Ambivalence in Later-Life Family Networks: Beyond Intergenerational Dyads

My iam Gi a din Uni e si y o Gene a E ic D. Widme Uni e si y o Gene a∗ Ing id A ne Connidis Uni e si y o Wes e n On a io∗∗ Anna-Maija Cas én Uni e si y o Eas e n Finland∗∗∗ Ri a Gou eia Uni e si y o Lisbon∗∗∗∗ Ba ba a Maso i Uni e si y o Applied Sciences and A s o Sou he n Swi ze land∗∗∗∗∗ Ambi alence in La e -Li e Family Ne wo ks: Beyond In e gene a ional Dyads NCCR LIVES, Cen e o he In e disciplina y S udy o Ge on ology and Vulne abili y, Uni e si y o Gene a, 54, e des Acacias, 1227 Ca ouge, Gene a, Swi ze land. ∗NCCR LIVES, Depa men o Sociology, Uni e si y o Gene a, Uni Mail, 40 bd du Pon -d’A e, CH-1211 Gene a 4, Swi ze land ([email p o ec ed]). ∗∗ Depa men o Sociology, Uni e si y o Wes e n On a io, London, On a io N6A5C2, Canada. ∗∗∗ Depa men o Social Sciences, Uni e si y o Eas e n Finland, PO Box 1627, Kuopio, 70211, Finland. ∗∗∗∗ Ins i u e o Social Sciences, Uni e si y o Lisbon, A enida P o esso Aníbal Be encou n∘9, Lisboa 1600-189, Po ugal. ∗∗∗∗∗ NCCR LIVES, Cen e o Compe ence on Aging, Uni e si y o Applied Sciences and A s o Sou he n Swi ze land, S abile Piazze a, Via Violino 11, 6928 Manno, Swi ze land. © 2018 The Au ho s. Jou nal o Ma iage and Family pub- lished by Wiley Pe iodicals, Inc. on behal o Na ional Coun- cil on Family Rela ions. This is an open access a icle unde he e ms o he C e- a i e Commons A ibu ion-NonComme cial License, which pe mi s use, dis ibu ion and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed and is no used o comme cial pu poses. Key Wo ds: ambi alence, con lic , emo ional suppo , amily ne wo ks, inequali ies, olde adul s. In la e li e, changing condi ions ela ed o heal h, pa ne ship, and economic s a us may igge no only suppo bu also con lic and ambi alence, wi h he consequen enego ia- ion o amily ies. The aim o his s udy is o in es iga e bo h con lic and emo ional sup- po in he amily ne wo ks o olde adul s, aking he esea ch beyond he le el o in e - gene a ional dyads. We used a subsample o 563 elde s (aged 65 yea s and olde ) om he Swiss Vi e/Leben/Vi e e su ey. Mul iple co - espondence analysis and in-dep h case s udies we e used o iden i y he key social condi ions ha ela e o he p e alence o con lic ed and suppo i e dyads in amily ne wo ks. Findings showed ha he balance o con lic and emo- ional suppo in olde adul s’ amily ne wo ks a ied acco ding o he composi ion o hei am- ily ne wo k as well as hei age, heal h, income, and gende . Heigh ened a en ion o he concep o ambi a- lence since he u n o he cen u y challenged he iew o amily ies as exclusi ely suppo i e o posi i e and highligh ed con adic ions in he ies be ween adul child en and hei aging pa en s (Connidis, 2012, 2015; Connidis & 768 Jou nal o Ma iage and Family 80 (June 2018): 768–784 DOI:10.1111/jom .12469 Ambi alence in La e -Li e Family Ne wo ks 769 McMullin, 2002; Lüsche , 2002, 2005; Lüsche & Ho , 2013; Lüsche & Pilleme , 1998). The ambi alence concep emphasizes he coex- is ence o con lic and suppo as inhe en pa s o amily dynamics (Connidis, 2012, 2015; Connidis & McMullin, 2002; Lüsche , 2002, 2005; Lüsche & Ho , 2013; Lüsche & Pilleme , 1998; Willson, Shuey, Elde , & Wick- ama, 2006). To da e, he ambi alence esea ch has been mainly ci cumsc ibed o he dyadic le el and a ely conside s how ambi alence in in e gene a ional ela ionships is embedded in pa e ns o con lic and emo ional suppo in la ge amily ne wo ks. Using a ep esen a- i e sample o olde adul s li ing in Gene a, Swi ze land, his a icle explo es con lic and emo ional suppo in la e -li e amily ne wo ks, iden i ies ou pa e ns, and in es iga es how hose pa e ns a e embedded in he demo- g aphic, social, and economic condi ions ha a ec indi iduals and hei amily ies. I goes beyond in e gene a ional dyads by s udying ambi alence a he le el o amily ne wo ks and by making connec ions o b oade social o ces, pa icula ly social inequali y o s uc u ed social ela ions. The esul s a e discussed in ela ion o key issues in social ge on ology, including olde adul s’ socioemo ional selec i i y (Ca s ensen, 1992) and he unin ended consequences o amily suppo (Connidis & McMullin, 2002; Lüsche & Pilleme , 1998; Willson e al., 2006). Backg ound Ambi alence Beyond In e gene a ional Dyads Ambi alence is a mul ile el concep ha empha- sizes con adic ions a he le els o indi iduals and ela ionships; social ins i u ions, including amilies; and socie y, including he wel a e s a e and he s uc u ed social ela ions o inequali y based on gende , class, ace o e hnici y, age, and abili y (Connidis, 2015). Indi iduals s i e o nego ia e hese mul ile el con adic ions and he esul ing coexis ence o con lic and emo- ional suppo in hei a ious pe sonal ela ion- ships, including ies in he amily ealm. These nego ia ions occu a he meso le el o amilies, and hey need o be explo ed a his le el— ha is, beyond he le el o isola ed dyads. Typologies ha e p o en o be a use ul ool o explo ing ambi alence in in e gene a ional am- ily dyads as was shown by Eu opean esea ch (e.g., Fe