gender equality (Zentner and Eagly 2015). In contexts
oftransitional genderideology(Press 2004),whereprogressin achievingequality
between men and women has stalled (England et al. 2020), universityeducated
women are often caught between mod ern and tra di tional scripts, encour aged to gain
(some)incomebutstillexpectedtodothelion’sshareofhousework(Gui2020). In
Germany,whereaconservativeculturalandinstitutionalcontextstillsanctionsanon
egalitarian division oflabor,marriage remainsalargelytraditional affair(Bellani,
EspingAndersen, and Pessin 2017;MüllerandDräger2019). Even within highly
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1983Online Dating and Educational Inequalities in Marriage Formation
educatedhomogamouscouples,partnersstillspecialize,partakinginanunequaldivi
sionofpaidandunpaidwork(Buschneret al.2018).Given limitedavailabilityof
eligiblepartners(EckhardandStauder2019),higheducatedwomenlivingintradi
tionalcontextsmaystillneedtodisplaytraditionalfamilyvalues(Blossfeld2009),
whethergenuinelyselfendorsedorresultingfromanticipatingafuturepartnerwho
islesslikelytotakeonacaregivingrole(BargandBeblo2012;Croftetal.2020).
By providing ample matching possibilities and a less constraining dating space, 
where partnering choices could bet ter align with peo ple’s gen u ine pref er ences and
expectations(Geser2007), internetdatingmayallow universityeducatedwomento
bemoreselective.Withmuchgreateropportunitiestomeettheirdemand,highlyedu
catedwomenwhowish tomarryyetequallyshare paidandfamilyworkcould use
internetdatingoptionstond more genderprogressivemen.Welleducatedwomen
actively searching for marital partners online would then hypothetically establish
semitraditional arrange ments in which part ners match on tra di tional views regard ing
marriage(e.g.,seenas an unbreakablebond,aninstitution central tofamilylife)as
wellas progressivegender roleattitudes.Highlytrainedwomenwouldthusreap the
symbolicandstatusenhancingrewardsofmarriage(Cherlin2020)whilealsonegoti
atingamoreegalitariangenderideology(andperhapspractices)withinthehousehold.
Comparatively,inadditiontobeingless(digitally)skilledatidentifyingandconverting
onlineopportunitiesintopartneringsuccess,lowereducatedwomenwhoalsohopefor
ashareddivision oflaborbut arelesseconomicallyindependent(i.e.,lessattractive
marriagepartners)mayfacemoredifcultymeetingtheirdemandandthusendupin
lesswellmatchedunions.
Whatfollows,giventheimportanceofcompatibilityandpartnerssharingpartnership
oriented values for relationship development and the transition to marriage (Chi
et al. 2020;Houts etal.1996;KellyandConley1987),isthatthehighlyeducated
(particularlywomen)whousedigitaltoolsfordatingaremorelikelytomarrythan
thelowereducated.Thispatternwouldalsooccurbeyondtheeffectofsociodemo
graphichomogamy.Pastwork showedthatsortingon socialbackgroundmattered
moreintheearlyphaseof partnering, when individualsselectedsimilar othersto
guaranteeshared interests,whereassimilarityinvalueswasmoreimportanttothe
couple’sprogressiontoamorecommittedstageoftheirunion(KerckhoffandDavis
1962). Therefore, higheducated women in couples that formed online, with high
valuescompatibility,areexpectedtoprogressmuchfastertowardmarriagethanthe
lowereducated, irrespective of status compatibility(e.g.,sharing the same educa
tional level, or the same place of ori gin).
Hypotheses
Tosummarize,Ianticipatethathighlyeducatedadultswhomettheirpartneronline
will more often transition to marriage than the lesseducated, particularly among
women(Hypothesis1).Theonlinedatingmarriageadvantageofthehighlyeducated
willholdbeyondtheeffectof sociodemographic homogamy and willbepartially
explained by matching on traditional marriage values and egalitarian gender role
beliefs(Hypothesis2).
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1984 G. Potarca
Are Online and Oine Dating Spaces Mutually Exclusive?
Theargumentspresentedearlierdonotimplyanyoverlapbetweenconventionaland
onlinedating, butthelines betweenthetwo maynotbethatclearinreallife.For
example,coupleswhomet ofinemaystillusedigitalcommunication(e.g.,social
media, messaging apps, video calls) as a relationship maintenance tool (Bergdall
et al. 2012).Nevertheless,evenifofinedatersunavoidablyendupinteractingdig
itally,whatdistinguishesthemfrom onlinedatersishowtheyselect theirpartners.
Aspreviouslystated,onlinecontextsprovideaccesstonumerouspotentialencoun
tersandpossibilitiesofscreening,whichpeoplepotentiallyusetoselectmorewell
suitedpartners.Searchingforapartneronlinecouldgeneratedifferentmatchesthan
exclusivelysearching ofine,withimplicationsforfurtherprogressionto marriage
likelyunalteredbyanysubsequent useofdigital communicationbetweenalready
established part ners.
Selection Into Online Dating
Drawing causal infer ences about the effect of online dat ing on mar riage for ma tion
wouldideally requireanexperimental designrandomly assigningadultstodifferent
search strat e gies. In the absence of such ran dom assign ment, obser va tional stud ies of
differencesin maritalchancesbetween ofineandonline datersmightbe subjectto
selectionbiases. Inadditiontospecicities insociodemographicprole(e.g.,Hitsch
et al. 2010)—including an overrepresentation of men (Schulz et al. 2008)—people
adoptingonlinematingstrategiesmayhaveinherentlydifferentxedorvariablechar
acteristics,whichmightinturnaffectunionformation. Severalidiosyncratic factors
maydetermineboththechoiceofdatingstrategyandmaritalsuccess.First,thepsy
chologicalproleofonlinedaterscouldplayarole.Althoughinitialaccountsdepicted
individualsusinginternetdatingasstereotypicallyshyandsociallyanxious(McKenna
et al. 2002;Whittyand Carr2006),subsequent studieshavefoundsuch technology
userstoscorelowondatinganxiety(ValkenburgandPeter2007)andhighonsociabil
ity(Kimetal.2009),extraversion,oropenness(TimmermansandDeCaluwé2017).
Otherstudiesfoundnoconnectionbetweenpersonality,selfesteem,anddigitaldating
(Blackhartetal.2014;Oroszetal.2018),butscholarshaveacknowledgedthatthepsy
chologicalproleofonlinepartnerseekerslikelychangesasnewtechnologiesemerge
andthusshouldbeaccountedfor(WhittyandYoung2016).
Second,researchhasshownthatpeopleuseonlinemeetingtoolsfordifferent
purposes,withmotivationsrangingfromcasualsextocommittedlongtermunions
(Gudelunas2012;Sumteretal.2017).Individualswhoattachgreatervaluetolong
termrelationships (e.g., more marriageoriented,ascribing more social value to
romanticunions)mayspecicallychooseinternetdatingtosearchmoreeasilyfor
marriagecandidates.Prepartneringdifferencesinvalueorientationbetweenonline
andofinedaters—andnotnecessarilybettermatchingoccurringonline—couldthen
explaindifferentspeedsofprogressiontowardmarriage.Todealwiththesemultiple
sourcesofbias,Icontrolforacomprehensivesetofpotentiallyconfoundingobserved
factorsandalsoconductacomplementarysetofanalysesexaminingprepartnering
pat terns.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1985Online Dating and Educational Inequalities in Marriage Formation
Data and Methods
Data
IusedtheGermanFamilyPanel(pairfam),release11.0(Brüderletal.2020),alongi
tudinalsurveydatasetcontainingdetailedyearlyinformationonindividuals’socio
demographicprole,preferences,andvalues,aswellasthecontextinwhichtheymet
theirpartner(ifinaunion).Thepairfam data con tain infor ma tion on the part ner ship
trajectoriesofaninitialsampleof12,402randomlyselectedmenandwomenwhoare
nationallyrepresentativeofcohortsbornin1971–1973,1981–1983,and1991–1993.
DemoDiff,whichconsists ofanoversampleof 1,489 EasternGerman respondents
bornin1971–1973and1981–1983(Kreyenfeldetal.2012), was ini ti ated in par al lel
with pairfam’s Wave II. DemoDiffwasseparately conductedforthreewaves until
WaveV,whenitwasfullyintegratedintopairfam.InWaveXI,arefreshmentsam
plewasadded,includingapproximately5,000respondentsfrombirthcohorts1981–
1983and1991–1993aswellasanew, youngercohort(2001–2003);becausethe
currentanalysistargetsrespondentswithatleasttwomeasurementpoints(forwhom
change over time can be traced), none of the respon dents in the refresh ment sam ple
wasincludedinthesample.Responseratesforpairfamwereapproximately30%to
45%ateachwave,whichiscommonforlargescalesurveysconductedinGermany
(Brüderletal.2019).Adetaileddescriptionofthestudyanditscohortstratiedran
domsamplecanbefoundinHuininketal.(2011).Thedataideallysuittheobjectives
ofthisstudy:theyrecordinformationonadulthood,astage thatnotonlyisdemo
graphicallydenseinfamilyformationevents(Rindfuss1991) but also entails a high
degreeoffamiliaritywiththeinternetanditsmultiplesocialuses(Helsperandvan
Deursen 2015). In addi tion, the panel design of pairfamperfectlytstheaimoftrack
ing partnering tran si tions across time.
Toanalyze competingtransitions to marriage,longtermcohabitation, orunion
dissolution (thelatter includingthevepartnerships thatendedin partner’sdeath)
asopposedtoremaininginanonresidentialunion,Icreatedapersonpartnershiple
basedonthe11availablewaves(2008/2009–2018/2019).Participantswerecensored
oncetheyexperiencedoneoftheeventsofinterestor,ifnotransitionoccurred,atthe
lastinterview.Fortheconstructionofthedataset,Irstdiscardedthosewhowere
continuouslysinglewhileinthepanel(n =5,260).Certainrespondents(n = 157) had
morethanonerelationshipspellwiththesamepartner.Assumingthatconditions
leadinguptomarriageformationdifferedovertime,Ichosenottoexcludethesecond
transition.Needingtoaccountforinitialassortmentinmaritalandgendervalues(i.e.,
similarityasclosetothepointofrelationshiponsetaspossible)ratherthanconverged
ideologiesovertime(AxinnandBarber1997;Hakim2003),Ididnotincluderespon
dentsinongoingmarriages(whocouldhaveprovidedonly postmarriageinforma
tiononvalues)norotherretrospectivelyrecordedrelationships(e.g.,ineventhistory
calendars),forwhichinformationonmeetingcontextwasalsomissing.Thisledto
theexclusionof6,474cases.Ialsoremovednonheterosexualrespondents(n = 129),
thosewhocontributedonlyoneobservationperpartnership(n =2,708),andpartner
shipsthatstarted whentherespondent wasyounger thanage18(n =624). Finally,
Iexcludedparticipantswhohadmissing informationonkey variables(n = 66). To
avoidhavingcasesofzerosurvivaltime(n =116)dismissedintheanalysis,Iadded
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1986 G. Potarca
asmallunit(0.5)tothetimevariable,asisstandardpractice(Alexandersson2015).
