MATTHIJS KALMIJN University of Amsterdam

Adult Children’s Relationships With Married

Parents, Divorced Parents, and Stepparents:

Biology, Marriage, or Residence?

The author compared the strength of therelationships that adult children have with dif-ferent types of parents: biological parents whoremained married, stepparents, and biologicalparents who divorced. He analyzed Dutchlife history data containing detailed measuresof living arrangements and used multilevelmodels to make comparisons both betweenand within children (N = 4,454). The resultsrevealed large differences in the strength of tiesacross parent types, but these were stronglyreduced when differences in the length of sharedresidence during childhood were taken intoaccount. Nonetheless, even after differences ininvestment opportunities were considered, therewere negative effects of divorce and positiveeffects of biological relatedness. The ‘‘marriageprotection’’ effect was stronger, especiallyfor fathers, than the biological relatednesseffect, pointing to the primacy of marriage overbiology for parent–child relations in adulthood.

One of the consequences of the rapid increasein divorce and remarriage is that ties betweenparents and adult children are increasingly

Department of Sociology, University of Amsterdam,Oudezijds Achterburgwal 185, 1012 DK Amsterdam, theNetherlands (matthijskalmijn@gmail.com).

This article was edited by Deborah S. Carr.

Key Words: divorce, intergenerational relationships, lifehistory, marriage, stepfamilies.

diverse. A father or mother can have childrenfrom a current union, stepchildren, and biolog-ical children from a previous union. For adultchildren, there is increasing diversity too, asmore and more children have one or two step-parents in addition to biological parents. Someof these stepparents were present during child-hood, whereas other stepparents entered later inthe child’s life course. Although this diversitymay not be so common as to warrant policyinterest right now, it will become more andmore common as the Baby Boom cohorts, whohave experienced high levels of divorce andremarriage, begin to enter old age.

There are two reasons why this diversity isimportant to study. First, increasing diversity inparent–child ties introduces new problems anddilemmas for parents and adult children. Forexample, children must decide how they allo-cate time and support among different types ofparents, and there are few normative or logicalguidelines about how to do this (Ganong & Cole-man, 1994, 2006). Similarly, parents may befaced with ‘‘collective ambivalence’’—a feel-ing of uncertainty about how to behave—towardtheir children (Ward, Spitze, & Deane, 2009).This may lead to differences in the way adultchildren are treated within the family, which mayin turn have repercussions in the form of inequal-ities in child well-being (Pillemer, Suitor, Pardo,& Henderson, 2010). Second, increasing diver-sity raises important theoretical questions aboutparent–child ties in general and, in particular,about the role of biology, marriage, and shared

Journal of Marriage and Family 75 (October 2013): 1181–1193 1181DOI:10.1111/jomf.12057

1182 Journal of Marriage and Family

residence for understanding parent–child ties(Anderson, Kaplan, & Lancaster, 1999; Hofferth& Anderson, 2003; King, 2009). Because thereare now several different types of parent–childties, underlying theories about the strength ofsuch ties can be tested more directly than waspossible in the past, when parent–child ties weremore homogeneous in nature.

Two types of studies have addressed theissue of increasing diversity. First, there is animportant stream of studies on the effects ofparental divorce. Some of these studies havefocused on young children and describe how theinvolvement of fathers in the child’s life changesafter divorce (Cheadle, Amato, & King, 2010;Swiss & Le Bourdais, 2009). Other studies havefocused on adult children and show that childrenwho experienced a parental divorce when theywere growing up have weaker relationshipswith their father than children whose parentsremained married (Albertini & Garriga, 2011;Aquilino, 2006; de Graaf & Fokkema, 2007).Relationships between adult children and theirmothers are negatively affected by divorce aswell, although these effects tend to be weaker(Kalmijn, 2012). The negative effect of parentaldivorce is often interpreted in terms of theinterrupted investment possibilities that fathersface after divorce. Because divorced fathers areless able to spend time with their children whenthe children are young, fathers often invest lessin their children, and the relationship graduallydeteriorates over time. Another argument is thatdivorced fathers miss the protective role ofmarriage: After divorce, fathers not only losea spouse but also a female kinkeeper (Kalmijn,2007b; Stephens, 1996).

A second stream of studies has examinedthe role of stepparenting (Ganong & Cole-man, 1994; Ganong, Coleman, & Jamison, 2011;King, 2006). Most of these studies focused onyoung children and used a within-family design.Some authors found no differences in fatherinvolvement between stepchildren and biologi-cal children in such blended families (Hofferth& Anderson, 2003), but other authors havefound that fathers are closer on average to theirbiological children than to their stepchildren(Evenhouse & Reilly, 2004; Schnettler & Stein-bach, 2011). The negative effect of being astepparent has often been interpreted in termsof the importance of biological relatedness forparental investments in children. Other studieshave compared different types of fathers of the

same child. For example, King has comparedthe child’s relationship with his or her stepfa-ther, his or her nonresident biological father,and his or her resident biological mother (King,2006, 2007, 2009; White & Gilbreth, 2001).These studies show that ties to stepfathers areusually stronger than ties to nonresident bio-logical parents, suggesting that shared residenceis more important than biological relatedness(King, 2006). Most of these studies focusedon young children who were still living withtheir parent(s). Less is known about differencesbetween stepparents and biological parents at anadult age.

In this study, I compared adult children’srelationships with three types of parents:(a) biological parents who remained married,(b) biological parents who divorced when thechild was young, and (c) stepparents. Compar-isons were made for both fathers and mothers sothat there were six types of parents to consider.To make these comparisons, I used a recentlyconducted survey in the Netherlands in whichinformation was gathered on all parent figureswho played a role in the respondent’s childhood(Oudejans & Kalmijn, 2013). Because of thedetailed retrospective data on childhood livingarrangements, I was able to examine how manyyears of shared residence each parent had withthe child. Moreover, for each parent, informa-tion was collected on the current relationship,in particular on how much contact there is, howmuch support is being exchanged, and on theperceived quality of the tie. In the analyses, Ipooled all types of parents and used multilevelmodels to analyze parent–child relationships(i.e., random effects regression and fixed effectsregression models). In the random effects mod-els, I made comparisons between children andparents simultaneously. In the fixed effects mod-els, I made comparisons within (adult) children,that is, different parents belonging to the samechild. Note that when I talk about stepparents, Imean stepparents with whom the child lived inthe past.