ing, Michels, Boll, & Filipp, 2009; an Gaalen & Dyks a, 2006). Lüsche ’s (2002, 2005) heo e ical ypology o ambi alence is one o e y ew a emp s o cap u e he in e ac ion be ween suppo and con lic in in e gene a- ional ies; i includes he ollowing ou ypes o dis inc ela ional pa e ns: emancipa ion, solida i y, cap i a ion, and a omiza ion. These ypes iden i y he a ious ways in which amily membe s nego ia e ambi alence in in e gene a- ional ela ionships (Lüsche & Ho , 2013). In he emancipa ion ype, indi iduals accep con- lic along wi h coope a ion and suppo and ind new and mo e e ec i e ways o ela ing. In he solida i y ype, pa en s and child en emphasize oge he ness and suppo as ways o a oiding con lic and open ambi alence. Cap i a ed pa en –child ies a e s uck in con lic , en angled in an ongoing ba le o e ambi alence, whe eas a omized in e gene a ional dyads disengage o a oid con lic and ambi alence. Lüsche ’s ypology has con ibu ed o a la ge body o heo- e ical and empi ical wo k in social ge on ology (e.g., Connidis, 2015; Ka z & Lowens ein, 2010; Lang, 2004; Le iecq, Bailey, & Dahlen, 2008; Phillips, Ogg, & Ray, 2003). Howe e , ela ed empi ical wo k ocuses on in e gene a ional dyads. A key challenge in he s udy o ambi a- lence in social ge on ology is o explo e he ou pa e ns o Lüsche ’s (2002, 2005) he- o e ical ypology a he meso le el o amily ne wo ks. We he e o e aim o place dyads in he wide con igu a ion o amily ne wo ks and o hen ela e amily ne wo k ypes o he la ge social con ex in which hey a e embed- ded. Empi ically iden i ying a ious ypes o amily ne wo ks and hei a ailable esou ces may e eal a iabili y among amily ypes in he capaci y o nego ia e ambi alence in ways ha enhance emo ional suppo in olde adul s’ amilies. To in es iga e ambi alence a he meso le el o amilies, we anspose Lüsche ’s (2002, 2005) heo e ical ypology in o he amewo k o social ne wo k analysis, in which dyads a e conside ed as in e dependen pa s o a ne wo k whose ea u es accoun o much o wha hap- pens in any o hem (Wasse man & Faus , 1994). Such ools ha e been used in social ge on ology o explo e amily suppo s uc u es (Co nwell, 2009, 2011; Gi a din & Widme , 2015). We p opose using hese ools o assess he ela i e dominance o con lic ed and suppo i e dyads in he o e all dynamics o olde adul s’ amily ne wo ks (see Table 1). 770 Jou nal o Ma iage and Family Table 1. T ansposi ion o Lüsche ’s Typology In o he F amewo k o Social Ne wo k Analysis P opo ion o con lic ed dyads P opo ion o suppo i e dyads La ge Low La ge Emancipa ion Solida i y Low Cap i a ion A omiza ion Taking a social ne wo k analysis pe spec i e, he nego ia ion o ambi alence a he meso ( amily) le el esul s in ou pa e ns (Widme , 2016). Family ne wo ks in which emancipa ion p e ails display a la ge numbe o dyads ha a e cha ac e ized by bo h con lic and emo ional suppo . These dyads accep he coexis ence o emo ional suppo and con lic , and hey ind ways o ela e e ec i ely. In amily ne wo ks o he solida i y ype, a majo i y o he dyads a e suppo i e, and e y ew show con lic . Emo- ional suppo is ac i ely p omo ed, and con lic is a oided, in a majo i y o hese amily dyads. Cap i a ion occu s when many o he dyads in a amily ne wo k a e bu dened by con lic , wi h limi ed emo ional suppo ; his si ua ion is mo e likely o occu when amily membe s mus s ay oge he due o s ong amily obliga ions and o limi ed pe sonal esou ces. Finally, a omiza ion is cha ac e ized by he p esence o ew amily dyads ha a e ei he suppo i e o con lic ed. Emo ional suppo migh pe e ou as e du ing he long e m, and ensions a e esol ed h ough collec i e emo ional disengagemen , pu ing olde adul s a isk o loneliness. Iden i ying such ne wo k pa e ns based on Lüsche ’s (2002, 2005) ypology makes i possible o assess ambi alence in a a ie y o amily con ex s and i s connec ion o emo ional suppo . Composi ion o Family Ne wo ks and O he Li e Cou se Condi ions o Ambi alence Family ne wo ks a e di e se in hei compo- si ion. Some include only a spouse, child en, and g andchild en, bu o he s include a wide a ie y o ies such as siblings, dis an kin, in-laws, s ep- ela i es, and iends (Gi a din & Widme , 2015; T eas & Ma cum, 2011). Di - e en amily ies may ha e unequal likelihoods o being ambi alen . Siblings, in-laws, dis an kin, s ep- ela i es, and iends ypically ha e lowe social expec a ions o p o ide emo ional suppo and o s ay connec ed; hese mo e olun a y ela ionships a e ins ead based on a ini y and sha ed in e es s (Campbell, Con- nidis, & Da ies, 1999; Schne le & Wöhle , 2014). In cases in which such ies a e p e alen in a amily ne wo k, one may expec a la ge p opo ion o suppo i e dyads, whe eas he ense dyads a e disengaged— he solida i y ype. In con as , amily ne wo ks composed o child en and g andchild en a e cha ac e ized by s ong obliga ions o main ain close emo- ional connec ions om which amily membe s canno eadily escape (Finch & Mason, 1993). Such ies may, in u n, c ea e ensions. Thus, emancipa ion is a likely esul o such amily ne wo ks, which a e expec ed o include a la ge p opo ion o suppo i e