Asaresultofallrestrictions,Irananalysesonasampleof3,561partneredindivid
uals,with4,043observations(i.e.,partnershipspells)and1,240recordedmarriage
for ma tion events.
Measures
The depen dent var i able is sur vival time fromrelationship entryuntilthe occurrenceof
marriage,cohabitation,orbreakup.Themajorityofrespondentswhotransitionedtomar
riageexperiencedaspellofcohabitationbeforemarrying.Thenumberofcoupleswho
metonlineandtransitionedstraightintoamaritalunionwas,however,toosmall(n = 14)
to war rant a dif fer en ti a tion between direct and indi rect mar riage. I cen sored time at the
dateofthelastinterviewifnotransitionwasobserved—thatis,ifindividualsremainedin
non res i den tial part ner ships. The var i able was constructed on the basis of infor ma tion on
relationshipduration,cohabitationduration,ormarriageduration(inmonths).
To cap ture meet ing set ting,Iusedinformationonhowrespondentsmettheirpart
ners.Themeasureallowsforasingleanswerfromthefollowingoptions:(1)school,
training,work;(2) hobby,club,sports;(3) bar,nightclub;(4) friends oracquain
tances;(5)relatives;(6) apersonalad;(7)theinternet;(8)vacation;and(9) other.
Allnondigitalsettingsaregroupedunderasingle(0)ofinecategory.OnlyinWave
IVdidpairfam begintodistinguish betweentwoonline settings(meetingthrough
aninternetpartnerndingservicevs.meetingthroughonlinesocialnetworks,chat
rooms,andsoon);startingwithWaveX,itfurtheraddedthepossibilityofhavingmet
throughdatingapps.Tomaximizethedataathand,andinlinewithearliertheoretical
arguments,Iusedabroadonlinecategoryforall11waves.
Thesecondkeypredictorisedu ca tional level,categorizedasnontertiaryeduca
tion(rangingfromnodegree,completedtherstandsecondstageofbasiceducation,
highschooleducation,tocompletedpostsecondaryeducationaltrainingmeanttopre
pareforlabormarketentryand/ortertiaryeducation)andtertiaryeducation(abach
elor’sdegreeand/orpostgraduatestudies).Therstcategoryislargelycomposedof
respondentswithsecondaryeducation;thenumberofindividualswithprimaryedu
cationwhomettheirpartneronlineistoosmalltoconsiderseparately.Giventhatthe
analysistargetedcouplesintheirlastyearofparticipationorduringtheyearofexpe
riencingaspecicevent,educationwastimeconstantandxedatthemostrecently
observedlevel.Eventhoughremovingrespondentsyoungerthan18wasintendedto
minimizetheamountofintraindividualchangesineducation,thisdecisionaddition
allyensured,forinstance,thatrespondentswhoenteredapartnershipwhilehaving
onlyasecondarydegreebutwhoenrolledinandcompletedahighereducationpro
gramduringthecourseoftherelationshipwereregardedashighlyeducated.
Basedonrespondents’andtheirpartners’education,Ithenconstructedapredictor
of edu ca tional homog amy, sin gling out pairs in which the two part ners had the same
levelofeducation.Furthermore,Iusedinformationoncountryoforiginforbothpart
nerstocomputeanindicatorofhomogamyonorigin(i.e.,borninthesamecountry).
Ialsoconsideredanindicatorofhomogamybasedonmothers’countryoforigin(and
analysesincludingthismeasurerevealedidenticalresults);givenagreateramountof
missingness in these data, how ever, the for mer was pre ferred. Information on reli gion
and paren tal social back ground for both part ners was unavail able in the data.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1987Online Dating and Educational Inequalities in Marriage Formation
Tomeasurepartnermatchingonfamilyvaluesandtominimizetheriskofendog
eneityandconvergencein valuesovertime,Iusedtherstrecordedmeasurement
ofbothpartners’valueorientations,similartoNitscheandGrunow(2016).Because
familyvaluesweremeasuredeveryotheryear startingwithWaveI,Iusedcouple
informationfromWaveVII,forinstance,iftherelationship was rstobservedin
WaveVI.Whendataonvaluesweremissing,toavoidtrimmingthesampleevenfur
ther(giventhatnotallrespondentsgaveconsenttotheirpartners’participation,and
not all part ners agreed to par tic i pate them selves), I instead used infor ma tion from the
nearestwavewhenfamilyvalueswererecorded.
To measure marriage value orientation, I rst computed a scale of tra di tional
mar riage ori en ta tion,constructedbysummingscoresforthefollowingitems:“You
shouldget marriedif youpermanentlylivewithyourpartner,”“Marriageisalife
longunionthatshouldnotbebroken,”and“Couplesshouldmarryatthelatestafter
achildis born.”Thesestatementsweremeasured ona5point scaleranging from
(1) “disagree completely” to (5) “agree completely.” Cronbach’s alpha was .673
among respon dents and .679 among part ners. To cap ture atti tudes toward women’s
labormarketparticipation—specically,towardthereconciliationbetweenmaternal
employmentandchildcare—Iusedanitemmeasuringagreementwiththestatement,
“Achildunder6willsufferfromhavingaworkingmother.”Eventhoughpairfam
includedanadditionalitemrequestingagreementonwomen’spaidwork(“Women
shouldbemoreconcernedabouttheirfamilythanabouttheircareer”),theformerwas
pre ferred given its greater var i a tion in responses.
Finally,togaugeattitudesregardingmen’sinvolvementinthedomesticsphere,I
reliedonanitemmeasuringagreementwiththestatement,“Menshouldparticipate
inhousework tothesameextent aswomen.”Comparableto Nitsche andGrunow
(2016),Icombinedtheanswersprovidedbybothpartnersforallthreetypesofval
uesandconstructedfourcategories:(1)themaleandthefemalepartnersharemodern
viewsonfamily,(2)botharetraditionallyoriented,(3)onlythewomanhasmodern
attitudes(mismatch1),and(4)onlythemanisprogressivelyoriented(mismatch2).
Nevertheless,becausecouplesinwhichthefemalepartnerendorsestraditionalviews
onfamilylifeandthemalepartnerismoreprogressivewereveryfew,Icombined
categories(2)and(4)tojointlyrefertocouplesinwhichthefemalepartnerhascon
ser va tive beliefs, irrespective of her part ner’s value ori en ta tion.
Givenahighproportion of data missingforpartners’ familyvalues(e.g.,only
44.3% provided data on attitudes toward maternal employment), I investigated
whethercertainfactorspredictedtheabsenceofsuchinformation.Resultsreported
inTableB1intheonlineappendix(sectionB)revealthatthelowereducated,those
withamigrationbackground,andrespondentsinmorerecentlyformedunionswere
morelikelyto have missingdataonpartners’values. Forwomenonly, thosewho
mettheirmatchofinealsoseemedtobeoverrepresented.Oneimplicationofthese
patterns—particularly the overrepresentation of the lowereducated—for analyses
assessingtheroleofvaluecompatibilityonthelikelihoodofmarriageisthatthemag
nitudeofeducationalgapsmaydifferfromthoseseenintherstsetofanalyses.To
checkwhethermainorinteractiveeffectsofmeetingcontextandeducationchange
comparedwithresultsobtainedforthefullanalyticalsample,Ireranthemainmodel
onthisrestrictedsample,withfairlycomparableresults(seeTable 3).
Theanalyses alsoincludedthefollowingcontrolmeasures: whetheremployed,
migrationbackground,residenceinEastGermany,andtheage(linearandsquared)
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1988 G. Potarca
andyearwhen therelationship began.Thelatterwasrecodedinto twocategories,
distinguishingbetweencouplesformedbefore(1989–2011)andafter2012.Thesec
ondcategorycoversunionsinitiatedonlineinthepost–datingappperiod(e.g.,Tinder
waslaunchedin2012).Indicatorsofwhetherrespondentswerepreviouslymarried
orhadchildrenwhentheyrstgottogetherwereomittedgiventheirhigh correla
tionwithageattherelationshipstart.Theywereneverthelessincludedindescriptive
analyses.
Foradditionalanalysesexploringtheroleofselectionintoonlinedating,Iconsid
eredtheinclusionoftwoothervariables:thesocial sta tus value of a part ner ship,cap
turedviaaquestionasking,“Howstronglydoyouexpecttoexperienceanincreased
socialstatus becauseofyourpartner?,” with answercategoriesfrom(1) “notatall”
to(5)“absolutely”;andtheveconstructsofper son al ity(neuroticism,extraversion,
agreeableness,conscientiousness,andopenness),measuredonavalidated21itemver
sionoftheBigFiveInventory(RammstedtandJohn2005).Foreachitem,respondents
ratedtheiragreement usinga5pointLikerttype scale rangingfrom(1)“absolutely
incorrect”to(5)“absolutelycorrect.”Personalitywasmeasuredthreetimesthroughout
thepanel(i.e.,inWavesII,VI,andX).Forrespondentswhoremainedinthepanellong
enoughtobesurveyedmorethanonce,Iconsideredonlytherstmeasurement.
Methods
Toassesseducationalgapsinhowmeetingonlineinuencesthetransitiontomar
riage, I relied on Fine-Gray com pet ing risks mod els(FineandGray 1999), which
focusedonthe subhazard ofexperiencinganeventof interest(i.e.,marriage) asa
functionof timespentinarelationship.MorerealisticallythanaCoxproportional
hazardmodel,italsoaccountedfortwopossiblealternativerisks(i.e.,cohabitation
anddissolution).Thesubhazardofmarriagewasdenedasfollows:
(1)
hmarriage (t)
is theinstantaneousprobabilityofmarriageoccurringattimet, pro vided
that no event occurred before t(Clevesetal.2011).Themodelforthesubhazardof
marriagethentookthefollowingform:
hmarriage (t|x)=hmarriage,0(t)exp(xβ).
 (2)
hmarriage,0
(t
)
representsanonparametricestimationofthebaselinesubdistributionhaz
ard for mar riage, whereas βsareregressioncoefcientsinlogsubhazardratioform.
The cumulative incidence function dening the incidence of marriage occurring
whileaccountingforalternativeriskswaslatercalculatedasfollows:
C
IFmarriage (t)=1−exp −
0
t
∫hmarriage (t)dt
{ }
, (3)
where
0
t
∫hmarriage (t)dt  is the cumulative marriagespecic subhazard function. As
hmarriage (t)=lim
Δt→0
Pr (t≤T<t+Δt, event =marriage|T>t or (T≤t and event ≠marriage)
Δt
⎛
⎝
⎜⎞
⎠
⎟.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1989Online Dating and Educational Inequalities in Marriage Formation
opposedtotheCox regression variantfordealing with competingrisks,the Fine
Grayapproachretainsparticipantswhoexperienceanalternativepartnershiptransi
tionintheriskset.Anadditionaladvantage,relevanttothisstudy’sgoalofassessing
the effect of meet ing online and edu ca tion on the cumu la tive inci dence func tion, is
themorestraightforward handlingofcovariateeffects. Becausesomerespondents
reportedseveralpartnerships,Iusedrobuststandarderrorstocorrectforthenoninde
pendenceofpartneringepisodesclusteredwithinindividuals.Theanalysiswastted
usingStata’sstcrreg com mand.
