Post-Divorce Dual-Household Living Arrangements and Adolescent Wellbeing

Adolescents are increasingly living in two households, alternating between family contexts. It is timely to consider how these contexts may affect adolescent’s psychological wellbeing. We use data from the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU), England, Germany, the Netherlands, and Sweden, including data on occurrence and extent of dual-household residency, to correlate 15 family types with adolescent’s internalizing problems, self-esteem, and life satisfaction. Analyses show that (i) adolescents in intact families exhibit better wellbeing than peers in different types of dissolved families, (ii) adolescents in reconstituted families exhibit less wellbeing than adolescents living with a single parent only, (iii) living in two households, where both parents are single or either of them is repartnered, is not associated with better wellbeing than living with a single parent only, (iv) adolescents in alternate living generally seem to do as well as their peers in intact families, but (v) there is a tendency that alternate living in a symmetrical family context, i.e., where both parents are either single or living with a new partner, is more positive for the adolescent than if one parent is in a new relationship and the other is not.


Introduction
Following the Second Demographic Transition, European societies face high prevalence of divorce, changing rates of repartnering, and an increased prevalence of post-divorce paternal involvement, shared living arrangements, and shared custody (Amato 2000(Amato , 2010Amato and James 2010;Bernardi et al. 2013;Cancian et al. 2014;Gähler and Palmtag 2015;Kitterød and Lyngstad 2014;Ottosen and Stage 2012;Spruijt and Duindam 2009). The family constellations that children and adolescents live in have grown ever-more complex. Experiencing parental union dissolution and growing up in a complex family may take emotional toll on adolescents trying to navigate multifaceted feelings towards their parents. At the same time, dissolved households often have fewer financial resources and parents face more stringent time constraints, leaving less time and money to invest in their adolescents. Yet, all dissolved household are not necessarily cut from the same cloth. Categories such as single-parent households and reconstituted family households may hide large underlying differences between children who may reside in two homes, with complex combinations of household structures. Failing to account for such differences in living arrangements may obscure important differences in child wellbeing. Yet, when studying how adolescents fare outside a two-biological-parent setting, family scholars have often used parsimonious categories that only differentiate between intact households and one or a couple of alternatives: e.g., singleparent families and reconstituted, or step-, families.
The main barrier to improve our knowledge on how adolescents fare when their parents do not live together has been the lack of good data on complex families. We seldom have information on whether, and to what extent, an adolescent resides in two households. Further, in the rare occasions when these data are available, the structure of both households often remains unknown. This lack of knowledge is unfortunate, because residence arrangements following parental divorce may have substantial bearing on children and adolescents' wellbeing (e.g., Bastaits et al. 2018;Fransson et al. 2016;Poortman 2018) as well as successful transition to adulthood (see Andersen 2017 for review). For example, does living with a single parent in only one household mean the same to adolescents as living with a single parent in one household and a parent in a reconstituted family in another household? Do adolescents in alternate living, i.e., living equally much with both parents in separate households with two single parents, have similar wellbeing as adolescents whose parents both live with new partners?
To address this gap in the literature, our analyses use detailed information on the occurrence and extent of adolescents' multiple household residency. Based on data from the Children of Immigrant Longitudinal Study from Four European Countries (CILS4EU) (see below), we define 15 family categories conditioned on both parents' living arrangements (when applicable). Wave 1 of the CILS4EU data includes information on adolescents' psychological wellbeing and both biological/adoptive parents' living arrangements, covering Sweden, The Netherlands, United Kingdom, and Germany, for representative national classroom-based samples of pupils aged around 14.
Three out of ten adolescents in our analytical dataset do not reside with both their biological parents. Most of those who have experienced a parental separation or divorce still live in only one household, mostly with a single or repartnered mother, but a substantial proportion lives in two households and here alternatives and combinations are many, e.g., two single parents, two repartnered parents, or one of each. To an increasing extent, adolescents live equally much in these households, i.e., there is no longer a main household to which the adolescent belongs. Measures of internalizing problems, self-esteem, and life satisfaction allow us to provide a detailed picture of the relationship between living arrangements and adolescent's wellbeing.
The chapter offers three main contributions to the literature on living arrangements and adolescents' wellbeing. First, by using a more detailed typology of family types, we document the prevalence of 15 different forms of living arrangements across four European countries. Second, we provide descriptive evidence on systematic differences in adolescent wellbeing across the forms of living arrangements. Third, we examine whether, and to what extent, mediating contextual factors and family characteristics can account for initial differences, and show that after controlling for these characteristics, salient differences in adolescent wellbeing remain across living arrangements.