dyads bu also some con lic ed and ambi alen dyads. O he condi ions associa ed wi h he li e cou se may al e he balance o con lic ed and emo ionally suppo i e dyads in amily ne wo ks. Ha ing inancial esou ces, being young-old, being in good heal h, and ha ing a pa ne con ibu e o main aining emo ional suppo h ough mu ual exchanges (Co nwell, 2009, 2011; O e , 2012). These condi ions a e expec ed o sus ain a la ge p opo ion o suppo i e dyads in he amily ne wo k and o hus p omo e he solida i y ype. Howe e , he a ailabili y o esou ces is also associa ed wi h ac o s ela ed o ime alloca ion and emo ional closeness among amily membe s (Taylo & No is, 2000). Thus, emancipa ion— he coex- is ence o con lic ed and suppo i e dyads in amily ne wo ks—may occu when esou ces a e a ailable; al hough esou ces con ibu e o suppo exchanges, hey may also aise ensions ela ed o ai ness and indi idual p e e ences in esou ce dis ibu ion. Al e na i ely, ha ing ewe inancial esou ces, being oldes -old, being in poo heal h, and ha ing los a pa ne may chal- lenge he balance o suppo i e dyads in amily ne wo ks in a o o con lic ed dyads. Olde pa - en s’ diminished esou ces inc ease hei need o emo ional suppo and, po en ially, heigh en he bu den placed on amily membe s—an ambi alen si ua ion ha has o be nego ia ed in amily ne wo ks (Connidis, 2012; Connidis & McMullin, 2002; Lüsche & Pilleme , 1998; Willson e al., 2006). Such ci cums ances may exace ba e ension among amily membe s, as hey ha e o adjus hei needs and expec a ions ega ding emo ional suppo (Connidis, 2003; Co nwell, 2009, 2011; Hillcoa -Nallé amby & Phillips, 2011). This may lead o cap i a ion, Ambi alence in La e -Li e Family Ne wo ks 771 wi h a la ge p opo ion o dyads wi hin he amily ne wo k ha a e cha ac e ized by con lic and limi ed emo ional-suppo , especially when esou ces (e.g., income and ime) a e sca ce and when he only po en ially suppo i e amily membe s a e child en (as in he case o widowed o di o ced pa en s). A omiza ion is expec ed o be mos likely when ew o no amily membe s a e p esen o when a se e e lack o esou ces makes emo ional-suppo exchanges di icul , leading o disengagemen (O e , 2012). No e ha such condi ions may p omo e gende di e - ences, as olde women a e mo e likely han olde men o be widowed, o ha e a lowe income, and o be in poo unc ional heal h because o hei gende ed posi ion in socie y and hei longe li e expec ancy (A be , Da idson, & Ginn, 2003; Moen, 1996; Willson e al., 2006). In sum, du ing la e li e, indi iduals may expe ience a a ie y o economic, amily, and heal h condi ions ha lead o dis inc and a - ious balances o con lic and emo ional sup- po in hei amily ne wo ks. This s udy aims o unde s and whe he and how he composi- ion o amily ne wo ks and li e cou se condi- ions a e associa ed wi h emancipa ion, solida - i y, cap i a ion, and a omiza ion as pa e ns ha cha ac e ize ambi alence in olde adul s’ am- ily ne wo ks. This issue is examined h ough he p e alence o con lic ed and suppo i e dyads in amily ne wo ks. We hen obse e whe he olde adul s’ amily-ne wo k composi ion, a ail- able esou ces, pa ne ship s a us, age, heal h, and social posi ion (as indica ed by gende , class, and ci izenship) accoun o he a ia ions in such pa e ns wi hin amily con igu a ions. To his end, we use mul iple co espondence anal- ysis (MCA) and in-dep h case s udies o explo e how he in e play o hese condi ions is ela ed o he balance o con lic and emo ional suppo in la e -li e amily ne wo ks. Me hod Da a and Sample The da a came om he Vi e/Leben/Vi e e s udy, which is a la ge, in e disciplina y su ey on he li e and heal h condi ions o people aged 65 yea s and olde ; his s udy was ca ied ou in 2011 and 2012 in i e o 26 can ons in Swi ze land (see O is e al., 2016). S a i ied by sex and age, he sample o 3,635 communi y-dwelling o ins i u ionalized pa icipan s was ep esen a i e o he gene al popula ion aged 65 yea s o olde . Da a we e collec ed using a sel -assessed ques ionnai e and an in-home, ace- o- ace in e iew wi h a s anda dized in e iew schedule. To illus a e ou indings wi h mo e in-dep h in o ma ion, we also d ew om a pool o ich li e-his o y da a based on li e calenda s ha he esponden s comple ed wi h he help o he in e iewe s. Gi en he p ac ical issues conce ning da a a ailabili y, ou analyses ocused on he Gene a subsample (n=704). We d opped 126 indi- iduals wi h cogni i e impai men s om he analysis because hey we e no able o answe he ques ionnai e on hei own ( esul ing in a subsample wi h n=578). An addi ional 15 we e d opped because hey did no answe he ques- ions abou hei amily ne wo ks. The e o e, he inal subsample included 563 esponden s. The mean age in he Gene a subsample was 78 yea s ( ange, 65–101): 40% we e aged 65 o 74, 35% 75 o 84, and 25% we e 85 and olde . O he esponden s, 49% we e women, and 66% we e na i e bo n; 61% had an a e age le el o educa ion (i.e., achie ed a high school o equi alen deg ee), and 23% had a high le el o educa ion (i.e., achie ed a leas a uni e si y deg ee). Rega ding hei las occupa ional s a us (p io