Results
Table 1providesdescriptive statisticsbymeetingsetting.Of4,043 observedpartner
ships formed between 1989 and 2018, 13.7% consisted of couples that met online.
Additionalexplorations(notshown)revealedthatinthelastveyears,oneinve(i.e.,
20.5%)unionsbeganonline.Table 1 fur ther more shows that com pared with individuals
incouplesformedofine,thosewhomettheirpartneronlinewerelessoftenhighlyedu
cated.Formenwithnontertiaryeducation,homogamywassignicantlymorecommon
forrelationshipsinitiatedonlinethanelsewhere.Highlyeducatedmen(butnotwomen)
were less homogamous and more likely to partner down on education online than
ofine(formoredetailedinformationoneducationalpairingsacrossmeetingsettings
thatalsodistinguishesrespondentswithprimary,secondary,andtertiaryeducation,see
TableC1intheonlineappendix,sectionC).Furthermore,menincouplesthatformed
onlinewerelesslikelytohaveapartnerofthesameorigin(i.e.,countryofbirth)but
attachedmoresocialstatusvaluetotheirunionthantheonesincouplesformedofine.
Forwomen,sociodemographichomogamy didnotvaryacrossmeetingcontext.Fur
thermore,respondentswhomet their partneronlineweremore frequently employed
(especiallymen),older,andmoreoftenpreviouslymarriedandwithchildrenatthestart
oftherelationshipthanthosewhomettheirpartnerofine.Thetwogroups,however,
didnotdifferintermsofpersonality.
Table 1alsoindicatesthatamongmen,unionsformedonlineweresignicantly
more likely to include partners who share traditional marriage views. Additional
crosstabulations across genderandeducational level(seeTableC2intheonline
appendix,sectionC)revealthatthiswasalsothecaseamongwomenbutonlyamong
thehighlyeducated.Morespecically,TableC2showsthatwhereaslowereducated
womenwerelesslikelytomatchwiththeirpartnerintermsofconservativemarriage
valuesiftheymetthemonline(13.3%)thanofine(25.2%),compatibilityregarding
traditionalmarriageviewsforthehighlyeducatedwasmorelikelytooccurifthey
metonline(29.3%)thanofine(19.1%).Furthermore,amonghighlyeducatedmen
andespeciallywomen,bothpartnersholdingprogressiveviewsregardingmothers’
participationinpaidworkwasmorecommonamongonlineinitiatedcouples.Among
respondentswithnontertiaryeducation,however,coupleswerelessfrequentlypro
gressiveandmoreofteninunionswhereonlythefemalepartnerheldmodernviews
onmaternalemployment.Whenitcomestovaluespertainingtomen’sinvolvement
inthedomesticsphere,particularlyamonglowereducatedmenandhighereducated
women, there is a greater chance of both part ners shar ing mod ern val ues and a lower
chanceofmismatchesamongcouplesformedonlinethanofine.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1990 G. Potarca
Table 1 Sampledescriptivestatistics,bymeetingcontext:pairfamWavesI–XI(2008–2019)
MaleSample FemaleSample
Ofine Online Sig. Ofine Online Sig.
TypeofTransition(%)
 Notransition 16.6 11.5 15.5 13.6
Marriage 31.4 35.8 29.7 28.2
 Cohabitation 34.1 36.6 38.5 45.0
 Breakup 17.9 16.0 16.3 13.3
Respondent’sEducation(%)
 Tertiary 42.9 38.3 42.8 37.2 †
EducationalHomogamy(%)
 Nontertiary×homogamy 63.7 73.3 * 62.0 63.4
 Tertiary×homogamy 56.4 50.5 58.7 60.0
OriginHomogamy(%) 91.0 86.4 * 88.6 88.6
MarriageValues’Match(%) **
 Bothmodern 55.4 45.2 60.2 59.4
 Bothoronlythewomantraditional 28.4 43.3 22.4 20.3
 Onlythewomanmodern 16.2 11.5 17.4 20.3
GenderValues’Match:Women’sPaid
Work(%)
 Bothmodern 26.5 24.8 28.7 31.8
 Bothoronlythewomantraditional 45.5 49.5 46.4 41.7
 Onlythewomanmodern 28.0 25.7 24.9 26.5
GenderValues’Match:Men’sDomestic
Work(%)
 Bothmodern 71.3 77.1 70.7 75.2
 Bothoronlythewomantraditional 14.1 10.5 12.0 9.8
 Onlyshemodern 14.6 12.4 17.3 15.0
Employed(%) 77.5 83.1 * 63.8 64.7
MigrationBackground(%) 15.8 12.3 16.7 14.9
PreviouslyMarried(%) 6.6 12.3 ** 12.5 23.6 ***
Childrenatt1(%) 9.1 14.8 ** 22.9 32.0 ***
LivinginEasternGermany(%) 28.6 29.6 30.2 31.4
Yearatt1:After2012(%) 29.2 46.1 *** 31.7 43.0 ***
PartnershipDurationatMarriage:
Range0.5–283(mean) 71.97
(45.95)
39.68
(26.60)
*** 74.96
(46.57)
47.33
(29.75)
***
Ageatt1:Range18–46(mean) 25.69
(6.34)
28.61
(7.31)
*** 25.51
(6.76)
29.09
(7.35)
***
Neuroticism:Range1–5(mean) 2.52
(0.74)
2.39
(0.76)
2.92
(0.83)
3.01
(0.82)
Extraversion:Range1–5(mean) 3.50
(0.79)
3.42
(0.78)
3.65
(0.79)
3.67
(0.77)
Agreeableness:Range1–5(mean) 3.20
(0.68)
3.22
(0.55)
3.26
(0.75)
3.20
(0.71)
Conscientiousness:Range1.5–5(mean) 3.75
(0.64)
3.73
(0.58)
3.92
(0.61)
3.82
(0.67)
Openness:Range1.4–5(mean) 3.62
(0.69)
3.56
(0.61)
3.66
(0.73)
3.74
(0.72)
SocialStatusValueofUnion:Range
1–5(mean) 2.07
(1.03)
2.26
(1.06) **
1.87
(0.97)
1.86
(0.98)
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1991Online Dating and Educational Inequalities in Marriage Formation
MaleSample FemaleSample
Ofine Online Sig. Ofine Online Sig.
NumberofObservations 1,675 243 1,816 309
NumberofIndividuals 1,512 236 1,610 293
Notes:t1 =therstyearofpartnership.Standarddeviationsareshowninparentheses.
†p <.10;*p <.05;**p <.01;***p < .001
Table 1 (continued)
Fine-Gray Competing Risks Models
I now present the results of a competingrisks analysis predicting transitions into
marriage(subhazardsreportedinTable 2). Figure1 also pro vi des men’s and women’s
predictedcumulativeincidence curvesofentryintomarriagebasedonthisestima
tion,acrosseducationalgroupandbymeetingsetting.Forthegraph,Irestrictedthe
rela tion ship dura tion to 180 months because the occur rence of events among cou ples
whometonlinerarelyextendedbeyondthiswindow.Aspreviouslynoted,themodel
treatstransitioningintoacohabitingunionanddissolvingthepartnershipasalterna
tiveriskstomarrying.Resultsinconnectiontoeventsotherthanmarriagearesecond
arytothisstudy,andIthereforedonotexpandonthem.Nevertheless,resultsinTable
2showthatmeetingcontextandeducationwerenotassociatedwiththetransitionto
cohabitationforeithergender andthathigheducatedwomenhadamarginallysig
nicantlowerriskofuniondissolutioniftheymettheirpartneronlineversusofine.
Focusingontheeventof interest,Figure1rstindicatesthatfor menwhomet
theirpartnerofine,there wasnopronouncedgapbetweenthosewithtertiaryand
nontertiaryeducation.Amongwomenwhofoundtheirpartnerofine,however,there
wasasignicanteducationalgap,withthelowereducatedmorelikelytotransition
intomarriage thanthehighereducated.Thegraphthen showsthatmeeting online
wasassociatedwithgreaterchancesofmarryingformen,irrespectiveofeducation;
forwomen,thehighlyeducatedwhomettheirpartneronlineweresignicantlymore
pronetotransitiontomarriagethanthelesseducated.Becauseofsmalldifferences
intheincidenceofmarriageacrossmeetingcontextforthosewithnontertiaryedu
cation,thereversaloftheoriginalpatternwasduetothesignicantincreaseexperi
encedbywomenwithtertiaryeducation.
The esti ma tes reported in Table 2provideamorepreciseindicationofthemagni
tude of this increase. Given that the mod els include both main and inter ac tive effects
ofeducation and meeting context,thesubhazardformeeting online, for instance,
representstheestimatedeffectforthereferencecategoryofeducation(i.e.,tertiary
education).Therefore,thesubhazardratioof1.859shows thatwithtime andmul
ti ple sociodemographic covariates con trolled for and with com pet ing events also
allowedtooccur,themarriagesubhazardforhighlyeducatedwomenwhomettheir
partneronlinewas85.9%ofthatforhighlyeducatedwomenwhomettheirpartner
ofine.Theestimatedeffectof meetingonlineforwomen withnontertiaryeduca
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1992 G. Potarca
Table 2 SubhazardratiosfromFineGraycompetingrisksmodelsof(1)marriage,(2)cohabitation,and
(3)breakup:pairfamWavesI–XI(2008–2019)
MaleSample FemaleSample
Marriage Cohabitation Breakup Marriage Cohabitation Breakup
Meeting Online 1.553* 0.949 0.660 1.859*** 0.971 0.604†
(0.278) (0.187) (0.222) (0.290) (0.168) (0.174)
Nontertiary 1.065 0.904 0.978 1.216* 0.845 0.862
(0.087) (0.090) (0.117) (0.095) (0.091) (0.102)
Meeting Online ×
Nontertiary 1.011 0.718 1.526 0.617* 1.103 1.548
(0.248) (0.198) (0.598) (0.141) (0.258) (0.544)
Employed 1.943*** 1.228 0.653*** 0.975 1.568*** 0.849
(0.263) (0.170) (0.081) (0.076) (0.160) (0.095)
Migration
Background
1.422*** 0.484*** 1.125 1.153 0.750* 1.063
(0.152) (0.079) (0.163) (0.120) (0.105) (0.151)
LivinginEastern
Germany 0.798** 1.163 0.959 0.927 1.323** 0.907
(0.067) (0.120) (0.133) (0.071) (0.126) (0.113)
Yearatt1After
2012 0.686** 1.154 0.763* 0.810 1.137 0.656**
(0.095) (0.112) (0.097) (0.113) (0.110) (0.086)
Ageatt11.412*** 1.087 0.628*** 1.496*** 1.012 0.687***
(0.083) (0.064) (0.040) (0.090) (0.049) (0.042)
Ageatt1,Squared 0.995*** 0.998 1.007*** 0.993*** 1.000 1.006***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Partnership
Duration 1.062*** 0.835*** 1.057*** 1.066*** 0.727*** 1.088***
(0.006) (0.022) (0.016) (0.005) (0.023) (0.019)
Partnership
Duration,
Squared
0.9995*** 1.002*** 0.999*** 0.9996*** 1.005*** 0.998***
(0.000) (0.001) (0.000) (0.000) (0.001) (0.000)
Partnership
Duration,Cubed 1.000001*** 0.99999* 1.000003*** 1.000001*** 0.99997*** 1.000004***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
LogPseudo
Likelihood −4,100.8 −4,298.6 −2,354.2 −4,267.3 −5,337.2 −2,402.8
Wald χ2(df)502(12) 525(12) 172(12) 478(12) 611(12) 87(12)
Prob> χ20.000 0.000 0.000 0.000 0.000 0.000
Numberof
Observations 1,918 2,125
NumberofEvents 613 660 339 627 838 337
Numberof
Competing
Events 999 952 1,273 1,175 964 1,465
NumberofCensored
Cases 306 323
Numberof
Individuals 1,697 1,855
Notes:Robuststandarderrorsareshowninparentheses.t1 =therstyearofpartnership.