The contribution of this study to the existingliterature lies in its focus on children’s relation-ships to their parents when the children are adultand living on their own. Most of the work onstepparents has focused on children who are stillliving at home. An important implication of thisshift in focus is that we obtain a view of thelong-term effects of shared residence. Previous

Adult Children’s Relationships With Parents 1183

studies have examined effects of the current resi-dence situation of the parent. For example, King(2006) showed that young children are closerto a (resident) stepfather than to a nonresidentfather. Moreover, she showed that close ties toa stepfather are more beneficial for child out-comes than close ties to a nonresident biologicalfather. In King’s view, these findings support thehypothesis that residence is more important thanbiology. I focus on the relationships that childrenhave with their parents when children are adultsand living on their own, and I relate this to theresidence history of the parent and child. Specifi-cally, I examined effects of the duration of sharedresidence when young on the present quality ofthe tie between parent and adult child. In addi-tion, I examined the extent to which differencesin the quality of different types of parent–childties can be explained by residential histories.

Another contribution of this research is that Iwas able to examine the effects of both residenceand marriage. In previous work that focusedon young children, marriage and residenceoverlap, except in the case in which parents areliving together unmarried (Hofferth & Anderson,2003). In my study, the focus was on adultchildren, and this leads to a different view of theeffects of marriage. More specifically, the effectof marriage here pertains to the issue of whetheror not the father of the child is still married tothe mother of the child. As I discuss below, Iexpected the marital tie to have an additionalbeneficial effect on the father–child relationshipwhen children are older, quite apart from howmuch time the father lived with the child in thepast (the residence effect). Note that I abstainedfrom analyzing the difference between marriedand cohabiting parents because my concern waswith the question of whether the father is stilltied to the mother residentially, not with the legalnature of that tie.

Finally, I added a different setting to theliterature by examining these issues as theypertain to the Netherlands. Although my workis largely occupied with testing hypothesesthat are applicable more generally, it remainsimportant to broaden the empirical base. TheNetherlands is a good example of a modernwestern European country. Dutch divorce andmarriage rates are average for western Europe(Kalmijn, 2007a), and the frequency of contactbetween adult children and older parents is alsocomparable to that in other western Europeancountries (Hank, 2007).

BACKGROUND AND HYPOTHESES

One debate in family studies revolves around thequestion of whether parents ‘‘favor’’ childrenwho are biologically related to them. Whenparents invest time and effort in their children,their children are more likely to survive. If oneassumes that the tendency to invest in childrenis at least in part genetically determined, one canthen argue that investing in biological childrenhas a reproductive advantage (Anderson et al.,1999; Hofferth & Anderson, 2003; Schnettler& Steinbach, 2011). In other words, if parentsinvest in biological children, their genesare more likely to be transmitted to futuregenerations. In evolutionary models, genesare the only relevant actors in the evolution(Dawkins, 1976). Hence, one could argue thatgenes that discriminate between biological andnonbiological offspring have a greater chance ofsurviving the evolutionary process than genesthat do not discriminate. As a result, parentswould be more likely to invest in biologicaloffspring than in stepchildren.

Even if the evolutionary theory does notapply to the present case, biology is relevantbecause it also carries a social distinction.People make a difference between biologicaland stepchildren in their way of thinking aboutthemselves and about others. This distinctiontherefore becomes relevant in the way parentsbehave toward others as well. One example ofthis lies in the ambiguous social role of thestepfather. Because a stepfather is often notconsidered to be the ‘‘real’’ father, he may beexcluded from the domain of parenting by themother, and the children themselves may tryto maintain a distance from him (Ganong &Coleman, 1994; Hofferth & Anderson, 2003).These differences lead to lower investments inchildren, which in turn will translate into weakerties with the children at an adult age. Researchalso has shown that normative obligations tosupport older parents are weaker when parentsare not biologically related (Ganong & Coleman,2006). The first hypothesis thus was that adultchildren have a stronger tie to their biologicalparents than to their stepparents (the biologyhypothesis).

A second debate revolves around the impor-tance of marriage for parent–child ties (Hofferth& Anderson, 2003; Stephens, 1996). Severalauthors have argued that, especially for fathers,marriage and children are a ‘‘package deal’’(Kalmijn, 2007b; Stephens). The reasoning

1184 Journal of Marriage and Family

behind this is that mothers invest more in fam-ily relationships than fathers. Mothers arrangefamily meetings, they initiate contact with adultchildren, they keep track of birthdays, they sendholiday cards—in short, they are the kinkeep-ers (Di Leonardo, 1987). This division of laborin marriage coincides with the marital divisionof labor in other domains, such as householdlabor and paid labor. When fathers are marriedto the mother of their children, they benefit fromthis division of labor because children tend tovisit parents together and they provide socialand emotional support to both parents simul-taneously. When fathers divorce, they lose notonly a spouse but also a kinkeeper. As a result,fathers may be less able to maintain ties withtheir children when the children are adults andthe fathers are not living with the mother any-more. A similar but weaker argument can bemade for parents in general, regardless of gen-der. When parents are divorced, adult childrencannot see their parents simultaneously, leadingto less time available per parent (Kalmijn, 2012).Adult children with divorced parents may alsoface conflicting loyalties after divorce, espe-cially when there was much conflict, and thismay lead to less closeness to one of the par-ents, either the father or the mother (Amato &Afifi, 2006). The second hypothesis was thatadult children have a stronger tie to a parentwho remained married than to a parent whodivorced. This effect is stronger for fathers thanfor mothers (the marriage hypothesis).