o e i emen ), 30% we e uppe , 27% we e whi e colla , 14% we e in e media y, 14% we e blue colla , 9% we e sel -employed, and 6% we e inac i e. O he esponden s, 62% had an a e age o high income. As o hei pools o ela i es, 61% we e ma ied o had a pa ne (ei he cohabi a ing o li ing sepa a ely), 82% had a leas one li ing child, and 68% had a leas one li ing sibling. Rega ding sel - a ed heal h, 54% assessed hei heal h as good o e y good, 38% assessed i as ai , and 8% assessed i as bad o e y bad. A la ge majo i y o he esponden s (75%) we e in good unc ional heal h ( obus on he eigh ac i i ies o daily li ing [ADL] scale), 16% epo ed ha ing di icul ies in one o mo e o he eigh ADL ca ego ies, and 9% we e dependen acco ding o he ADL scale. Few we e ins i u ionalized (7%). Measu es Types o amily ne wo ks. D awing on s anda d name gene a o s o amily ne wo ks (Widme , Aeby, & Sapin, 2013), we asked esponden s o lis a maximum o i e indi iduals whom hey conside ed signi ican amily membe s a he ime o he in e iew. A e naming hese amily 772 Jou nal o Ma iage and Family membe s, he esponden s we e asked o indi- ca e he ype o ela ionship (e.g., pa ne , sis- e , o daugh e ) ha hey had wi h each o he ci ed pe sons. A o al o 14 amily e ms (com- monly called name in e p e e s) we e iden i ied, epo ed by a leas 5% o esponden s. To map he main ypes o amily ne wo ks, we applied s anda d ac o - and clus e -analy ic p ocedu es o he amily ne wo ks (Widme , 2016). The clus e ing app oach is commonly used in social ge on ology o iden i y social-ne wo k ypes, based on a ious kinds o social ela ionships (e.g., amily, iends, neighbo s, and commu- ni y g oups; Fio i, An onucci, & Co ina, 2006; Li win, 2001; Wenge , 1991). The iden i ica ion o amily ne wo ks h ough signi ican amily membe s is, howe e , a ely conside ed. To his end, we p oceeded using wo s eps. We i s an an explo a o y ac o analysis—using p incipal componen analysis wi h a imax o a ion—on he 14 amily e ms, plus a esidual ca ego y in o which he o he e ms we e ga he ed. Fol- lowing he s anda d p ac ice o ac o analysis (Tabachnick & Fidell, 1996), we e ained he six ac o s ha had eigen alues g ea e han 1; hese explained 55% o he a iance. We hen inpu hese six ac o s’ sco es in o a hie a - chical clus e ing analysis based on Euclidean dis ances and he Wa d clus e ing algo i hm (Leba , Mo ineau, & Pi on, 2002). We selec ed a solu ion wi h six clus e s, based on bo h in e p e abili y and clus e - alidi y measu es such as he Calinski-Ha abasz and silhoue e indexes (E e i , Landau, Leese, & S ahl, 2011). In he i s ype, which we named conjugal (39%), he esponden s cen e ed on hei child en and on hei cu en pa ne . The son amily ne wo k (8%) ocused on biolog- ical sons, hei pa ne s, and hei child en, whe eas he daugh e amily ne wo k (11%) la gely included biological daugh e s and hei child en. In sibling amily ne wo ks (15%), he esponden s mainly ci ed hei siblings as signi ican amily membe s. Those in kinship amily ne wo ks (8%) included a a ie y o el- a i es. Finally, hose in spa se amily ne wo ks (19%) ei he men ioned no signi ican amily membe o lis ed only a ew iends whom hey conside ed o be amily membe s. Pa e ns o con lic and emo ional suppo . A e lis ing he signi ican amily membe s, a se o ques ions was asked abou he emo ional suppo and con lic among he lis ed amily membe s (Widme e al., 2013). We measu ed he emo- ional suppo in he dyads using he ollowing ques ion: “Who would gi e emo ional suppo o X [i.e., o he esponden and hen o each o he indi idual in he esponden ’s amily ne wo k, conside ed one by one] du ing ou ine o mino oubles?” We mapped he con lic in he dyads wi h he ollowing ques ion: “Each amily has i s con lic s and ensions. In you opinion, who makes X [i.e., he esponden and hen each o he indi idual in he esponden ’s amily ne wo k, conside ed one by one] ang y?” Thus, each ocal pe son could men ion mo e han one membe in he amily ne wo k who emo ionally suppo s o annoys each o he o he membe s, including him- o he sel . In o he wo ds, he esponden s e alua ed no only hei own ela ionships wi h each amily membe bu also all he ela ionships among hei signi ican amily membe s. Nex , o es ima e emo ional suppo and con- lic in amily ne wo ks, we assessed he densi y o emo ional suppo and he densi y o con lic ac oss all o he amily dyads. These densi y measu es e e o he p opo ion o suppo i e and o con lic ed dyads among all he dyads in a pe sonal o amily ne wo k (Hanneman & Rid- dle, 2005). A high densi y o emo ional suppo indica es ha a majo i y o amily dyads could be easily ac i a ed o emo ional suppo , and a high densi y o con lic indica es a la ge p o- po ion o ense dyads. The a iable “pa e ns o con lic and emo ional suppo ” was ope a- ionalized by combining he wo densi y mea- su es in o a single ca ego ical a iable. Fo his pu pose, we i s dicho omized hese wo mea- su es: high e sus low densi y o emo ional sup- po and high e sus low densi y o con lic . Fo suppo i e ies, we chose he median sco e o .33 (33%) as he h eshold sepa a ing high and low