†p <.10;*p <.05;**p <.01;***p < .001
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1993Online Dating and Educational Inequalities in Marriage Formation
Fig. 1 Men’sandwomen’spredictedcumulative incidenceof marriage,bymeeting contextandeduca
tionallevel.ThegureisbasedonestimatesreportedinTable 2.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1994 G. Potarca
tionisgivenby1.859× 0.617 = 1.147. This means that for those with less edu ca tion,
meetingtheirpartnerthroughtheinternetwasassociatedwithasignicantlysmaller
increase(14.7%)intheincidenceofmarriage.TheseresultsconrmHypothesis1,
whichpredictedthatespeciallyforwomen,usingdigitaltoolsformateselectionposi
tionsthehighlyeducatedatamarriageadvantagecomparedwiththelowereducated.
Furthermore,themechanism suggested as triggeringimprovementinthemari
talchancesofhighlyeducatedwomenusinginternetdatingtondpartnerswasthe
bet ter matching on tra di tional mar riage atti tudes and pro gres sive views on gen der.
Table 3presentsmodelstestingwhethervaluecompatibilitymediatedtheassociation
between meet ing online and the tran si tion to mar riage.
Asnotedearlier,givenasignicantreductioninsamplesizewhenincludingmea
suresofpartners’ideologicalpairings,Irerantheanalysesonthisrestrictedsample
(Model1).The resultsarefairlysimilartothosepreviouslypresented,witha few
exceptionsobservedamongwomen.1Regardingthemodelbuildingstrategy,Model
2addsmeasuresofeducationalandoriginhomogamy;Model3includestheeffectof
matchingonmarriagevalueorientation;andModels4and5addtheeffectofmatch
ingongendervalues intermsof mothers’paidworkand men’sdomesticinvolve
ment,respectively.
To assess their indi vid ual mediating con tri bu tion, I esti mated the two mea sures
ofgendervaluematchinginseparatemodels.Furthermore,toalignwiththeoretical
arguments,Iusedadifferentbaselinecategoryforthetwotypesoffamilyvalueide
ology,suchthatresultspresenttheeffectofcouplessharingtraditional(vs.modern)
marriagevaluesandtheeffectofcouplessharingmodern(vs.traditional)genderval
uesalongsidetheeffectofcouplesinwhichonlythefemalepartnerheldprogressive
familyvalues.Giventhateducationaldifferencesinwomen’sonlinedatingadvan
tageweredrivenbythelargeincreaseintheincidenceofmarriageamongthehighly
educated(withlittle variationinthemarital chancesofthelowereducated across
meeting context), I mainly inspected changes in the subhazard of meeting online
(i.e.,theeffectcorrespondingtouniversityeducatedwomen).Nevertheless,tovisu
alizehowtheadditionofeachcovariateinuencestheonlineeducationalgradientin
marrying,FiguresD1(formen)andD2(forwomen)intheonlineappendix(section
D)presentthepredictedcumulativeincidenceofmarriagebyeducationandmeeting
contextfromthesemodels.
Asexpected,resultsshowthatsociodemographichomogamydidnothaveasignif
icanteffectinitself,nordiditsinclusionsubstantivelyaltertheassociationbetween
meetingonlineandentryintomarriage.Addingameasureofmatchingonmarriage
valuesinModel3,however,reducedtheeffectofmeetingonlineforbothhighlyedu
catedmen(from58%inModel2to42.1%)andwomen(from54.7%to48%).Forthe
former,theeffectofmeetingonlinewasalsonolongerstatisticallysignicant.For
thelatter,accountingforhowcouplespairupintermsofmarriageattitudesslightly
diminishedtheofineadvantageofwomenwithnontertiaryeducation(from26.5%
to22.4%).ResultsinconnectiontoModel3furthermorerevealthatforwomen,cou
1 Themaineffectofmeetingonlinedecreasedinsizebutremainedsignicant.Theinteractiveterm“meet
ing online ×nontertiary”loststatisticalsignicance—potentiallybecause thelowereducatedwere more
likelytohavemissinginformationonpartners’values—butretaineditssize.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1995Online Dating and Educational Inequalities in Marriage Formation
Table 3 SubhazardratiosfromFineGraycompetingrisksmodelsofmarriage(cohabitationandbreakupasalternaterisks):pairfamWavesI–XI(2008–2019)
MaleSample FemaleSample
Model 1
(restricted
sam ple 1) Model 2 Model 3 Model 4 Model 5
Model 1
(restricted
sam ple 1) Model 2 Model 3 Model 4 Model 5
Meeting Online 1.610* 1.580* 1.421 1.457†1.429 1.510* 1.547* 1.480* 1.460†1.475†
(0.336) (0.327) (0.310) (0.319) (0.310) (0.287) (0.296) (0.292) (0.287) (0.293)
Nontertiary 1.133 1.122 1.102 1.128 1.108 1.265* 1.265* 1.224* 1.277* 1.231*
(0.113) (0.114) (0.113) (0.119) (0.113) (0.124) (0.124) (0.119) (0.129) (0.123)
Meeting Online ×Nontertiary 0.986 1.006 1.134 1.105 1.095 0.679 0.679 0.719 0.746 0.731
(0.314) (0.320) (0.365) (0.355) (0.357) (0.212) (0.209) (0.230) (0.241) (0.233)
EducationalHomogamy 1.070 1.070 1.072 1.061 1.122 1.155 1.141 1.156
(0.106) (0.105) (0.107) (0.104) (0.112) (0.117) (0.115) (0.116)
CountryOriginHomogamy 0.775 0.775 0.774 0.791 1.174 1.241 1.229 1.224
(0.177) (0.189) (0.189) (0.196) (0.281) (0.310) (0.306) (0.298)
MarriageValues’Match
(ref.= both mod ern)
 Bothtraditional 1.421** 1.461*** 1.431*** 1.556*** 1.627*** 1.549***
(0.156) (0.166) (0.154) (0.185) (0.195) (0.184)
 Onlythewomanmodern 1.196 1.230 1.201 1.558*** 1.608*** 1.591***
(0.174) (0.177) (0.174) (0.194) (0.200) (0.195)
GenderValues’Match:
Women’sPaidWork
(ref.= both tra di tional)
 Bothmodern 1.223†1.277*
(0.143) (0.146)
 Onlythewomanmodern 1.003 1.072
(0.123) (0.133)
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1996 G. Potarca
MaleSample FemaleSample
Model 1
(restricted
sam ple 1) Model 2 Model 3 Model 4 Model 5
Model 1
(restricted
sam ple 1) Model 2 Model 3 Model 4 Model 5
GenderValues’Match:Men’s
DomesticWork(ref.= both
tra di tional)
 Bothmodern 0.913 0.844
(0.110) (0.123)
 Onlythewomanmodern 0.728†0.695*
(0.132) (0.126)
LogPseudoLikelihood −2,484.4 −2,483.4 −2,478.8 −2,477.3 −2,477.2 −2,369.5 −2,368.7 −2,360.2 −2,358.3 −2,358.4
Wald χ2(df)292(12) 293(14) 308(16) 306(18) 332(18) 231(12) 234(14) 255(16) 260(18) 257(18)
Prob> χ20.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
NumberofIndividuals 866 866 866 866 866 830 830 830 830 830
NumberofObservations 910 910 910 910 910 871 871 871 871 871
NumberofEvents 415 415 415 415 415 396 396 396 396 396
NumberofCompetingEvents 417 417 417 417 417 410 410 410 410 410
N(censored) 78 78 78 78 78 65 65 65 65 65
Note:Robuststandarderrorsareshowninparentheses.
†p <.10;*p <.05;**p <.01;***p < .001
Table 3 (continued)
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1997Online Dating and Educational Inequalities in Marriage Formation
ples in which the man had con ser va tive mar riage val ues, irrespective of woman’s
ideology,hadahigherchanceofmarryingthancouplesinwhichbothpartnerswere
moremodern. Formen,it wasmainlycouples wherebothpartners hadtraditional
marriagevaluesthatweresignicantlymorepronetomarriage.
ResultsfromModel4furthermoreindicatethatinthecaseofwomen,whenpart
ners’relativeideologyregardingmaternalemploymentwascontrolledfor,themar
riagesubhazardforthehighlyeducateddecreasedtwoadditionalpercentagepoints.
Althoughaddingameasureofcompatibilityinmarriagevalueorientationprompteda
greaterdecreaseintheeffectsizeofmeetingcontext,itwasaccountingformatching
onbothmarriageattitudesandgenderideologythatrenderedtheeffectstatistically
insignicant(atthep <0.05level).Resultsalsorevealthatfemalerespondentswhose
partnerssharedmodernviewsonwomen’spaidworkweresignicantlymorelikely
to marry compared with women in more traditional couples. For men, this effect
wassmallerinbothsizeandsignicance,anditsinclusiondidnotreducebutrather
slightlyincreasedtheeffectofmeetingonline.
Finally,resultsfromModel5showthatwhen accountingforvaluematchingin
termsofmen’sinvolvementindomestictasks,theassociationbetweenmeetingcon
textandriskofmarriagedecreasedagain forwomen,albeittoa lesserextentthan
inModel4.FigureD2(onlineappendix)alsovisuallyindicatesthattheinclusionof
withincouplegenderideologyconcerning maternalemploymenthadagreateroff
setting effect on the edu ca tional gra di ent of women who met their part ner online than
the inclu sion of value matching concerning men’s par tic i pa tion in domes tic tasks.
The results in Table 3alsorevealthatwomenincouplesinwhichonlytheyhadpro
gressiveviewsonmen’scontributiontofamilywork(butnotwomenincouplesin
whichbothpartnersheldmodernviews)weresignicantlylesslikelytomarrythan
womeninmoretraditional couples.Formen,theeffectofthis mediatorwasmore
mod est, and its inclu sion did not dimin ish the effect of meet ing online. The results
thereforelargelyconrmHypothesis2,whichsuggestedthatespeciallyforwomen,
partners’combinedideologyintermsofbothmarriagevalueorientationandgender
rolespartiallyexplainstheonlinedatingmarriageadvantageofthehighlyeducated
comparedwiththelowereducated.