Background
Following the Second Demographic Transition in the developed world, people deferred marriages, cohabitated, had fewer kids later, and broke up more (Lesthaeghe 2010;Van de Kaa 1987). Multiple partner fertility increased the share of children growing up with step-parents and step-siblings (Gähler and Palmtag 2015). Across the period, how children and parents affect each other's lives has changed. Fathers increased involvement and custody and living arrangements where children spend equal amount of time with both parents (alternate living) became common (e.g., Cancian et al. 2014;Kitterød and Lyngstad 2014;Ottosen and Stage 2012;Spruijt and Duindam 2009). Consequently, single-mother families are no longer the default state for children growing up in non-traditional families (Amato 2000(Amato , 2010Amato and James 2010;Bernardi et al. 2013). These children experience multiple family constellations and often live part-time in two parallel family structures. Yet, despite the increase in family complexity, the literature on family heterogeneity struggles with incorporating these changes. Notably, empirical studies of children living in non-traditional households tend to describe only one resident household, thereby not capturing the dual context that increasingly defines many childhoods.
Previous studies consistently show that children from non-intact families are disadvantaged on a variety of outcomes (e.g., Amato 2010), including emotional and psychological wellbeing (e.g., Amato and Sobolewski 2001;Barrett and Turner 2005;Mitchell et al. 2015;Sun and Li 2002), with recent work suggesting that children in single custody households are worse off compared to children who spend equal time with both parents (see Baude et al. 2016;Bauserman 2002;Nielsen 2014 for recent meta-reviews finding weak support). These analyses commonly use simple dichotomies, e.g., intact family versus parental divorce, single parent versus reconstituted family, or alternate living versus sole custody parents, although children of separated or never married parents face an increasing array of family circumstances. One of the most prevalent trends in current family life is that children to an increasing extent move between two households, with varying family compositions. Yet, current knowledge is almost entirely based on information from only one of these households, or on shared living arrangements without including composition of the dual set of involved households (e.g., Fransson et al. 2016;Turunen 2017; see Baude, Pearson and Drapeau 2016 for a recent review). Thus, how this family complexity, and how interactions between dual, and simultaneous, family types, relate to children's wellbeing remains understudied. Recent work has emphasized the need for a better and deeper understanding of how family complexity and heterogeneity, across ethnic groups and socio-economic strata, moderate the impact of parental separation on children's wellbeing and life chances (Amato 2000(Amato , 2010Amato and James 2010;Bernardi et al. 2013;Grätz 2017). Our analyses in the present chapter adds to the ongoing discussion by showing how child wellbeing varies across dual-household family types.

Parental Divorce and Adolescent Wellbeing
Why do children and adolescents from non-intact families on average report lower levels of wellbeing, here defined as a positive view on the self and the absence of internalizing problem behavior? The literature suggests two explanations. First, the emotional turmoil following a divorce may impact children negatively. Children and youth often react to divorce with shock if the divorce is unexpected, anger because the parents no longer live together, grief over the missing family, and regret at the loss of a parent (Öberg and Öberg 1991). Household disruption theory suggests that children may feel abandoned, worry about the future, and blame themselves for the divorce (Hetherington 1979;Pryor and Rodgers 2001;Rutter 1979). Lower wellbeing accompanies such post-divorce uncertainty (Hetherington 1992). Further, children who experience parental divorce have elevated stress-levels, which also may directly lower wellbeing (Evans and Kim 2007).
Second, many children and youth lose resources from parental divorce and from living in a single parent family. Economic conditions often deteriorate for parents following family dissolution (Andreβ et al. 2006), and the increase in family constellations including other than two biological parents with common children have increasingly become concentrated among the less privileged (Gähler and Palmtag 2015;Härkönen 2017), i.e., parents with lower education, class position, and income. The economic and material disadvantage and/or loss is associated with lower psychological wellbeing in children (Gähler and Garriga 2013). Moreover, because single parents are sole breadwinners and caretakers of the household, they may have less time to spend with their children (e.g, Gibson-Davis 2008), which could also affect wellbeing negatively (Del Bono et al. 2016;Milkie et al. 2015). From a resource perspective, divorce may cause parents to have fewer financial and personal resources to invest in children and youth, which in turn lowers wellbeing.