densi y o emo ional suppo , as he aw sco es we e no no mally dis ibu ed. On a e - age, he esponden s epo ed a much lowe p o- po ion o ense dyads han o suppo i e dyads, so we se he h eshold o con lic a .10 (10%), which means ha he e was ension in one o e e y 10 possible dyads. Then we combined hese wo dicho omized densi y measu es in o a single a iable—pa e ns o con lic and emo- ional suppo —wi h he ollowing ou le els: emancipa ion (high densi ies o emo ional sup- po and con lic ), solida i y (high densi y o emo ional suppo and low densi y o con lic ), cap i a ion (low densi y o emo ional suppo and high densi y o con lic ), and a omiza ion Ambi alence in La e -Li e Family Ne wo ks 773 (low densi ies o emo ional suppo and con lic ). Family ne wo ks wi h he emancipa ion ype (23%) had high densi ies o bo h emo ional sup- po and con lic —on a e age, 57% o dyads we e pe cei ed as suppo i e, and 33% we e con lic o ien ed. Among amily ne wo ks show- ing solida i y (31%), he dyads we e, on a e - age, 58% suppo i e and only 2% ense. Family ne wo ks wi h he cap i a ion ype (9%) had a densi y o con lic (20%) ha was simila o he densi y o emo ional suppo (21%), which di - e s om he emancipa ion ype, in which sup- po i e dyads we e clea ly dominan . Fo amily ne wo ks wi h he a omiza ion ype (37%), he e was a e y low densi y o bo h emo ional sup- po and con lic , as only 11% o he dyads we e suppo i e and 2% we e con lic ed. Pool o a ailable ela i es. The esponden s we e asked o epo whe he hey had a pa - ne (ei he cohabi a ing o li ing sepa a ely) a he ime o he in e iew (0 =“no pa ne ,” 1=“ha e a pa ne ”), had a leas one li ing child (0 =“no child en,” 1 =“ha e child en”), and had a leas one li ing sibling (0 =“no b o h- e so sis e s,”1=“ha e b o he s o sis e s”). Age. We di ided ocal indi iduals by age in o one o he ollowing h ee g oups: aged 65 o 74 yea s (“young-old”), 75 o 84 yea s (“old-old”), and 85 yea s and olde (“oldes -old”; Suzman & Riley, 1985). Heal h. To measu e heal h, we ocused on unc- ional heal h a he han on mo e gene al heal h indica o s such as sel - a ed heal h because he loss o au onomy in old age ep esen s one o he mos challenging condi ions o bo h he olde adul s and hei amily membe s. Such a si ua ion wi hin a amily ne wo k may igge he p o ision o a ious o ms o suppo as well as ension. In addi ion, we pe o med he same analysis wi h sel - a ed heal h and ob ained simila esul s (no shown). To assess unc- ional heal h, we asked esponden s how much di icul y (0 =“no di icul y,” 1 =“able wi h di icul y,” 2 =“unable o pe o m”) hey had in pe o ming i e basic ac i i ies—washing, d essing and und essing, ea ing and cu ing ood, mo ing in and ou o bed, and mo ing a ound indoo s (Ka z, Fo d, Moskowi z, Jackson, & Ja - ee, 1963). We also asked hem abou hei di i- cul y in pe o ming he ollowing h ee mobili y ac ions: going up and down s ai s, mo ing a ound ou side, and walking a leas 200 me e s (Rosow & B eslau, 1966). These eigh i ems had a eliabili y le el (C onbach’s 𝛼)o .91.We c ea ed he ollowing h ee unc ional-heal h ca ego ies: ADL- obus (able o pe o m all eigh ac i i ies alone), ADL wi h di icul y (ha ing di icul y pe o ming one o mo e o he ac i i ies alone), and ADL-dependen (ha ing one o mo e ADL incapaci ies and needing someone’s help o pe o m hem). Social posi ion in socie y. We used income (1 =“low,” 2 =“a e age,” 3 =“high”), gende (0 =“ emale,” 1 =“male”), and ci izenship (wi h place o bi h as a p oxy; 0 =“na i e- bo n,” 1 =“ o eign-bo n”) o measu e social posi ion in Swiss socie y. We adjus ed incomes o household size, wi h a alue o 1 o he household head and 0.5 o each addi ional household membe (A kinson, Rainwa e , & Smeeding, 1995; Gab iel, O is, S ude , & Bae iswyl, 2015). Da a Analysis To assess he associa ions ha he ou pa e ns o con lic and emo ional suppo had wi h he ypes o amily ne wo ks and wi h he o he li e cou se condi ions (e.g., he pool o a ail- able ela i es, age, heal h, income, gende , and ci izenship), we compu ed an MCA—using he Fac oMineR package in R (Lê, Josse, & Hus- son, 2008). This enabled us o obse e how he in e play o hese di e en a iables was associa ed wi h emancipa ion, solida i y, cap i- a ion, and a omiza ion. MCA is a nonlinea mul i a ia e analysis me hod o ep esen ing he unde lying s uc u es in a se o obse a ions, as desc ibed by a se o ca ego ical a iables (Abdi & Valen in, 2007; A olio e al., 2013). MCA elies on an assump ion o in e depen- dence a he han on a causali y p inciple, so i is an ideal app oach when a ious ac o s in e - ac in a bidi ec ional o ci cula way. In sum, his explo a o y me hod p o ides a be e unde - s anding o how he esponse ca ego ies a e in e ela ed and, he e o e, enables he iden i i- ca ion o pa e ns. To his end, MCA ex ac s he main dimensions ha s uc u e he ela ionship be ween he esponse ca ego ies. In MCA, he i s wo o h ee ex ac ed dimensions explain as much a iance as possible, and hey a e usually su icien o syn hesize he mos impo an in o - ma ion con ained in he con ingency ables, all 774 Jou nal o Ma iage and