Supplementary Analyses
First,toaccountfor the potentialselectivity of respondentswhomettheir partner
online, I esti mated mod els with addi tional covariates that had the poten tial to shape
boththecontextofpartnerselectionandtheprobabilityofmarriage,suchasperson
alitytraitsorthesocialstatusvalueofapartnership.Givenalargeamountofnon
overlappingmissingvalues,whichwouldhaveexcessivelytrimmedthesubsample
of respon dents who met their part ner online, I opted for mod els testing the effect of
eachsetofcovariatesseparately(TableE1,onlineappendix,sectionE).Becauseof
asignicantreductioninsamplesizewhenaddingpersonalitymeasures,Irstrepli
cated Model 1 on the sub set of respon dents who pro vided infor ma tion on these items.
Theresultsformenshowthatthecoefcientformeetingonlinewasstillsubstantial
butstatisticallysignicantatthep <.10levelonly.Forwomen,meetingonlinewas
stillstronglyandsignicantlyassociated withthesubhazard ofmarriage.Forboth
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1998 G. Potarca
groups,the inclusionofpersonality factors slightly reduced the effect of meeting
contextbutdid not entirelyoffset itseffect.Furthermore,these analysesindicated
the robust ness of pre vi ous results to includ ing a mea sure of social value ascribed
toromanticunions.Lesseducatedwomenwhoserelationshipsformedinnondigital
settingsstillhadahigherprobabilityofmarriage,andmen(irrespectiveofeducation)
andhigheducatedwomenwhoserelationshipsstartedonlinehadincreasedoddsof
mar riage.
Second,toseewhetherhighlyeducatedwomenusingonlinedatingwerenotjust
morepronetoconverttheirunionintomarriagebutalsohadgreaterchancesofnding
apartnertobeginwith,Ifocusedontheprepartneringstageandinvestigatedentry
into a roman tic part ner ship among sin gles searching for a part ner online ver sus those
searchingonlyelsewhereinadiscretetimesurvivalanalysis(Allison2014).Begin
ning with Wave III, pairfamaskssingle respondentswhetherthey usetheinternet
tondapartneriftheyrespondedtoapriorquestionassessingiftheywouldliketo
haveapartner,usinga5pointscalerangingfrom(1)“notatall”to(5)“absolutely.”
Ifrespondentsselectedthemiddlevalue(3)orhigher,thenthefollowupquestionon
whethertheyusetheinternettondapartnerwasaddressed.Theadvantageofthis
lteringdesignisthattheresultingsampleincludesonlythoseindividualswhoare
seekingarelationship,minimizingthepossibilitythatonlinedatershaveselectively
stronger(orweaker)partneringintentionsthanofinedaters.Italsohelpsdisentangle
twomechanismsthatpotentiallyleadtolesspartnering:namely,thechoiceofstaying
singleordifcultyndingasuitablepartner.Theobservationalwindowrangesfrom
thestart ofsinglehood(sincethedissolutionofthepreviouspartnership,ifany)or
age18(ifnopreviousrelationshipswererecorded)untiltheyearofenteringaunion,
withrightcensoringoccurringiftheeventofinterestwasnotexperienceduntiltheir
lastyearofpanelparticipation.
Additionalinformation about measuresandmethodis provided insectionFof
theonlineappendix.Theresultsofthisanalysis(reportedinFigureF1,whichplots
predicted prob a bil i ties of starting a part ner ship across gen der, edu ca tion, and search
context)revealthat theuseof theinternetledto anegative shiftinthepartnering
chancesofallgroupsbutlesssoforhighlyeducatedwomen.Thisndingalignswith
previousworkshowingthatusingdigitaltoolsfordatingwasnotnecessarilylinked
tomorepartnerships(Rosenfeld2018). It is pos si ble that online, too much choice and
diversityofoptionsoverwhelms(Schwartz2005), slowing the pro cess of selecting
andinvestinginasingleconnection.Maledatersexperiencedgreaterdelaysinpart
nershipformationthanfemaledaters,conrmingthatmentakemoretimetosearch
and per haps engage in more casual encoun ters when presented with a mul ti tude of
options(YuandKuo2016).
Amongmen,thedataalsoexposeapositiveeducationalgradientirrespectiveof
searchcontext.Amongwomen,whenonlyseekingapartnerofine,thehighlytrained
werelesssuccessfulthanthosewith lowereducation; indigital markets,however,
theydisplayedsubstantiallygreaterpartneringchances,and a positiveeducational
gradientemerged.Additionalanalyses,includingxedeffectsmodels(notshown),
dismissconcernsregarding(un)observedheterogeneity,eitherxedortimevarying,
affecting the relation between online search strategy and partnering events across
edu ca tion and gen der. In short, results reveal that seek ing a part ner in the vir tual
worldperpetuatedpartneringinequality(Schmitz2017) among men while reverting
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

1999Online Dating and Educational Inequalities in Marriage Formation
theeducationalhierarchy amongwomen.Withinthegroup ofsingles usingonline
dating,higheducatedwomenexperiencedthehighestchanceofndingamate.This,
alongsidepreviousresults,conrmsthatwelleducatedwomenarethemostskilled
at tap ping into the abun dant and diverse pool of can di dates pro vided online to part ner
andultimatelymarry.
Finally,inadditionalanalysesbasedonthesamesampleofsinglerespondents,I
soughttoestablishwhetherhighlytrainedwomenwithamoreliberalmindsetregard
inggenderroleswere selectivelydrawn toonlinedating.ResultsreportedinTable
F1intheonlineappendixindicatethatasexpectedgiveneducationaldifferencesin
familyvalues(Kulik2016;MyersandBooth2002),lowereducatedmenandwomen
weregenerallymoretraditionallyorientedintermsofmarriagevaluesandlesspro
gressivewithregardtothedistributionofgenderroles.Online,despiteanoverrep
resentationofhighlyeducatedwomenwithtraditionalmarriageviews,therewasno
oversupplyofhighlyeducated women with progressivegendervalues. Moreover,
thereseemedtobeasmallincreaseinlowereducatedwomenvaluingmen’sinvolve
ment in domes tic tasks among those searching for part ners online. Therefore, the
greater probability of universityeducated women being part of couples with pro
gressiveviewsongenderrolesiftheunionstartedonlinethanofineismorelikelya
consequenceofthematchingmechanismsoccurringonlinethanaresultofselectivity
on val ues.
Discussion
With the wide spread use and accep tance of online instru ments for connecting with
potentialpartners(Smith2016),investigatingpartnermatchingandmaritalunionfor
mationintoday’sdigitaleraishighlywarranted.Identifyingtheimpactthattheinter
nethashadonmarriageisalsoanimportantstepinprovidingscienticevidencethat
responds to grow ing con cerns regard ing the gen eral impact of new tech nol o gies on
sociallife(Chesley2005, 2006).Withindemographicresearchonfamilyformation
andinequalityinparticular,welackinvestigationsintohowonlinedatingtoolsare
shapingcurrenteducationalgradientsinmarriage(CarboneandCahn2014;Kalmijn
2013).Tollthisknowledgegap,Iinvestigatedentryintomarriageamongcouples
whometonlinecomparedwiththosewhometofineviaFineGraycompetingrisks
modelsoftimeuntilmarriage (vs.cohabitationorbreakup).Toprobe whetherthe
internet’s effects on mar i tal pros pects across edu ca tion is related to bet ter matching in
terms of mar riage val ues and gen der role atti tudes, over and above sociodemographic
similarity,Ialsoassessedpartners’ideologicalpairingsasakeyexplanatorymecha
nisminadirectempiricaltest.Finally,theanalysesaddressedconcernsofselection
intoonlinedating either by accountingforabroadset of potentially confounding
observedfactorsorbyexaminingprepartneringpatternsinadditionalanalyses.
Basedon uptodatepaneldataprovidingdistinctivelyrichmultiactorinforma
tiononthepartnering choices ofGermanadultsover time, the investigations rst
foundthatincontrastwiththepositivemarriagegapcurrentlyobservedinmostWest
erncountries(VanBaveletal.2018),butinlinewithearlierndings(Mulderetal.
2006),universityeducatedGermanwomenwhomettheirpartnerintraditionalways
werelesspronetomarrythanthosewithnontertiaryeducation.Formen,thediffer
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2000 G. Potarca
encebetweenthelowerandhighereducatedwasminimalirrespectiveofwherethey
mettheirpartner.Furthermore,men ofbothtertiary andnontertiaryeducationhad
signicantly higher chances of marryingiftheirunionstartedonlinethanofine.
Bothlowerandhighereducatedmendisplayinggreaterpartnerresemblanceontra
ditionalmarriagevaluesiftheirrelationshipstartedonlineversusofine(TableC2,
onlineappendix)mayexplainwhyonlinedatinguniformlyincreasedmaritalchances
amongmen.Forwomenincoupleswhometonline,thegradientwasinverted.Poten
tiallymoredigitallyskilledatmakingbetterchoiceswhenfacedwithamultitudeof
diverseoptions(OllierMalaterreetal.2019),highlyeducatedwomenwhomettheir
matchonlineweresignicantlymorepronetotransitionintoamorecommitted(i.e.,
marital)unionthanlowereducatedwomen.
Accounting for partners’ matching on family values partly offset the marital
advantage of universityeducated German women who met their partner online.
Whereasstatuscompatibilityintermsofeducationorplaceoforiginplayedamini
malrole(KerckhoffandDavis1962), matching in terms of mar riage and gen der role
beliefsmatteredinpartiallyexplainingthehigherchanceofmarryingamonghighly
educatedwomen.First,ndingsspeakofthepivotalrolethatmarriagestillplaysin
Germany(KlärnerandKnabe2017)andamonguniversityeducatedwomeningen
eral(Cherlin2020).Second,inviewofsupplementaryresultsindicatingthathighly
educatedwomen withtraditional marriageattitudesselfselectintodigitalpartner
search(TableF1,onlineappendix),internetdatingseemstobepositionedasalong
termdatingcontext(Lietal.2013),selectivelyappealingtoindividualswithastron
ger focus on lifelong unions and intimacy goals (Sanderson et al. 2007). Rather
thanbeingaspacewherepeoplemakedecisionsbasedonlyonimpulseormarket
like rationalizations, digital partner markets—particularly dating websites—may
infactbe hyperromanticized spaces where traditionalideals of longlasting love
still dominate (Bergström 2011; Dröge and Voirol 2011) but where con ser va tive
genderviews cansimultaneously bechallenged.Asexpected, onlinedating asan
unrestrictedspaceformateselectionincentivizedthehighlyeducated—particularly
womenwithanegalitarianvisionofdoingfamily—toselectpartnerswithasimilar
mindset(Press2004).
Nevertheless, the nding that couples’ alignment on modern views regarding
maternalemploymentimpactedprogresstowardmarriageyetmatchingonprogres
sive atti tudes regard ing men’s con tri bu tion to house hold chores car ried less weight
(andevenhinderedentryintomarriage,althoughtheeffectwasnonsignicant)sug
gests that for marriage to ensue, people cannot exceedingly depart from what is
sociallyascribedasfamily.Despitetheslightoverrepresentationoflowereducated
womenwhobelievedintheequalsharingofdomesticworkamongthoseusingdigi
taltoolsfordating(TableF1intheonlineappendix),prepartneringpreferencesdid
nottranslateintoaparticularlygreaternumberofvaluecompatibleunions.Giventhe
possibilitythatinitialpartnerpreferencesandattitudeschangebasedonthestructural
accessibility ofpartners whosharecompatiblefeatures—that is,theease ofmatch
(Houtsetal.1996)—futureresearchshouldidentifypotentialadaptationsinpartner
pref er ences occur ring through out the whole selec tion pro cess.