In their work on stepfathers, Hofferth andAnderson (2003) also used the perspectives ofmarriage and biology to understand involvementof fathers in (young) children. Because theyfound few differences between stepchildrenand biological children in the same family(i.e., in blended families), they concluded thatmarriage was a more influential factor thanbiology. It is also possible, however, that bothhypotheses are true. Biological children canhave an advantage when their father is stillmarried (a marriage advantage), and biologicalchildren can have an advantage to stepchildren(a biological advantage). To see which influenceis stronger, consider all three types of parents:(A) married biological fathers, (B) divorcedbiological fathers, and (C) stepfathers. If themarriage effect (A–B) is stronger than thebiology effect (A–C), this implies that C > B.Hence, children’s ties to stepfathers from a unionthat still exists should be stronger than ties to

biological fathers who divorced. Interestinglyenough, this comparison can also be made withina child. Moreover, the effect can be examinedat an adult age, something that has not beendone in most earlier work on stepparenting.The third hypothesis, therefore, was that adultchildren have stronger ties to a stepfather thanto a biological father who divorced (primacy ofmarriage over biology). For mothers, a similarargument can be made in principle, but I arguethat the marriage effect applies in particularto fathers, and hence it is less clear what theoutcome will be.

Another and often-used hypothesis aboutparent–child ties centers around the notion ofshared residence (King, 2006). Previous studieshave focused on shared residence when childrenare young. In these cases, residence refersto the effect of living with a parent on therelationship with the parent. When the focusis on adult children, residence refers to theeffect of the length of time the child livedwith the parent on the quality of the tie at alater age, when the child is living independently.The amount of time that the parent and childspend together when the child is growing upis an important indicator of the investmentsthat parents make in children (Sayer, Bianchi,& Robinson, 2004). Investments are usuallydefined as the efforts that parents make tocontribute to the emotional and physical well-being of their children. According to research onpersonal relationships, investments in others arealso investments in the relationship (Rusbult,Martz, & Agnew, 1998). Hence, investmentsin children will not only contribute to childwell-being but will also lead to a strongerparent–child relationship in the long run. Thefourth hypothesis was that the longer the parentand child shared residence during childhood,the stronger the tie between parent and adultchild (duration hypothesis).

The duration hypothesis may also be relevantfor comparing types of relationships. Morespecially, I argue that the history of sharedresidence in part explains differences betweendifferent types of father–child relationships.One reason why divorced fathers in particularwould experience a deterioration of the tiesto their children lies in the fact that fathersrarely get custody. Although many divorcedfathers remain involved in the lives of theirchildren, this involvement will typically beless intensive than it was before the divorce.

Adult Children’s Relationships With Parents 1185

An early decline in involvement can beseen as a decline in investment in children,which, according to principles of reciprocity,will lead to less frequent support from andcontact with children when children are older(Aquilino, 2006; de Graaf & Fokkema, 2007).Suggestive evidence for this notion is that thelong-term effects of divorce on father–childrelationships are generally less negative whenthe parents divorced at an older age (Aquilino,1994; Kalmijn, 2012). Hence, the long-termnegative effect of divorce may in part bedue to a difference in the length of sharedresidence. Similar arguments can be madeabout stepparents: Relationships between adultchildren and stepparents may be weaker, butthis can be due to the more limited time thata stepparent was involved in the child’s life.A stepparent who was present from birth, forexample, would not be perceived as differentfrom a biological father (Hamilton, Cheng,& Powell, 2007). Because stepparents oftenlive a shorter period of time with the childthan biological parents (although not necessarilyshorter than divorced nonresident parents), thelong-term effect of ‘‘biology’’ may in part be dueto a difference in the length of shared residence.The last hypothesis was that the negative long-term effects of divorce and stepparenting onparent–child ties are partly due to a shorterduration of shared residence (spuriousnesshypothesis).

METHOD

The data I used come from the survey Lon-gitudinal Internet Studies for the Social Sci-ences, which is publicly available throughhttp://www.lissdata.nl. This survey is based ona representative probability sample of approxi-mately 5,000 households in the Dutch popula-tion. The response rate at the household levelwas 48%, which is about average for responserates in the Netherlands (Scherpenzeel, 2009).All household members 16 years of age andolder were asked to complete short Internetquestionnaires. Respondents were paid whenthey completed a questionnaire. The Nether-lands is in the top 10 countries with the highestInternet penetration rates (World Telecommuni-cation/ICT Indicators Database Online, 2012).Households without Internet access (or with-out broadband) received a broadband Internetconnection and were loaned a computer if they

did not have one. Older respondents withoutInternet access and a computer received trainingin how to use the Internet. Each month, a dif-ferent questionnaire was presented, covering aspecific topic (e.g., work, health, family). I useda special biographical module that was held inJuly and August 2012 and that contained detailedretrospective questions on family background(Oudejans & Kalmijn, 2013). The response rateon this specific module was 84% (N = 5,247).The module was presented only to householdmembers who were the head of the householdor the partner of the head (single persons wereincluded). None of these respondents lived withparents.

Respondents were asked about their livingarrangement at birth and what changes theyexperienced in their living arrangement beforeleaving home and living independently. If theirliving arrangement changed, detailed informa-tion was gathered on all subsequent households.For one of these ‘‘secondary’’ households,additional questions were asked about the parentfigures. If there were two or more secondaryhouseholds, the questionnaire focused on theone in which the child spent most of his or herchildhood. If one of the secondary householdswas a stepfamily, however, this householdwas given precedence in order to not miss anystepparent figure in the child’s life. Of the 5,247respondents, 706 did not live with both their bio-logical parents at least some time before leavinghome. Of these, 229 lived in stepfamilies.