Family Table 2. Disc imina ion Measu es o Va iables in Dimensions a e Ro a ion Va iables Dimension 1 loading Dimension 2 loading Ac i e Gende .25 .01 Income .33 .02 Ci izenship .01 .01 Ha ing a pa ne .49 .04 Ha ing a leas one li ing child .02 .56 Ha ing a leas one li ing sibling .13 .10 Family ne wo ks .25 .75 Pa e ns o con lic and emo ional suppo .12 .17 S a us o unc ional heal h .30 .00 Passi e Age – – % o explained ine ia 12.61% 8.40% No e. As a passi e a iable, age has no con ibu ed o he cons i u ion o axes. N=476. while emaining pa simonious (Desbois, 2008). To iden i y hese s uc u ing dimensions, one elies on disc imina ion measu es: he a ios o co ela ions be ween he a iables conside ed in MCA and he chosen dimensions. These mea- su es indica e he con ibu ion o each a iable in he de ini ion o each dimension; la ge disc im- ina ion measu es indica e s onge con ibu ions o he dimension’s de ini ion (see Table 2; A o- lio e al., 2013). All o he esponse ca ego ies can be plo - ed along he wo dimensions (axes in he plo ) o he MCA. The in e p e a ion o he esul s is based on he p oximi ies and he dis ances be ween he esponse ca ego ies in he plo ; hose ha a e close o one ano he p esen simi- la pa e ns o esponses, and hose ha a e dis- an ha e dissimila pa e ns (Abdi & Valen in, 2007; A olio e al., 2013). To be e iden i y which g oups o esponse ca ego ies we e mo e clea ly associa ed wi h one o he wo main dimensions, we o a ed hem, as he esponse ca ego ies a e be e aligned along he main dimensions a e o a ion (Sa acco, Cha en , & Kuen z, 2010). We inpu all a iables excep age as ac i e (di ec ly con ibu ing o he cons i u- ion o he MCA’s axes) and inpu age as a pas- si e a iable (G eenac e & Blasius, 2006)— ha is, we p ojec ed age a pos e io i in he al eady exis ing plo , bu i did no con ibu e o he axes’ de ini ions. This is because we unde s ood age as a p oxy o he aging p ocess, o he psychological de elopmen , and o a a ie y o social p ocesses s emming om changes in social oles, s a uses, heal h, gene a ional expe- iences and iden i y, and s ages in he li e cycle, a he han as a causal mechanism pe se (Se - e s en & Godlewski, 2016; Se e s en & Maye , 1997). We conduc ed all analyses in R (R De el- opmen Co e Team, 2011). Resul s Resul s o he Mul iple Co espondence Analysis Table 2 displays he disc imina ion measu es o he a iables ( he a ios o co ela ions) in he wo i s ex ac ed dimensions ha had eigen alues g ea e han 1, as well as each dimension’s pe cen ages o explained ine ia (i.e., a iance) a e o a ion. These wo cho- sen dimensions syn hesized he mos impo - an in o ma ion o all he ac i e a iables. As shown in Table 2, a iables measu ing pa ne - ship (.49), income (.33), unc ional heal h s a us (.30), and gende (.25) s ongly con ibu ed o he cons i u ion o Dimension 1. Types o am- ily ne wo ks (.75) and pa en hood (.56) s ongly con ibu ed o he cons i u ion o Dimension 2. In sum, he a ailabili y o a ious esou ces, ha ing a leas on li ing child, and he compo- si ion o amily ne wo ks we e he main con- ibu o s o he wo-dimensional MCA space. To es each a iable’s goodness o i , we compu ed con idence ellipses o each esponse ca ego y (Husson, Josse, Lê, & Maze , 2017). Con idence ellipses desc ibe he dis ibu ions o he cen e o g a i y o each esponse ca ego y and can hus be used o deduce a con idence egion o each ca ego y ia boo s apping. These analyses e ealed ha all he selec ed a iables’ esponse ca ego ies we e signi ican ly dis inc om each o he , as hei ellipses did no o e lap, wi h he excep ions o he daugh e and son am- ily ne wo ks—which we e loca ed in he same quad an (see Figu e 1)—and coun y o bi h. Figu e 1 shows ha he i s dimension (ho - izon al axis) disc imina ed among indi iduals based on he a ailabili y o esou ces such as income, pa ne ship, gende , age, unc ional heal h, and ha ing a leas one li ing sibling. P ecisely, he esponse ca ego ies ha indica ed an ad an ageous posi ion in socie y (high o Ambi alence in La e -Li e Family Ne wo ks 775 Figu e 1. P ojec ion o he Fi s Two Dimensions o he Mul iple Co espondence Analysis, A e Ro a ion. a e age income, male, na i e bo n), a la ge pool o ela i es (ha e a pa ne , ha e b o he s o sis e s, and ha e child en), young-old age, and good unc ional heal h (ADL- obus ) we e loca ed in he nega i e coo dina es o he ho i- zon al axis. On he o he hand, hose e lec ing a low social posi ion (low income, emale, and o eign bo n), a small pool o a ailable ela i es (no pa ne , no b o he s o sis e s, and no child en), oldes -old age, and poo unc ional heal h (ADL wi h di icul y o ADL-dependen ) we e posi ioned in he posi i e coo dina es o he ho izon al axis. The esou ces dimension cap u ed 13% o he o al a iance. Rega ding he second dimension ( e ical axis), Figu e 1 e eals he s ong disc imina- o y powe o pa en hood. Indeed, his axis di e en ia ed indi iduals who had child en (in he posi i e coo dina es o he e ical axis) om hose who did no (in he nega i e coo dina es o he e ical axis), and hei esul ing amily ne wo ks. Figu e 1 shows ha he no