Inlinewithearlierevidenceindicatingthatsimilarityingenderroleattitudesis
moreconsequential for women’srelationship outcomes(Ogolskyetal.2014), the
resultspresentedhereadditionallyshowthatmatchingongendervalueshadaweaker
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2001Online Dating and Educational Inequalities in Marriage Formation
mediating effect on men’s subhazard of marriage and that accounting for it (as
opposedtoaccountingformatchingonmarriagevalues)didnotplayaroleinoff
settingmen’soverallonlinedatingmarriageadvantage.Furthermore,thedatareveal
thatmen—andnotwomen—withtertiaryeducationcrossededucationalboundaries
morefrequentlyindigitalthannondigitalsettings(TableC1,onlineappendix)and,
insodoing,perpetuatedstereotypicallygenderedpairingsof highereducatedmen
datinglowereducatedwomen,asseeninpreviousstudies(e.g.,Skopeketal.2011).
Ratherthansignalanincreaseinopenness,heightenedonlineeducationalexogamy
found in previous studies (Thomas 2020) may in fact hide a return to traditional
pairingsamongcertain highlyeducated men.TableC2alsoreveals thatcompared
withthosewhomet theirpartnerofine,higheducated menwhomettheirmatch
onlinewerelesslikelytobepartofcoupleswithliberalviewsonmen’scontribution
tofamilywork,especiallywhenthesemenpartnereddown(asrevealedinadditional
analyses,notshown).
Whetheronecouldgeneralizethesendingstoothernationalcontextsisanopen
question. On the one hand, in more egalitarian contexts, with a generally greater
supplyofegalitarianmen,onlinedatingislesslikelytoestablishitselfasasingular
spacewherehighearningwomengoinsearchofmenwhoarewillingtomarryand
sharehouseholdlabor.Inthesecontexts,wemaythereforeobservefewerdifferences
betweenonlineandofinematchingpatterns.Ontheotherhand,incountriessuch
astheUnitedStates,withawideningpositiveeducationalmarriagegap(Parkerand
Stepler2017)andgrowingeconomicclass divides(Schneider andHastings2017),
digitaldating technologiesmay assisthighstatus individualsinselectingsimilarly
educatedpartnersmoreefciently,exacerbatingratherthaninvertingcurrenteduca
tionalgradients.Therangeofdatingwebsitesoperatingineachcountrymayplayan
additionalrole,withplatformsexclusivelyaimedatprofessionalsoracademics(e.g.,
EliteSinglesintheUnitedStates)potentiallyacceleratingthemaritalsuccessofthe
highlyeducated.Finally,incontextswheremarriageislessnormativelyandinstitu
tionallyendorsed,attachmenttomarriageastheultimatefamilyformmaybelower.
Nevertheless,virtualspacesofpartnerselectionmightstillemergeasthelastoutpost
for peo ple in search of mar riage.
Other directions of inquiry could also be explored. It is worth investigating
whether internet dat ing shifts the tim ing and occur rence of union for ma tion for
membersofsexualminorities, whohavebeen showntobenetmorefromonline
matching(RosenfeldandThomas2012), but for whom the sam ple was too small to
beexaminedinthisanalysis.Moreover,theroleofonlinedatingmatchingonother
characteristics(e.g.,economicresources,religion,personality,leisureinterests,good
looks,orsociability)shouldalsobescrutinized.Tofurthertestwhetherinternetdat
ing facilitates relationships with greater betweenpartner compatibility,the actual
distributionofdomestictaskswithinonlineinitiatedcoupleswillbeexaminedina
followupstudy.Furthermore,thecurrentstudydidnotpresumetograspthecom
plexitiesofconstantlyevolvingoptionsofmeetingonline(EllisonandBoyd2013)
andtherelationshipsformedwithintheirbounds.Futurewavesofthedatausedin
thisanalysisneverthelesspromisetoenableamoredetailedlookatspecicmodesof
onlinesocialinteraction.Finally,eventhoughtheanalysisaccountedforanextended
setofpotentiallyconfoundingfactors,theassociationbetweeneducationandmar
riagemaystillbesubjecttoundetecteddifferencesandthusofanoncausalnature.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2002 G. Potarca
Despiteitslimitations,thisresearchconrmsthatinternetdatingfostersanuneven
distribution ofopportunitiesforpartneringandmarriage(Schmitz 2017),reafrm
ingthe importanceof socialcontextsofmeetingandmating(Blau1978;Blauand
Schwartz1997;Kalmijn andFlap2001)—particularlythecontinuouslyexpanding
digitalenvironment—forthesocialdemographyofunionformation.■
Acknowledgments This arti cle beneted from the sup port of an Ambizione grant from the Swiss National
Science Foundation (grant num ber: PZ00P1_174197). This research was also supported by the Swiss
National Centre of Competence in Research LIVES – Overcoming Vulnerability: Life Course Perspectives,
nanced by the Swiss National Science Foundation (grant num ber: 51NF40-160590). Earlier ver sions of
this arti cle were presented at the annual meet ing of the Amer i can Sociological Association (New York,
August 2019), the Euro pean Population Conference (Brussels, June 2018), and the ISA RC06-41 Conference
(Singapore, May 2018). This paper uses data from the Ger man Family Panel pairfam, coor di nated by Josef
Brüderl, Sonja Drobnič, Karsten Hank, Franz Neyer, and Sabine Walper. pairfam is funded as a long-term
pro ject by the Ger man Research Foundation (DFG). The author acknowl edges sup port from Josef Brüderl,
Nina Schumann, Rüdiger Lenke, and the entire pairfam team. This paper has grown thanks to every one
who pro vided feed back at var i ous stages of writ ing, includ ing anon y mous review ers, Clémentine Rossier,
Michael Rosenfeld, Wilfred Uunk, and Josh Vidich. My deepest grat i tude goes to Dorian Kessler for his
gen er os ity, con stant encour age ment, and always excel lent advice.
References
Alexandersson,A.(2015).Florida annual can cer report: 2015 inci dence and mor tal ity.Tallahassee:Flor
idaDepartmentofHealth.Retrievedfromhttps://fcds.med.miami.edu/inc/publications.shtml
Allison,P.D.(2014).Event his tory and sur vival anal y sis(2nded.).LosAngeles,CA:SAGEPublications,
Inc.
Axinn,W.G.,&Barber,J.S.(1997).Livingarrangementsandfamilyformationattitudesinearlyadult
hood. Journal of Marriage and the Family, 59, 595–611.
Baizán,P.,Aassve,A.,&Billari,F.C.(2003).Theinterrelationsbetweencohabitation,marriageandrst
birthinGermanyandSweden.Population and Environment, 25, 531–561.
Barg, K., & Beblo, M. (2012). Does “sorting into specialization” explain the differences in time use
betweenmarriedandcohabitingcouples?AnempiricalapplicationforGermany.Annals of Economics
and Statistics, 2012(105/106),127–152.
Becker,G.S.(1993).A trea tise on the fam ily: Enlarged edi tion(2nded.).Cambridge,MA:HarvardUni
versityPress.
Bellani,D.,EspingAndersen,G.,& Nedoluzhko,L.(2017). Neverpartnered:A multilevel analysisof
life long singlehood. Demographic Research, 37, 53–100. https://doi.org/10.4054/DemRes.2017.37.4
Bellani,D.,EspingAndersen,G.,&Pessin,L.(2017).Whenequitymattersformaritalstability:Compar
ingGermanandU.S.couples.Journal of Social and Personal Relationships, 35, 1273–1298.
Bellou,A.(2014).Theimpactofinternetdiffusiononmarriagerates:Evidencefromthebroadbandmar
ket. Journal of Population Economics, 28, 265–297.
Bergdall,A.R.,Kraft,J.M.,Andes,K.,Carter,M.,HateldTimajchy,K.,&HockLong,L.(2012).Love
andhookingupinthe newmillennium:Communication technologyandrelationshipsamongurban
AfricanAmericanandPuertoRicanyoungadults.Journal of Sex Research, 49, 570–582.
Bergström,M.(2011).Casualdatingonline: Sexualnormsandpractices onFrench heterosexualdating
sites. Zeitschrift für Familienforschung, 23, 319–336.
BitkomResearch.(2017).Jederzweiteonlinerglaubtandiegroßeliebeviainternet[Everysecondonline
person believes in the great love via the internet]. Retrieved from https://www.bitkom.org/Presse/
Presseinformation/JederzweiteOnlinerglaubtandiegrosseLiebeviaInternet.html
Blackhart,G. C.,Fitzpatrick,J.,&Williamson,J.(2014).Dispositionalfactorspredictinguseofonline
dat ing sites and behav iors related to online dat ing. Computers in Human Behavior, 33, 113–118.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2003Online Dating and Educational Inequalities in Marriage Formation
Blau,P.M.(1978).Amer i can occu pa tional struc ture(1sted.).NewYork,NY:FreePress.
Blau,P.M.,Beeker,C.,& Fitzpatrick,K.M. (1984).Intersectingsocial afliationsandintermarriage.
Social Forces, 62, 585–606.
Blau,P.M.,&Schwartz,J.(1997).Crosscutting social cir cles: Testing a mac ro struc tural the ory of inter-
group rela tions.NewBrunswick,NJ:Routledge.
Blossfeld,H.P.(2009).Educationalassortativemarriagein comparativeperspective.Annual Review of
Sociology, 35, 513–530.
Blossfeld,H.P.,&Jaenichen,U.(1992).Educationalexpansionandchangesinwomen’sentryintomar
riageandmotherhoodintheFederalRepublicofGermany.Journal of Marriage and the Family, 54,
302–315.
BouchetValat,M.,&Dutreuilh,C.(2015).Fewersinglesamonghighlyeducatedwomen.Agenderrever
salofhypergamyacrosscohortsinFrance.Population, 70, 665–688.
Bourdieu,P.(1989).Socialspaceandsymbolicpower.Sociological Theory, 7, 14–25.
Bourdieu,P.(1998).Practical rea son: On the the ory of action(1sted.)(R.Johnson,Trans.).Stanford,CA:
StanfordUniversityPress.
Bourdieu,P.(2008).The bach e lors’ ball: The cri sis of peas ant soci ety in Béarn(R.Nice,Trans.).Chicago,
IL:UniversityofChicagoPress.
Bourdieu,P.(2013).Distinction: A social cri tique of the judge ment of taste.Oxford,UK:Routledge.
Brüderl,J.,&Diekmann,A.(1994).Bildung,geburtskohorteundheiratsalter:Einevergleichendeuntersu
chungdesheiratsverhaltensinWestdeutschland,OstdeutschlandunddenVereinigtenStaaten[Educa
tion,birthcohort,andmarriageage:AcomparativestudyofmarriagebehaviorinWestGermany,East
Germany,andtheUnitedStates].Zeitschrift für Soziologie, 23, 56–73.