The data were analyzed as parent–child dyads.When I analyzed contact, quality, and support,the sample was limited to parents who are stillliving. The number of living parents was 4,454,of whom 501 were divorced parents and 117were stepparents. The first dependent variable inthe parent–child analysis was the frequency offace-to-face contact between parent and child.This was coded into midpoint values assignedto the frequency categories (e.g., weekly =52). To avoid the skewness of the resultingvariable, the frequencies were logged. Thesecond dependent variable was the quality ofthe relationship between the father and the child.The quality was measured on a 5-point scale(1 = very good, 2 = good, 3 = reasonable,4 = not so good, 5 = poor). The lastvariable was support exchange with the parent.Seven items were presented: (a) receivinghousehold help from the parent, (b) givinghousehold help to the parent, (c) receiving other

1186 Journal of Marriage and Family

Table 1. Means and Standard Deviations of VariablesUsed in the Analyses (N = 4,454)

Variable M SD Minimum Maximum

Daughter .58Number of siblings 1.98 1.48 0 10Education (years) 12.43 2.91 6 17Age child 41.18 10.48 18 76With partner .57Log contact 2.66 1.26 0 5.86Quality 3.88 1.04 1 5Support 1.83 0.44 1 3Duration of

residence (years)17.02 3.29 0 18

practical support from the parent, (d) givingother practical support to the parent, (e) receivinginterest in one’s personal well-being from theparent, (f) showing interest in the parent’spersonal well-being, and (g) receiving goodadvice from the parent. For each type of support,respondents were asked how often the supporthad been exchanged in the past 3 months (1 =never, 2 = occasionally, 3 = often). The scale isthe sum of the items and the reliability was good(α = .76 for mothers, .78 for fathers, and .78 forstepparents). The means and standard deviationsof all variables are presented in Table 1.

The main independent variables wereobtained from the retrospective data on child-hood living arrangements. I considered six typesof parents: (a) biological mothers, (b) divorcedmothers, (c) biological fathers, (d) divorcedfathers, (e) stepmothers, and (f) stepfathers. Adivorce had to occur during childhood. Marriageand cohabitation were treated as the same,but there were few cohabiting couples withchildren in the parent generation. The effectof parent type was approached with a set ofdummy variables. Different types of contrastsfor these dummy variables were used dependingon the hypothesis (see below). I also made finerdistinctions in additional analyses (discussedin the text) between divorced parents who wereresident parents after divorce and divorcedparents who were nonresident parents. Finally,I used the number of years that the parent andchild lived together. The age at leaving homewas also used in constructing these variables,but years at home after age 18 were not countedgiven the high level of independence afterthis age.

For the random effects models, I includedcharacteristics of the adult child as controlvariables: the respondent’s age, sex, the numberof siblings, whether the child is living with apartner, and the level of education (coded asyears of schooling). These variables are knownto be related to contact frequency and supportexchange. Information on geographical distancewas not included because this may potentiallymediate the effects (Silverstein, 1995). Iexplored effects of the parent’s socioeconomicstatus, religiosity, and education, but none ofthese affected contact and quality. Becauseusing these variables introduces missing valuesin the analyses, I decided to not include them inthe models.

I used multivariate random and fixed effectsregression models. The fixed effects modelscompare parents within children, and thiscontrols implicitly for all unmeasured childcharacteristics. This is similar in spirit toKing’s (2006) analyses, which were based onyoung resident children. The random effectsmodels make comparisons between children andfathers simultaneously and control for measuredchild characteristics. Both types of modelstake into account the clustering of parent–childrelationships within adult children. Because ofthe relatively small samples, the random effectsmodels will be more powerful statistically.

RESULTS

In Table 2, I present information on the amountof time that different types of parents sharedwith the child. Biological parents who did notdivorce lived with the child for 17.7 years. Thereis also variance in this number, because theage at leaving home varied. When the parentsdivorced, the child lived with the biologicalfather only 11.1 years, while living 15.9 yearswith the biological mother. The variance in thesedurations is substantial. If one separates thesefigures depending on whether the parent was theresident parent or not, one sees even fewer yearsfor the nonresident divorced parent, as expected.Interesting is that when mothers were thenonresident parent, duration of coresidence washigher than when fathers were the nonresidentparent. This likely has to do with the fact thatfathers rarely get custody when children are veryyoung (Kalmijn & de Graaf, 2000). Stepparentshad the shortest duration of shared residencewith the child: 5.6 years for stepmothers and 8.0

Adult Children’s Relationships With Parents 1187

Table 2. Duration of Shared Residence (in Years) by Typeof Parent

Type of parent M SD na

Biological mother 17.70 1.90 4,452Biological father 17.51 2.64 4,452Stepmother 5.61 4.42 79Stepfather 8.03 5.22 150Divorced mother 15.94 3.97 408

Resident after divorce 17.36 2.17 279Nonresident after divorce 13.50 4.38 70

Divorced father 11.12 5.55 408Resident after divorce 15.49 3.75 70Nonresident after divorce 9.82 5.29 279

aBased on full life history data, including parents whoare no longer living.

years for stepfathers. This shows that biologicaldivorced fathers share, on average, still moretime with the child than stepfathers.

In Table 3, I present the random effectsmodels. All three outcome variables werestandardized before the model was estimatedso that coefficients can be interpreted in termsof effect sizes. Before I turn to the tests ofthe hypotheses, I discuss the variance decom-position. Parents are nested within children,and hence the question arises of how muchof the variance is between children and howmuch is between parents within children. Toexamine this, I estimated an empty modeland calculate the intraclass coefficient ρ,which is the amount of variance that is dueto the between-child differences. For contact,ρ = .73; for quality, ρ = .61; and for support,ρ = .73. Hence, there was considerable simi-larity in the relationships that a given child haswith different parents. Nonetheless, the within-children correlation is not perfect, and this leavesroom for the effects I assumed in our hypotheses.