child en ca ego y and he amily ne wo ks ocused on amily membe s o he han child en—such as he sibling, kinship, and spa se amily ne wo ks—we e posi ioned on he nega i e side o he e ical axis, whe eas he ha ing child en ca ego y and he amily ne wo ks cen e ed p i- ma ily on li ing child en—such as he conjugal, daugh e , and son amily ne wo ks—we e p o- jec ed on he posi i e side o he e ical axis. The pa en al-s a us dimension explained 8% o he o al a iance. As Figu e 1 shows, he ou pa e ns o con lic and emo ional suppo we e di e en ly p ojec ed on he ou quad an s o he g aph. 776 Jou nal o Ma iage and Family Based on he coo dina es o he esponse ca - ego ies in he bidimensional space, he main p o iles associa ed wi h he ou pa e ns o con- lic and emo ional suppo became iden i iable. Emancipa ion was in he i s quad an . This pa e n was associa ed wi h conjugal amily ne wo ks; wi h being in he male, na i e-bo n, young-old, and ADL- obus ca ego ies; and wi h ha ing an a e age o high income. Cap i a ion was in he second quad an , in close connec ion wi h he inclusion o child en—bu wi h he no-pa ne ca ego y—and wi h daugh e and son amily ne wo ks, bo h ocused on chil- d en and g andchild en. In e ms o esou ces, hese esponden s had low social posi ions (low income, emale, and o eign bo n), we e in he oldes -old g oup and had poo unc ional heal h (ADL wi h di icul y o ADL-dependen ). Soli- da i y was posi ioned in he hi d quad an . This pa e n was connec ed wi h ha ing b o he s o sis e s, ha ing a pa ne , ha ing an a e age o high income, and being in he male, young-old, and ADL- obus g oups. Solida i y was also associa ed wi h a lack o child en, as i was posi ioned on he nega i e side o he e ical axis, in he same a ea as he sibling amily ne wo ks. Finally, a omiza ion was p ojec ed in he ou h quad an . This pa e n was associa ed wi h an absence o child en and wi h amily ne wo ks ha we e cen e ed p ima ily on kin, on a ew iends, o on no one, as in he spa se and kinship amily ne wo ks. A omiza ion was also ela ed o he lack o a ious esou ces, as i was connec ed wi h being in he oldes -old, ADL wi h di icul y o ADL-dependen , and emale g oups, and wi h ha ing no pa ne and a low income. Insigh s F om he Case S udies To illus a e he ways in which he pa e ns o con lic and emo ional suppo we e embedded in s uc u al condi ions, we p o ide ou case s udies ha we e selec ed on he basis o he quan i a i e esul s o bo h he ace- o- ace in e iews (wi h s anda dized in e iew sched- ules) and he li e-his o y da a. Emancipa ion in a conjugal amily ne wo k. The ocal indi idual, a man in his 90s, was li ing wi h his wi e in his own house a he ime o he in e iew. A e a ca ee as an in e na ional ci il se an , he was well o , in good unc ional heal h, and had wo sons, bo h o whom li ed nea by. His olde son was ma ied and had h ee child en, whe eas he younge one was single and childless. His amily belonged o he con- jugal ype, as he epo ed his wi e and wo sons as he only signi ican amily membe s. The sup- po i e dyads in his amily we e qui e dense and ecip ocal. As shown in Figu e 2a, his wi e was cen al in p o iding emo ional suppo , as she ga e suppo o all he amily membe s bu ecei ed suppo only om he ocal pe son ( he a ows poin o suppo p o ide s). The ocal indi idual exchanged emo ional suppo wi h his wi e and wi h he younge son. The olde son was less in eg a ed in he emo ional sup- po exchanges, as he p o ided no suppo and Figu e 2. Emancipa ion in a Conjugal Family Ne wo k. The dynamic o emo ional suppo ; densi y = .50; ecip oci y = .50 The dynamic o con lic ; densi y = .33; ecip oci y = .33 (a) (b) Ambi alence in La e -Li e Family Ne wo ks 783 Lang, F. R. (2004). The ilial ask in midli e: Ambi a- lence and he quali y o adul child en’s ela ion- ships wi h hei olde pa en s. In K. Pilleme & K. Lüsche (Eds.), In e gene a ional ambi alences: New pe spec i es on pa en –child ela ions in la e li e (pp. 183–206). Ox o d, England: Else ie . h ps://doi.o g/10.1016/S1530-3535(03)04008-1 Lê, S., Josse, J., & Husson, F. (2008). Fac oMineR: An R package o mul i a ia e analysis. Jou nal o S a is ical So wa e,25, 1–18. h ps://doi.o g/10 .18637/jss. 025.i01 Leba , L., Mo ineau, A., & Pi on, M. (2002). S a is- ique explo a oi e mul idimensionnelle [Mul idi- mensional explo a o y s a is ics] (3 d ed.). Pa is: Dunod. Le iecq, B. L., Bailey, S. J., & Dahlen, P. (2008). Ambi alence and coping among cus odial g and- pa en s. In B. Hayslip, J . & P. Kaminski (Eds.), Pa en ing he cus odial g andchild. Implica- ions o clinical p ac ice (pp. 3–16). New Yo k: Sp inge . Li win, H. (2001). Social ne wo k ype and mo ale in old age. The Ge on ologis ,41, 516–524. h ps:// doi.o g/10.1093/ge on /41.4.516 Lüsche , K. (2002). In e gene a ional ambi alence: Fu he s eps in heo y and esea ch. Jou nal o Ma iage and Family, 64, 585–593. h ps://doi .o g/10.1111/j.1741-3737.2002.00585.x Lüsche , K. (2005). Looking a ambi alences: The con ibu ion o a “new-old” iew o in e gene a- ional ela ions o he s udy o he li e cou se. In R. Le y, P. Ghisle a, J.