Brüderl,J.,Drobnič,S.,Hank, K.,Neyer,F.J.,Walper,S.,Alt,P.,...Wilhelm,B.(2020). The Ger man
Family Panel (pairfam), ZA5678 data le ver sion 11.0.0[Dataset].Cologne,Germany:GESISData
Archive.https://doi.org/10.4232/pairfam.5678.11.0.0
Brüderl,J.,Schmiedeberg,C.,Castiglioni,L.,Becker,O.A.,Buhr,P.,Fuß,D.,...Schumann,N.(2019).
The Ger man Family Panel: Study design and cumu lated eld report (Waves 1 to 10)(Release10.0).
Retrieved from https://www.pairfam.de/leadmin/user_upload/uploads/Neu_10/TP01%20Cumulat
ed%20Field%20Report%2C%20pairfam%202019.pdf
Buschner,A.,Adam,U.,&Schulz,F.(2018).Relativeeducation,parenthood,andcouples’divisionofpaid
work.EvidencefromGermancensusdata.Journal of Family Research, 30, 96–119.
Carbone,J.,&Cahn,N.(2014).Marriage mar kets: How inequal ity is remak ing the Amer i can fam ily(1st
ed.).Oxford,UK:OxfordUniversityPress.
Cesare,N.,Lee,H.,McCormick,T.,Spiro,E.,&Zagheni,E.(2018).Promisesandpitfallsofusingdigital
traces for demo graphic research. Demography, 55, 1979–1999.
Cherlin,A.J.(2020).Degreesofchange:Anassessmentofthedeinstitutionalizationofmarriagethesis.
Journal of Marriage and Family, 82, 62–80.
Chesley,N.(2005).Blurringboundaries?Linkingtechnologyuse,spillover,individualdistress,andfamily
sat is fac tion. Journal of Marriage and Family, 67, 1237–1248.
Chesley,N.(2006).Familiesinahightechagetechnologyusagepatterns,workandfamilycorrelates,and
gen der. Journal of Family Issues, 27, 587–608.
Chi,P.,Wu,Q.,Cao,H.,Zhou,N.,& Lin,X. (2020).Relationshiporientedvalues andmaritalandlife
satisfactionamongChinesecouples.Journal of Social and Personal Relationships, 37, 2578–2596.
Cleves,M.,Gutierrez,R.,Gould,W.,&Marchenko,Y.(2011).An intro duc tion to sur vival anal y sis using
Stata(3rded.).CollegeStation,TX:StataCorpLP.
Croft,A.,Schmader,T.,Beall,A.,&Schaller,M.(2020).Breadwinnerseeksbottlewarmer:Howwomen’s
futureaspirationsandexpectationspredicttheircurrentmatepreferences.Sex Roles, 82, 633–643.
Davis,S. N.,&Greenstein,T.N.(2009).Genderideology:Components,predictors,andconsequences.
Annual Review of Sociology, 35, 87–105.
Dieckhoff,M.(2008).Skillsandoccupationalattainment:AcomparativestudyofGermany,Denmarkand
theUK.Work, Employment and Society, 22, 89–108.
Drobnic,S.,Blossfeld, H.P.,& Rohwer,G.(1999). Dynamics ofwomen’s employment patterns over
thefamilylifecourse:AcomparisonoftheUnitedStatesandGermany.Journal of Marriage and the
Family, 61, 133–146.
Dröge,K.,&Voirol,O.(2011).Onlinedating:Thetensionsbetweenromanticloveandeconomicrational
ization.Zeitschrift für Familienforschung, 23, 337–357.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2004 G. Potarca
Eckhard,J.,&Stauder,J.(2019).Partnermarketopportunitiesandunionformationoverthelifecourse—A
com par i son of dif fer ent mea sures. Population, Space and Place, 25, e2178. https://doi.org/10.1002/
psp .2178
Ellison,N.B.,&Boyd,D.M.(2013).Socialitythroughsocialnetworksites.InW.H.Dutton(Ed.),Oxford
hand book of internet stud ies(pp.151–172).Oxford,UK:OxfordUniversityPress.https://doi.org/10
.1093/oxfordhb/9780199589074.013.0008
England,P.,Levine,A., &Mishel, E.(2020). Progresstoward genderequalityintheUnitedStateshas
slowed or stalled. Proceedings of the National Academy of Sciences, 117, 6990–6997.
Feld,S.L.(1984).Thestructureduseofpersonalassociates.Social Forces, 62, 640–652.
Feliciano,C.,Robnett,B.,&Komaie,G.(2009).GenderedracialexclusionamongWhiteinternetdaters.
Social Science Research, 38, 39–54.
Fine,J.P.,&Gray,R.J.(1999).Aproportionalhazardsmodelforthesubdistributionofacompetingrisk.
Journal of the Amer i can Statistical Association, 94, 496–509.
Finkel,E.J.,Eastwick,P.W.,Karney,B.R.,Reis,H.T.,&Sprecher,S.(2012).Onlinedating:Acritical
analysisfromtheperspectiveofpsychologicalscience.Psychological Science in the Public Interest,
13, 3–66.
Geser,H.(2007).Online search for ofine part ners: Matching plat forms as tools of indi vid ual empow er
ment and social retraditionalization(Workingpaper).Zürich,Switzerland:UniversitätZürich,Philos
ophischeFakultät,SoziologischesInstitut.Retrievedfromhttps://nbnresolving.org/urn:nbn:de:0168
ssoar323827
Grow,A.,Schnor,C., &VanBavel,J.(2017).The reversalofthegendergap ineducationand relative
divorcerisks:Amatterofalternativesinpartnerchoice?Population Studies, 71(Supp1.),15–34.
Gudelunas,D.(2012).There’sanappforthat:Theusesandgraticationsofonlinesocialnetworksforgay
men. Sexuality & Culture, 16, 347–365.
Gui,T. (2020).“Leftoverwomen” or single by choice:Genderrole negotiation of single professional
womenincontemporaryChina.Journal of Family Issues, 41, 1956–1978.
Hakim, C. (2003).A new approach to explaining fertility patterns: Preference theory. Population and
Development Review, 29, 349–374.
Hardey,M.(2002).Lifebeyondthescreen:Embodimentandidentitythroughtheinternet.Sociological
Review, 50, 570–585.
Helsper,E.J.,&vanDeursen,A.J.A.M.(2015).Thethirdleveldigitaldivide:Whobenetsmostfrom
beingonline?InL.Robinson,S.R.Cotton,J.Schulz,T.M.Hale,&A.Williams(Eds.),Communication
and infor ma tion tech nol o gies annual(Vol.10,pp.29–52).Bingley,UK:EmeraldGroupPublishing
Limited.https://doi.org/10.1108/S2050206020150000010002
Hiekel,N.,Liefbroer,A. C.,&Poortman,A.R.(2015). Marriageand separationrisksamong German
cohabiters:Differencesbetweentypesofcohabiter.Population Studies, 69, 237–251.
Hitsch,G.J.,Hortaçsu,A.,&Ariely,D.(2010).Matchingandsortinginonlinedating.Amer i can Economic
Review, 100, 130–163.
Houts,R.M.,Robins,E.,&Huston,T.L.(1996).Compatibilityandthedevelopmentofpremaritalrela
tion ships. Journal of Marriage and the Family, 58, 7–20.
Hudde,A.(2020). Homogamyin genderroleattitudesamongyoungcouples:EvidencefromGermany.
KZfSS Kölner Zeitschrift für Soziologie und Sozialpsychologie, 72, 403–428.
Hudde,A.,&Engelhardt,H.(2020).Intracouple(dis)similarityingenderroleattitudesandthetransition
toparenthoodinGermany.Euro pean Sociological Review, 36, 852–867.
Huinink,J.,Brüderl,J.,Nauck,B.,Walper,S.,Castiglioni,L.,&Feldhaus,M.(2011).PanelAnalysisof
IntimateRelationshipsandFamilyDynamics(pairfam):Conceptualframeworkanddesign.Zeitschrift
für Familienforschung, 23, 77–101.
Jalovaara,M.(2012).SocioeconomicresourcesandrstunionformationinFinland,cohortsborn1969–
81. Population Studies, 66, 69–85.
Kalmijn, M. (2013). The educational gradient in marriage: A comparison of 25 European countries.
Demography, 50, 1499–1520.
Kalmijn,M.,&Flap,H.(2001).Assortativemeetingandmating:Unintendedconsequencesoforganized
set tings for part ner choices. Social Forces, 79, 1289–1312.
Kelly,E.L.,&Conley,J.J.(1987).Personalityandcompatibility:Aprospectiveanalysisofmaritalstabil
ityandmaritalsatisfaction.Journal of Personality and Social Psychology, 52, 27–40.
Kerckhoff,A.C.,&Davis,K.E.(1962).Valueconsensusandneedcomplementarityinmateselection.
Amer i can Sociological Review, 27, 295–303.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2005Online Dating and Educational Inequalities in Marriage Formation
Kim, M., Kwon, K.N., & Lee, M. (2009). Psychological characteristics of internet dating service
users:Theeffectofselfesteem, involvement,and sociability ontheuseofinternet datingservices.
CyberPsychology & Behavior, 12, 445–449.
Klärner,A.,&Knabe,A.(2017).Onthenormativefoundationsofmarriageandcohabitation:Resultsfrom
groupdiscussionsinEasternandWesternGermany.Demographic Research, 36, 1637–1666. https://
doi.org/10.4054/DemRes.2017.36.53
Klein,M.(2015).TheincreasingunemploymentgapbetweenthelowandhigheducatedinWestGermany.
Structuralorcyclicalcrowdingout?Social Science Research, 50, 110–125.
Klesment,M.,& Van Bavel, J.(2017).The reversal ofthegendergapin education,motherhood, and
womenasmainearnersinEurope.Euro pean Sociological Review, 33, 465–481.
Kreyenfeld,M.,Huinink,J.,Trappe,H.,&Walke,R.(2012).DemoDiff:Adatasetforthestudyoffamily
changeinEastern(andWestern)Germany.Schmollers Jahrbuch, 132, 653–660.
Kulik,L.(2016).Theimpactofsocialbackgroundongenderroleideology:Parents’versuschildren’satti
tudes. Journal of Family Issues, 23, 53–73.
Li,N.P.,Yong,J.C.,Tov,W.,Sng,O.,Fletcher,G.J.O.,Valentine,K.A.,...Balliet,D.(2013).Matepref
erencesdopredictattractionandchoicesintheearlystagesofmateselection.Journal of Personality
and Social Psychology, 105, 757–776.
Lipps,O.,&Oesch,D.(2018).Theworkingclassleftbehind?TheclassgapinlifesatisfactioninGermany
andSwitzerlandoverthelastdecades.Euro pean Societies, 20, 549–571.
Lück,D.,&Ruckdeschel,K.(2018).Clearinitscore,blurredintheoutercontours:Culturallynormative
conceptionsofthefamilyinGermany.Euro pean Societies, 20, 715–742.
Marcińczak,S.,Musterd,S.,vanHam,M.,&Tammaru,T.(2016).Inequality and ris ing lev els of socio-
eco nomic seg re ga tion: Lessons from a pan-Euro pean com par a tive study. In T. Tammaru,S. Mar
cińczak,M.VanHam,&S.Musterd(Eds.), Socio-eco nomic seg re ga tion in Euro pean cap i tal cit ies:
East meets West( pp.358–382).Abingdon,UK:Routledge.