I first discuss contact frequency. Biologicalmothers who did not divorce are the referencecategory for other types of mothers; biologicalfathers who did not divorce are the referencecategory for other types of fathers. To obtainthese contrasts, two models were estimated, andthe results were merged into one table (Table 3).I observed negative effects of divorced mothers,showing that adult children have less frequentcontact with divorced mothers than with marriedmothers. Adult children also have significantlyless frequent contact with stepmothers than withbiological mothers. I did not see a difference

in contact between biological mothers andfathers. When switching the reference categoryto biological fathers, I observed a negative effectof stepfathers and divorced fathers, with thelatter effect being stronger than the former (z= 2.10). All the effects, with the exception ofthe difference between divorced and marriedmothers, are substantial in magnitude. In sum,adult children have much less contact withdivorced parents and with stepparents than withbiological married parents. This is true for bothfathers and mothers, but the order is different.For mothers, stepparenting is most negative;for fathers, divorce is most negative. At first,these results confirm the biology hypothesis andthe marriage hypothesis. I note that the effectof divorce is significantly stronger for fathersthan for mothers, showing that the marriagehypothesis applies mostly to fathers (z = 10.23).

In Model 2, I added duration of residence.I noted a significant effect of duration: Foreach year of shared residence, contact frequencyincreased by 0.055 SD. Hence, when I compared0 and 18 years of residence, the maximumrange, the effect was 0.99 SD, which is asubstantial effect. This confirms the sharedresidence hypothesis. The more time parent andchild shared when the child was young, the morecontact there currently is. More important, Iobserved that differences between different typesof parents are strongly reduced when sharedresidence is added to the model. The effectsof stepparents are reduced by 82% for fathersand 72% for mothers, and the effects of divorceare reduced by 49% for fathers and 29% formothers. Most striking is that the disadvantagefor stepfathers (compared to biological fathers)is no longer significant when one takes durationinto account. This confirms the spuriousnesshypothesis and refutes the biology hypothesis,at least for contact. The marriage hypothesis,especially for fathers, still stands, because thereremains a negative effect of divorce on contacteven when the shorter period of residence istaken into account. I do note, however, that abouthalf of the initially observed divorce effect is dueto the fact that divorced fathers spent less timeliving with the child.

Next, I turn to the models for quality. Theeffects of the type of parent were similar tothose for contact: Adult children have strongerties to biological parents than to stepparents anddivorced parents. For mothers, stepparentingis most negative; for fathers, divorce and

1188 Journal of Marriage and Family

Table 3. Random Effects Regression of Parent–Child Relationships on Selected Variables ( N = 4,454)

PredictorModel 1:Contact

Model 2:Contact

Model 3:Quality

Model 4:Quality

Model 5:Support

Model 6:Support

Daughter −0.025 −0.021 0.032 0.036 0.253∗ 0.258∗

(0.037) (0.037) (0.035) (0.035) (0.036) (0.036)Number of siblings −0.024† −0.023† −0.002 −0.002 −0.045∗ −0.045∗

(0.012) (0.012) (0.012) (0.012) (0.012) (0.012)Education −0.012† −0.012† 0.008 0.008 0.024∗ 0.024∗

(0.006) (0.006) (0.006) (0.006) (0.006) (0.006)Age child −0.004∗ −0.004∗ −0.011∗ −0.012∗ −0.009∗ −0.009∗

(0.002) (0.002) (0.002) (0.002) (0.002) (0.002)With partner −0.089∗ −0.091∗ −0.021 −0.023 −0.164∗ −0.167∗

(0.039) (0.038) (0.037) (0.037) (0.037) (0.037)Type of parent

Biological mother —c —c —c —c —c —c

Biological father −0.036∗ −0.030† −0.046∗ −0.042∗ −0.107∗ −0.101∗

(0.018) (0.018) (0.022) (0.021) (0.018) (0.017)Stepmothera −1.034∗ −0.291∗ −1.376∗ −0.708∗ −1.093∗ −0.417∗

(0.124) (0.138) (0.135) (0.151) (0.121) (0.134)Stepfatherb −0.522∗ −0.093 −0.665∗ −0.255∗ −0.664∗ −0.274∗

(0.087) (0.093) (0.095) (0.104) (0.084) (0.090)Divorced mothera −0.225∗ −0.160∗ −0.401∗ −0.334∗ −0.258∗ −0.198∗

(0.060) (0.060) (0.060) (0.060) (0.058) (0.058)Divorced fatherb −0.697∗ −0.358∗ −0.828∗ −0.515∗ −0.788∗ −0.480∗

(0.064) (0.069) (0.065) (0.073) (0.062) (0.067)Duration of residence 0.055∗ 0.049∗ 0.050∗

(0.005) (0.005) (0.005)Constant 0.501∗ −0.462∗ 0.480∗ −0.393∗ 0.262∗ −0.613∗

(0.124) (0.148) (0.118) (0.150) (0.119) (0.143)σ u .85 .85 .75 .75 .82 .82σ i .51 .50 .62 .61 .50 .49R2 .044 .060 .076 .086 .105 .114

Note: Numbers in parentheses are standard errors. Dependent variables are standardized. σ u is the standard deviationbetween children; σ i is the standard deviation within children (across parents).

aReference category is biological mothers. bReference category is biological fathers. cReported results are based ondifferent models, with different reference categories (see text).

†p < .10. ∗p < .05.

stepparenting are about equally negative. Model4 shows that duration has a strong and significanteffect. The longer a parent and child weretogether, the closer they are when the childis adult. Differences between types of parentsare reduced considerably when duration is takeninto account. These reductions are especiallystrong for stepfathers: Sixty-two percent of thedifference between stepfathers and biologicalfathers disappears when duration is taken intoaccount. For divorced fathers, the reduction isalso substantial (38%), but the remaining effectof divorce is still strong (b = −.52).

Finally, I turn to the models for supportexchange. The findings are in line with thosefor quality. Adult children exchange less supportwith stepfathers and divorced fathers than withbiological fathers. They also exchange lesssupport with stepmothers than with biologicalmothers and somewhat less support withdivorced mothers than with married mothers.The effect of duration is significant and explainsmuch of the differences: 62% of the gap betweenbiological and stepmothers and 59% of the gapbetween biological fathers and stepfathers. Thedisadvantage of being divorced for fathers isalso reduced substantially. The effects remain

Adult Children’s Relationships With Parents 1189

significant, however, suggesting that differencesare not entirely spurious.