-M. Le Go , D. Spini, &E.D.Widme (Eds.),Towa ds an in e disci- plina y pe spec i e on he li e cou se. Ad ances in li e cou se esea ch (Vol. 10, pp. 93–128). Ox o d, England: Else ie . h ps://doi.o g/10.1016/S1040- 2608(05)10003-3 Lüsche , K., & Ho , A. (2013). In e gene a ional ambi alence: Beyond solida i y and con lic . In I. Albe & D. Fe ing (Eds.), In e gene a ional ela- ions: Eu opean pe spec i es in amily and socie y (pp. 39–63). B is ol, England: Policy P ess. Lüsche , K., & Pilleme , K. (1998). In e gene a ional ambi alence: A new app oach o he s udy o pa en –child ela ions in la e li e. Jou nal o Ma - iage and he Family,60, 413–425. h ps://doi.o g/ 10.2307/353858 Moen, P. (1996). Gende , age, and he li e cou se. In R. H. Bins ock, L. K. Geo ge, V. W. Ma shall, G. C. Mye s,&J.H.Schulz(Eds.),Handbook o aging and he social sciences (4 h ed., pp. 171–187). San Diego, CA: Academic P ess. O e , S. (2012). The bu den o ecip oci y: P o- cesses o exclusion and wi hd awal om pe sonal ne wo ks among low-income amilies. Cu en Sociology,60, 788–805. h ps://doi.o g/10.1177/ 0011392112454754 O is, M., Guicha d, E., Nicole , M., Gab iel, R., Tholomie , A., Monno , C., ...Joye,D.(2016). Rep esen a ion o ulne abili y and he elde ly: A o al su ey e o pe spec i e on he VLV su ey. In M. O is, C. Robe s, D. Joye, & M. E ns -S aehli (Eds.), Su eying human ulne abili ies ac oss he li e cou se (pp. 27–64). Be lin, Ge many: Sp inge . h ps://doi.o g/10.1007/978-3-319-24157-9_2 Phillips, J., Ogg, J., & Ray, M. (2003). Explo ing con lic and ambi alence. In A. Lowens ein & J. Ogg (Eds.), The p ojec OASIS: Old age and au on- omy: The ole o se ice sys ems and in e gene a- ional amily solida i y, inal epo (pp. 193–226). Hai a, Is ael: Cen e o Resea ch and S udy o Aging. R De elopmen Co e Team. (2011). R: A lan- guage and en i onmen o s a is ical compu ing. Re ie ed om h p://www.R-p ojec .o g/ STATNET Rosow, I., & B eslau, N. (1966). A Gu man heal h scale o he aged. Jou nal o Ge on ology,21, 556–559. h ps://doi.o g/10.1093/ge onj/21.4.556 Sa acco, J., Cha en , M., & Kuen z, V. (2010). Ro a ion in mul iple co espondence analysis: A plana o a ion i e a i e p ocedu e. Wo king Pape s o GREThA,2010, 1–19. Re ie ed om h p://cahie sdug e ha.u-bo deaux4. /2010/2010- 04.pd Schne le , S., & Wöhle , T. (2014). On he suppo - ing ole o iendship o pa en s and non-pa en s in la e li e. A compa a i e analysis using da a om he h ee wa es o he Ge man aging su ey. In M. Löw (Ed.), Viel al und Zusammenhal : Ve handlungen des 36. Kong esses de Deu schen Gesellscha ü Soziologie in Bochum 2012 [Di e si y and Cohesion: Nego ia ions o he 36 h Cong ess o he Ge man Sociological Associa ion in Bochum 2012] (pp. 1–26). F ank u , Ge many: Campus. Se e s en, R. A., J ., & Godlewski, B. (2016). Con- cep s and heo ies o age and aging. In V. L. Beng - son & R. A. Se e s en, J . (Eds.), Handbook o heo ies o aging (3 d ed., pp. 9–25), New Yo k: Sp inge . Se e s en, R. A., J ., & Maye , K. U. (1997). The measu emen o age, age s uc u ing, and he li e cou se. Annual Re iew o Sociology,23, 233–261. h ps://doi.o g/10.1146/annu e .soc.23.1.233 Sil e s ein, M., Beng son, V. L., & Law on, L. (1997). In e gene a ional solida i y and he s uc u e o adul child–pa en ela ionships in Ame ican amilies. Ame ican Jou nal o Sociology,103, 429–460. h ps://doi.o g/10.1086/231213 Suzman, R., & Riley, M. W. (1985). In oducing he “oldes old.” The Milbank Memo ial Fund Qua - e ly:Heal h and Socie y,63, 175–186. h ps://doi .o g/10.2307/3349879. Tabachnick, B. G., & Fidell, L. S. (1996). Using mul i a ia e s a is ics (3 d ed.). New Yo k: Ha pe Collins. 784 Jou nal o Ma iage and Family Taylo , J. E., & No is, J. E. (2000). Sibling ela- ionships, ai ness, and con lic o e ans e o he a m. Family Rela ions,49, 277–283. h ps://doi .o g/10.1111/j.1741-3729.2000.00277.x T eas, J., & Ma cum, C. S. (2011). Di e si y and am- ily ela ions in an aging socie y. In R. A. Se e s en, J .&J.L.Angel(Eds.),Handbook o sociology o aging (pp. 131–141). New Yo k: Sp inge . h ps:// doi.o g/10.1007/978-1-4419-7374-0_9 Van Gaalen, R. I., & Dyks a, P. A. (2006). Solida - i y and con lic be ween adul child en and pa - en s: A la en class analysis. Jou nal o Ma iage and Family,68, 947–960. h ps://doi.o g/10.1111/ j.1741-3737.2006.00306.x Wasse man, S., & Faus , K. (1994). Social ne wo k analysis: Me hods and applica ions. Camb idge, England: Camb idge Uni e si y P ess. Wenge , G. C. (1991). A ne wo k ypology: F om heo y o p ac ice. Jou nal o Aging S ud- ies,5, 147–162. h ps://doi.o g/10.1016/0890- 4065(91)90003-B Widme , E. D. (2016). Family con igu a ions: A s uc- u al app oach o amily di e si y.NewYo k: Rou ledge. Widme , E. D., Aeby, G., & Sapin, M. (2013). Col- lec ing amily ne wo k da a. In e na ional Re iew o Sociology,23, 27–46. h ps://doi.o g/10.1080/ 03906701.2013.771049 Willson, A. E., Shuey, K. M., Elde , G. H. J ., & Wick ama, K. A. S. (2006). Ambi alence in mo he –adul child ela ions: A dyadic analysis. Social Psychology Qua e ly,69, 235–252. h ps:// doi.o g/10.1177/019027250606900302