Marsden,P.V.(1990).Networkdataandmeasurement.Annual Review of Sociology, 16, 435–463.
McClendon,D.(2018).Crossingboundaries:“Somecollege,”schools,andeducationalassortativemating.
Journal of Marriage and Family, 80, 812–825.
McClendon,D.,Kuo,J.C.L.,&Raley,R.K.(2014).Opportunitiestomeet:Occupationaleducationand
marriageformationinyoungadulthood.Demography, 51, 1319–1344.
McKenna,K.Y.A., Green,A. S.,&Gleason, M.E.J. (2002).Relationship formationon theinternet:
What’sthebigattraction?Journal of Social Issues, 58, 9–31.
Meeussen,L.,VanLaar,C.,&Verbruggen,M.(2019).Lookingforafamilyman?Normsformenaretop
plinginheterosexualrelationships.Sex Roles, 80, 429–442.
Mulder,C.H.,Clark,W.A.V.,&Wagner,M.(2006).Resources,livingarrangementsandrstunionfor
mationintheUnited States,theNetherlandsandWestGermany.Euro pean Journal of Population /
Revue Européenne de Démographie, 22, 3–35.
Müller, N., & Dräger,J. (2019). Economic roles and marriage timing: Acohort comparison between
womenandmeninEastandWestGermany.Longitudinal and Life Course Studies, 10, 347–374.
Myers, S. M., & Booth, A. (2002). Forerunners of change in nontraditional gender ideology. Social
Psychology Quarterly, 65, 18–37.
Nitsche, N., & Grunow, D. (2016). Housework over the course of relationships: Gender ideology,
resources, and the divi sion of house work from a growth curve per spec tive. Advances in Life Course
Research, 29, 80–94.
Nitsche,N.,&Grunow,D.(2018).Doeconomicresourcesplayaroleinbargainingchildcareincouples?
ParentalinvestmentincasesofmatchingandmismatchinggenderideologiesinGermany.Euro pean
Societies, 20, 785–815.
Ogolsky,B.G.,Dennison,R.P.,&Monk,J.K.(2014).Theroleofcouplediscrepanciesincognitiveand
behavioralegalitarianisminmaritalquality.Sex Roles, 70, 329–342.
OllierMalaterre,A.,Jacobs,J.A.,&Rothbard,N.P.(2019).Technology,work,andfamily:Digitalcul
turalcapitalandboundarymanagement.Annual Review of Sociology, 45, 425–447.
Oppenheimer,V.K.(1988).Atheoryofmarriagetiming.Amer i can Journal of Sociology, 94, 563–591.
Oppenheimer,V.K.(1994).Women’srisingemploymentandthefutureofthefamilyinindustrialsocie
ties. Population and Development Review, 20, 293–342.
Orosz,G.,Benyó, M.,Berkes,B., Nikoletti,E., Gál,É.,TóthKirály,I., & Bőthe,B. (2018).Theper
sonality,motivational,andneedbasedbackgroundofproblematicTinderuse.Journal of Behavioral
Addictions, 7, 301–316.
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2006 G. Potarca
Parker,K.,&Stepler,R.(2017).As U.S. mar riage rate hov ers at 50%, edu ca tion gap in mar i tal sta tus
wid ens (FactTank:NewsintheNumbersreport).Washington,DC:PewResearchCenter.Retrieved
from http://www.pewresearch.org/facttank/2017/09/14/asusmarriageratehoversat50education
gapinmaritalstatuswidens/
Potarca,G.(2017). Doestheinternet affectassortative mating? Evidencefromthe U.S.and Germany.
Social Science Research, 61, 278–297.
Press,J.E. (2004). Cutebuttsand housework: Agynocentric theory of assortativemating.Journal of
Marriage and Family, 66, 1029–1033.
Psacharopoulos,G.,&Patrinos,H.A.(2018).Returnstoinvestmentineducation:Adecennialreviewof
the global lit er a ture. Education Economics, 26, 445–458.
Qian,Y.(2018).Educationalassortative matingand incomedynamics incouples:Alongitudinaldyadic
per spec tive. Journal of Marriage and Family, 80, 607–621.
Rammstedt,B.,& John, O.P.(2005).Kurzversion des BigFiveInventory (BFIK): Entwicklungund
validierungeines ökonomischeninventarszurerfassungderfünffaktoren derpersönlichkeit [Short
versionoftheBigFiveInventory(BFIK):Developmentandvalidationofaneconomicinventoryfor
assessmentofthevefactorsofpersonality].Diagnostica, 51, 195–206.
Rindfuss, R. R. (1991). The young adult years: Diversity,structural change, and fertility—Population
AssociationofAmerica1991presidentialaddress.Demography, 28, 493–512.
Rosenfeld,M.J.(2017).Marriage,choice,andcouplehoodintheageoftheinternet.Sociological Science,
4, 490–510.
Rosenfeld,M.J.(2018).AreTinderand datingapps changingdatingandmatingintheUSA?InJ.Van
Hook,S.M.McHale,&V.King(Eds.),Families and tech nol ogy(pp.103–117).Cham,Switzerland:
SpringerInternationalPublishing.https://doi.org/10.1007/9783319955407_6
Rosenfeld,M.J.,&Thomas,R.J.(2012).Searchingforamate:Theriseoftheinternetasasocialinter
mediary.Amer i can Sociological Review, 77, 523–547.
Rosenfeld,M. J.,Thomas,R.J.,& Hausen,S.(2019).Disintermediatingyour friends:Howonlinedat
ingin theUnitedStatesdisplacesotherwaysofmeeting.Proceedings of the National Academy of
Sciences, 116, 17753–17758.
Sanderson,C.A.,Keiter,E.J.,Miles,M.G.,&Yopyk,D.J.A.(2007).Theassociationbetweenintimacy
goals and plans for ini ti at ing dat ing rela tion ships. Personal Relationships, 14, 225–243.
Sassler,S.,&Schoen,R.(1999).Theeffectof attitudesandeconomic activityonmarriage. Journal of
Marriage and the Family, 61, 147–159.
Schmitz,A.(2017).The struc ture of dig i tal part ner choice: A Bourdieusian per spec tive.Cham,Switzer
land:SpringerInternationalPublishing.
Schneider, D., & Hastings, O. P. (2017). Income inequality and household labor. Social Forces, 96,
481–506.
Schulz,F.,Skopek,J.,Klein,D.,&Schmitz,A.(2008).WernutztinternetkontaktbörseninDeutschland?
[WhousesinternetcontactexchangesinGermany?].Zeitschrift für Familienforschung, 20, 271–292.
Schwartz,B.(2005).The par a dox of choice: Why more is less.NewYork,NY:HarperPerennial.
Schwartz,C.R.,&Han,H.(2014).Thereversalofthegendergapineducationandtrendsinmaritaldis
so lu tion. Amer i can Sociological Review, 79, 605–629.
Skopek,J.,Schulz,F.,&Blossfeld,H.P.(2011).Whocontactswhom?Educationalhomophilyinonline
mate selec tion. Euro pean Sociological Review, 27, 180–195.
Smith,A.(2016).15% of Amer i can adults have used online dat ing sites or mobile dat ing apps(Internet&
Technologyreport).Washington,DC:PewResearchCenter.Retrievedfromhttp://www.pewinternet
.org/2016/02/11/15percentofAmericanadultshaveusedonlinedatingsitesormobiledatingapps/
Sobotka,T.(2011).Fertilityin Austria,GermanyandSwitzerland:Isthereacommonpattern?Comparative
Population Studies, 36, 263–304.
Spannagel, D. (2016). Soziale mobilität nimmt weiter ab [Social mobility continues to decline]. WSI
Mitteilungen, 2016, 613–620.
Sprecher,S.,Econie,A.,&Treger,S.(2019).Matepreferencesinemergingadulthoodandbeyond:Age
var i a tions in mate pref er ences and beliefs about change in mate pref er ences. Journal of Social and
Personal Relationships, 36, 3139–3158.
Suhr,F.(2020,August24).Infograk: Datingbörsengenerierenimmer mehrnutzer[Infographic: Dat
ing sites gen er ate more and more users]. Statista Infograken.Retrievedfromhttps://de.statista.com/
infograk/22641/prognosezuranzahldernutzervondatingportalen/
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021

2007Online Dating and Educational Inequalities in Marriage Formation
Sumter,S.R.,Vandenbosch,L.,&Ligtenberg,L.(2017).LovemeTinder:Untanglingemergingadults’
moti va tions for using the dat ing appli ca tion Tinder. Telematics and Informatics, 34, 67–78.
Thomae,M.,&Houston,D.M.(2016).Theimpactofgenderideologiesonmen’sandwomen’sdesirefor
atraditionalornontraditionalpartner.Personality and Individual Differences, 95, 152–158.
Thomas,R.J.(2020).Onlineexogamyreconsidered:Estimatingtheinternet’seffectsonracial,educa
tional, reli gious, polit i cal and age assortative mat ing. Social Forces, 98, 1257–1286.
Timmermans,E.,&DeCaluwé,E.(2017).ToTinderornottoTinder,that’sthequestion:Anindividual
dif fer ences per spec tive to Tinder use and motives. Personality and Individual Differences, 110, 74–79.
Valkenburg,P.M.,&Peter,J.(2007).Whovisitsonline datingsites? Exploringsomecharacteristicsof
online dat ers. CyberPsychology & Behavior, 10, 849–852.
VanBavel,J.,&Klesment,M.(2017).Educationalpairings,motherhood,andwomen’srelativeearnings
inEurope.Demography, 54, 2331–2349.
VanBavel,J.,Schwartz,C.R.,&Esteve,A.(2018).Thereversalofthegendergapineducationandits
consequencesforfamilylife.Annual Review of Sociology, 44, 341–360.
Waller,M.R.,&McLanahan,S.S.(2005).“His”and“her”marriageexpectations:Determinantsandcon
se quences. Journal of Marriage and Family, 67, 53–67.
Whitty,M.T.,&Carr,A.N.(2006).Cyberspace romance: The psy chol ogy of online rela tion ships.New
York,NY:PalgraveMacmillan.
Whitty,M.T.,&Young,G.(2016).Cyberpsychology: The study of indi vid u als, soci ety and dig i tal tech-
nol o gies.Hoboken,NJ:JohnWiley&Sons.
Yu,W.,&Kuo,J.C.L.(2016).Explainingtheeffectofparentchildcoresidenceonmarriageformation:
ThecaseofJapan.Demography, 53, 1283–1318.
Zagel, H., & Breen, R. (2018). Family demography and income inequality in WestGermany and the
UnitedStates:Acta Sociologica, 62, 174–192.
Zentner,M.,&Eagly,A.H.(2015).Asocioculturalframeworkforunderstandingpartnerpreferencesof
womenandmen:Integrationofconceptsandevidence.Euro pean Review of Social Psychology, 26,
328–373.
GinaPotarca
gina .potarca@unige .ch
NCCRLIVES/InstituteofDemographyandSocioeconomics,UniversityofGeneva,Geneva,Switzerland;
https://orcid.org/0000000212863781
Downloaded from http://read.dukeupress.edu/demography/article-pdf/58/5/1977/1204808/1977potarca.pdf by UNIVERSITE DE GENEVE user on 02 November 2021