Is marriage more influential than biologicalrelatedness? To test this, I compared divorcedfathers and stepfathers. After controlling forduration, children have more contact withstepfathers than with biological divorced fathers(−.093 vs. −.358, z = 4.05). The same resultwas obtained when I looked at quality (−.255vs. −.515, z = 2.70) and support exchange(−.274 vs. −.480, z = 2.58). Hence, on allthree counts adult children have a stronger tie totheir stepfather than to their divorced biologicalfather when the length of shared residence is heldconstant. In this sense, I support the primacy ofmarriage over biology for fathers. For mothers,there was only one significant difference: Thequality of the relationship is higher for biologicaldivorced mothers than for stepmothers (z =2.49). I note, however, that I did not expecta stronger marriage effect here. Moreover, fordivorced mothers there will often have beenshared residence after divorce. I turn to thisissue next.

So far, no distinction has been made betweendivorced parents who were resident parents anddivorced parents who were not. In additionalmodels, I examined this difference. To someextent, this postdivorce residence effect willbe captured by the length of residence, but itmay also have an independent effect. Sharedresidence is determined here only for the mostimportant secondary household. For fathers, Ifound positive and significant effects of beingthe resident (divorced) parent (.431 for contact,.886 for quality, and .604 for support, allps < .01). The effects of duration were stillpositive and significant in these models. Theseresults show that not only longer residencebut also postdivorce residence is influentialin maintaining good relationships with adultchildren. For divorced mothers, these effectswere smaller and less often significant, probablybecause of the smaller subsample (.206, p = .05,for contact; .295, p < .01, for quality; and .170,ns, for support).

The effects of the control variables are inline with prior research. Daughters exchangemore support than sons with their parents,and being married and older was associatedwith less support exchange. More highlyeducated children have somewhat less frequentcontact, but they exchange more support withtheir parents. In larger families, finally, I

observed less contact (per dyad) and lesssupport, but this did not translate into weakerrelationships in the sense of lower perceivedquality.

To what extent are my findings affectedby unmeasured differences between children?To address this, I made comparisons betweendifferent parents of the same children. InTable 4, I present the estimates of the fixedeffects regression models. I can make severalmeaningful within-child contrasts here. First,I compared biological fathers and stepmothersas well as biological mothers and stepfathers.These within-child comparisons apply implicitlyto persons who grew up (part of the time) with abiological parent and stepparent of the oppositesex. Second, I can make comparisons withingender: stepfathers with divorced biologicalfathers and stepmothers with divorced biologicalmothers. These divorced parents were implicitlythe nonresident parents, because the comparisonis made within children, and hence thecomparison is based on children who haveboth a stepparent and a divorced parent of theopposite sex.

The results presented in Table 4 show thatthere is no difference between stepfathers anddivorced biological fathers. This applies to allthree outcomes. When I controlled for durationof residence, however, stepfathers appear tohave an advantage, at least in terms of contactfrequency and support exchange. This againsuggests that marriage is more influential forfathers than biology. For the (smaller group of)stepmothers, the conclusion is different. Here,I see that adult children have stronger ties totheir divorced biological mothers than to theirstepmothers. These differences, however, are inpart explained by differences in the length ofshared residence.

I also made comparisons of parents whowere involved in the same postdivorce family:Divorced mothers and stepfathers are the mostcommon case. Here I again see an advantageof divorced biological mothers over stepfathers.This advantage is fully explained by the lengthof residence, however. The same conclusionapplies when I compared divorced fathers andstepmothers.

DISCUSSION

Using comparisons of adult children’s relation-ships with multiple types of parents, this research

1190 Journal of Marriage and Family

Table 4. Fixed Effects Regression of Parent–Child Relationships on Selected Variables (N = 4,454)

PredictorModel 1:Contact

Model 2:Contact

Model 3:Quality

Model 4:Quality

Model 5:Support

Model 6:Support

Type of parenta

Biological father −0.033† −0.025 −0.044† −0.036 −0.088∗ −0.080∗

(0.019) (0.018) (0.023) (0.022) (0.018) (0.018)

Stepmother −0.738∗ −0.398 −0.656† −0.293 −0.988∗ −0.634∗

(0.308) (0.302) (0.374) (0.369) (0.302) (0.296)

Stepfather −0.102 −0.106 −0.057 −0.061 −0.598∗ −0.602∗

(0.254) (0.248) (0.309) (0.303) (0.250) (0.243)

Divorced mother 0.259 −0.122 0.333 −0.073 −0.072 −0.469†

(0.270) (0.266) (0.328) (0.325) (0.265) (0.261)

Divorced father −0.236 −0.333 −0.092 −0.196 −0.699∗ −0.801∗

(0.271) (0.264) (0.329) (0.323) (0.266) (0.258)

Duration of residence 0.054∗ 0.058∗ 0.057∗

(0.006) (0.007) (0.006)

Constant 0.016 −0.888∗ 0.007 −0.958∗ 0.091∗ −0.850∗

(0.038) (0.102) (0.047) (0.125) (0.038) (0.100)

ContrastsStepfather–divorced fatherb .133 .227∗ .035 .135 .101 .199∗

Stepmother–divorced motherc −.996∗ −.276† −.989∗ −.220 −.916∗ −.165Stepfather–divorced motherc −.361∗ .016 −.390∗ .013 −.526∗ −.133Stepmother–divorced fatherb −.502∗ −.065 −.564∗ −.097 −.289∗ .167

σ u .97 .95 .96 .92 .95 .95σ i .51 .50 .62 .61 .50 .49R2 .010 .051 .002 .062 .044 .051

Note: Numbers in parentheses are standard errors. Dependent variables are standardized. σ u is the standard deviationbetween children; σ i is the standard deviation within children (across parents).

aReference category: Biological mother. bDivorced father was resident father when compared to stepmother; divorcedfather was nonresident father when compared to stepfather. cDivorced mother was resident mother when compared tostepfather; divorced mother was nonresident mother when compared to stepmother.

†p < .10. ∗p < .05.

was able to shed more light on alternative the-oretical arguments about parent–child ties. Thefocus was on the strength of the tie betweenparent and adult child—measured with con-tact frequency, support exchange, and perceivedquality—and this tie was linked to childhoodliving arrangements. The first hypothesis arguedfor an effect of biology, that is, the positive influ-ence of being biologically related to a parent.A second hypothesis argued for the protectiveeffect of marriage. Especially for fathers, beingmarried to the child’s mother strengthens the tiesto his children, not only when children are youngbut also when children are older and indepen-dent. Third, I argued that differences betweentypes of parents are due to the amount of invest-ments parents made while the children were

young. I examined this via the effect of durationof residence, assuming that longer periods ofshared residence would be associated with moreinvestments. Coparenting will probably makethis assumption less valid, but in the histories ofthe children analyzed here, coparenting was notthat common (Spruijt & Duindam, 2009).

I found strong evidence for the influence ofduration: The longer parents and children weretogether in childhood, the more contact therewas later, the more support was exchanged, andthe more positive the adult child was aboutthe relationship. There is not much variance induration for biological parents who remainedmarried, so the question is whether this effectis not simply due to an additional effect ofmarriage. I checked the duration effect for

Adult Children’s Relationships With Parents 1191

stepparents and divorced parents and foundsignificant duration effects in this subsampleas well. My second hypothesis about durationwas also confirmed. I found that differencesin duration explain about 40% to 60% of thedifferences in the strength of the ties betweendifferent types of parents. In particular, theinitial ‘‘disadvantage’’ of stepfathers is stronglyreduced when one holds constant the length ofshared residence. The disadvantage of divorcedfathers is also to some extent due to the factthat divorced fathers live a shorter period oftime with the child. Hence, a large part of theoriginal differences are spurious. For divorcedfathers, this is mainly due to the interruptionof their investment options after divorce. Forstepfathers, this is mainly due to the fact thatthey usually do not enter the child’s life veryearly.

There is still some evidence for the marriagehypothesis and the biology hypothesis. For twoof the three outcomes (quality and support,but not contact), the effects of marriage (i.e.,divorce) and stepparenting remain significantwhen controlling for duration. If both hypothesesare true to some extent, the question is, whichof the two hypotheses is more important?To examine this, I compared stepfathers anddivorced fathers. When I made this comparison,and when I controlled for duration, it appearsthat children have somewhat stronger ties totheir stepfather than to their divorced biologicalfather. This suggests that marriage is moreimportant than biology. My finding echoes theresults of King (2006), who found that youngchildren feel somewhat closer to their stepfatherthan to their divorced biological father. Sheinterpreted this in terms of the ‘‘primacy ofresidence.’’ I found the same pattern whenlooking at adult children, pointing, in my view,to the ‘‘primacy of marriage.’’ The interpretationis different because the marriage effect pertainsto the kinkeeping role of the (former) spouse ofthe father, whereas the residence effect pertainsto the day-to-day investments that parents andchildren are able to make in their relationship.

I end with some limitations and suggestionsfor further research. One limitation is therelatively small sample size. I tested all thedifferences with two types of models and forthree outcomes, and the results seem robust.Nevertheless, I await replications of my findingswith larger samples. With larger samples,there are also possibilities to make further

distinctions, for example, between differenttypes of stepparents. For example, we couldexplore the dissolution of remarriage and thedeath of the remarried biological parent. Isuspect that one reason why relationshipswith stepmothers are relatively weak lies inthese underlying factors, but the sample sizeprohibited me from analyzing this issue.

Finally, I would like to advocate studies thatincorporate both within-child and within-parentcomparisons. Within-child designs basicallycontrol for unmeasured child characteristics.For example, if stepchildren have behavioralproblems or if stepchildren are very independent,they may have developed weaker relationshipswith all their parents, regardless of the type ofparent. Such differences would be taken intoaccount when comparing different parents ofthe same child. This design does not, however,control for unmeasured parent characteristics.Within-family (or better within-parent) designs,on the other hand, control for unmeasured parentcharacteristics. For example, it could be possiblethat stepfathers have more behavioral problemsthan married fathers, and such differences mayaffect the comparisons between different typesof parent–child types in an unfair way. Whencomparing different children of the same parent,such differences are controlled, but differencesbetween children are not. Using both designswith one data set in one study would thereforeyield a more powerful conclusion than a studywith only one of the two designs.

REFERENCES

Albertini, M., & Garriga, A. (2011). The effectof divorce on parent–child contacts: Evidenceon two declining effect hypotheses. EuropeanSocieties, 13, 257–278. doi:10.1080/14616696.2010.483002

Amato, P. R., & Afifi, T. D. (2006). Feeling caughtbetween parents: Adult children’s relations withparents and subjective well-being. Journal ofMarriage and Family, 68, 222–235. doi:10.1111/j.1741-3737.2006.00243.x

Anderson, K. G., Kaplan, H., & Lancaster, J.(1999). Paternal care by genetic fathers andstepfathers: I. Reports from Albuquerque men.Evolution & Human Behavior, 20, 405–431.doi:10.1016/S1090-5138(99)00023-9

Aquilino, W. S. (1994). Later life parental divorce andwidowhood: Impact on young adults’ assessmentof parent–child relations. Journal of Marriage andthe Family, 56, 908–922. doi:10.2307/353602

1192 Journal of Marriage and Family

Aquilino, W. S. (2006). Noncustodial father–childrelationship from adolescence into young adult-hood. Journal of Marriage and Family, 68,929–946. doi:10.1111/j.1741-3737.2006.00305.x

Cheadle, J. E., Amato, P. R., & King, V.(2010). Patterns of nonresident father contact.Demography, 47, 205–225.

Dawkins, R. (1976). The selfish gene. Oxford, UK:Oxford University Press.

de Graaf, P. M., & Fokkema, T. (2007). Contactsbetween divorced and non-divorced parents andtheir adult children in the Netherlands: Aninvestment perspective. European SociologicalReview, 23, 263–277. doi:10.1093/esr/jcl032

Di Leonardo, M. (1987). The female world of cardsand holidays: Women, families, and the work ofkinship. Signs, 12, 440–453.

Evenhouse, E., & Reilly, S. (2004). A siblingstudy of stepchild well-being. Journal of HumanResources, 39, 248–276. doi:10.2307/3559012

Ganong, L. H., & Coleman, M. (1994). Remar-ried family relationships. Thousand Oaks, CA:Sage.

Ganong, L., & Coleman, M. (2006). Obligationsto stepparents acquired in later life: Relationshipquality and acuity of needs. Journals of Gerontol-ogy: Series B: Psychological Sciences and SocialSciences, 61, 80–88.

Ganong, L. H., Coleman, M., & Jamison, T.(2011). Patterns of stepchild–stepparent rela-tionship development. Journal of Marriageand Family, 73, 396–413. doi:10.1111/j.1741-3737.2010.00814.x

Hamilton, L., Cheng, S., & Powell, B. (2007).Adoptive parents, adaptive parents: Evaluatingthe importance of biological ties for parentalinvestment. American Sociological Review, 72,95–116.

Hank, K. (2007). Proximity and contacts betweenolder parents and their children: A Europeancomparison. Journal of Marriage and Family, 69,157–173. doi:10.1177/0192513X08322627

Hofferth, S. L., & Anderson, K. G. (2003). Areall dads equal? Biology versus marriage as abasis for paternal investment. Journal of Marriageand Family, 65, 213–232. doi:10.1111/j.1741-3737.2003.00213.x

Kalmijn, M. (2007a). Explaining cross-nationaldifferences in marriage, cohabitation, and divorcein Europe, 1990–2000. Population Studies, 61,243–263. doi:10.1080/00324720701571806

Kalmijn, M. (2007b). Gender differences in theeffects of divorce, widowhood and remarriageon intergenerational support: Does marriageprotect fathers? Social Forces, 85, 1079–1104.doi:10.1353/sof.2007.0043

Kalmijn, M. (2012). Long-term effects of divorceon parent–child relationships: Within-family com-parisons of fathers and mothers. European Soci-ological Review. Advance online publication.doi:10.1093/esr/jcs066

Kalmijn, M., & de Graaf, P. (2000). Gescheidenvaders en hun kinderen: Een empirische analysevan voogdij en bezoekfrequentie [Divorced fathersand their children: An empirical analysis of custodyand visitation]. Bevolking en Gezin, 29, 59–84.

King, V. (2006). The antecedents and consequencesof adolescents’ relationships with stepfathersand nonresident fathers. Journal of Marriageand Family, 68, 910–928. doi:10.1111/j.1741-3737.2006.00304.x

King, V. (2007). When children have two moth-ers: Relationships with nonresident mothers,stepmothers, and fathers. Journal of Marriageand Family, 69, 1178–1193. doi:10.1111/j.1741-3737.2007.00440.x

King, V. (2009). Stepfamily formation: Implicationsfor adolescent ties to mothers, nonresident fathers,and stepfathers. Journal of Marriage and Family,71, 954–968.

Oudejans, M., & Kalmijn, M. (2013). Life HistoryQuestionnaire. Tilburg, the Netherlands: CentER-data.

Pillemer, K., Suitor, J. J., Pardo, S., & Hen-derson, C. (2010). Mothers’ differentiation anddepressive symptoms among adult children. Jour-nal of Marriage and Family, 72, 333–345.doi:10.1111/j.1741-3737.2010.00703.x

Rusbult, C. E., Martz, J. M., & Agnew, C. R.(1998). The Investment Model Scale: Measuringcommitment level, satisfaction level, quality ofalternatives, and investment size. Personal Rela-tionships, 5, 357–391. doi:10.1111/j.1475-6811.1998.tb00177.x

Sayer, L. C., Bianchi, S. M., & Robinson, J. P.(2004). Are parents investing less in children?Trends in mother’s and father’s time withchildren. American Journal of Sociology, 110,1–43. doi:10.1086/386270

Scherpenzeel, A. (2009). Start of the LISS panel: Sam-ple and recruitment of a probability-based Internetpanel. Tilburg, the Netherlands: CentERdata.

Schnettler, S., & Steinbach, A. (2011). Howdo biological and social kinship play outwithin families in the US? Zeitschrift furFamilienforschung, 23, 173–195.

Silverstein, M. (1995). Stability and change in tempo-ral distance between the elderly and their children.Demography, 32, 29–46. doi:10.2307/2061895

Spruijt, E., & Duindam, V. (2009). Joint physicalcustody in the Netherlands and the well-being ofchildren. Journal of Divorce and Remarriage, 51,65–82.

Stephens, L. S. (1996). Will Johnny see Daddythis week? An empirical test of three theoretical

Adult Children’s Relationships With Parents 1193

perspectives of postdivorce contact. Journalof Family Issues, 17, 466–494. doi:10.1177/019251396017004003

Swiss, L., & Le Bourdais, C. (2009). Father–childcontact after separation: The influence of livingarrangements. Journal of Family Issues, 30,623–652. doi:10.1177/0192513X08331023

Ward, R. A., Spitze, G., & Deane, G. (2009). The morethe merrier? Multiple parent–adult child relations.Journal of Marriage and Family, 71, 161–173.doi:10.1111/j.1741-3737.2008.00587.x

White, L., & Gilbreth, J. G. (2001). When childrenhave two fathers: Effects of relationships withstepfathers and noncustodial fathers on adoles-cent outcomes. Journal of Marriage and Fam-ily, 63, 155–167. doi:10.1111/j.1741-3737.2001.00155.x

World Telecommunication/ICT Indicators DatabaseOnline [CD-ROM]. (2012). Geneva, Switzerland:International Telecommunication Union.