Dual-Household Living Arrangements and Wellbeing
These above stated explanations describe how children's wellbeing differ across intact and non-intact families. Yet, what could we expect from different post-divorce living arrangements? For example, how does family reconstitution affect children? On the one hand, step-parents may add resources to the child's household (Erola and Jalovaara 2017), such as income and help with household tasks. On the other hand, the relation between the step-parent and the child is sometimes problematic (e.g., Mitchell et al. 2015), characterized by tension and rivalry about the child's biological parent. In accordance with this more pessimistic view, most studies show that step-parents cannot compensate for biological parents. Instead, children in stepfamilies have lower wellbeing than children in intact families and sometimes also than children in single parent families (see Coleman et al. 2000;Sweeney 2010 for literature reviews).
During the last decades, alternate living and shared physical custody, where parents have equal rights and children live (almost) equally with both parents, has become increasingly popular in many countries. In these living arrangements, children and youth keep a continuous and close relationship with both parents. Closeness to parents following dissolution can be beneficial for adolescents, (e.g., Booth et al. 2010), who otherwise may miss out of bonding with the noncustodial parent who often is the father (King and Sobolewski 2006). Yet, the child or adolescent also constantly moves between two households, which may be burdensome. Nevertheless, empirical studies show that children with alternate living generally exhibit a higher emotional and psychological wellbeing than children mainly living with a single parent (Nielsen 2014), up to the level of children whose parents remain together (Fransson et al. 2016;Turunen 2017). To some extent this may be because alternate living arrangements are more common among privileged socio-economic groups and because these parents are often better able to cooperate well (Fritzell and Gähler 2017), but even after controlling for income and parental conflict, children seem to be better off in this living arrangement (Nielsen 2014). In a recent meta-analysis, Baude et al. (2016) show that children's wellbeing improves more the closer to 50-50 the child's division of time between the two households comes, and they find a positive association with behavioral and social adjustment but not with emotional adjustment. To our knowledge, however, no previous study on alternate living arrangements has accounted for the heterogeneity in family structure in the child's two households, i.e. whether the parents are single or if one, or both, has initiated a new union partnership.
One reason to study the role of union heterogeneity is that parents' union status (a)symmetry may impact on children's wellbeing, either conditional on or independent of alternate living. Family systems perspective (Hetherington 1992) and the ecology of human development perspective (Bronfenbrenner 1979) both suggest that when parents are not thriving, it affects children too. The symmetrical situation where both parents are single or live in a union may reflect situations where both parents are either satisfied or not actively resenting the other. Asymmetry, where one parent has repartnered while the other remains single may reflect situations more prone to animosity (for example, if a divorce was unilateral because the spouse responsible for terminating the marriage had found a new partner). Thus, symmetric situations are likely easier to handle for the child than an asymmetrical situation where one parent remains single and the other has moved on to a new relationship. These are all dimensions of how living in a non-intact family relates to adolescent wellbeing that we empirically study in this chapter.
Based on the arguments developed above, we expect adolescents in all non-intact dual-household contexts to have lower wellbeing than their peers in intact families. We expect, however, adolescents in alternate living arrangements, symmetrical dual-households in particular, to differ less from intact families than do adolescents with single household living arrangements and whose parents have different types of households. We generally expect family symmetry to trump asymmetry for adolescent's wellbeing. Moreover, we expect the wellbeing of adolescents in reconstituted families to be on par with, or lower than, the wellbeing for adolescents in single parent families. Finally, based on the notion that frequent interaction with both parents following family dissolution is beneficial for the adolescent, we expect living in dual households to be associated with higher wellbeing for adolescents than only living with one parent in one household.

Country Contexts
Our dataset includes four Western European countries, including both countries with a long tradition of easy-access divorce laws (Sweden) and countries with more litigative approaches to divorce (Germany, the UK). In terms of adolescents' risk of ever experiencing parental union separation, the countries covered represent varying extent of children born in unions who can expect to have their parents not residing together at age 15 (Andersson et al. 2017). A dual-household multi-country perspective has so far been absent from the literature, leaving it unclear how well national studies generalize to an international context. Residential complexity could differ across national contexts (cf., Sobotka 2008), not only because of variation in family demography and family law but also because of family policy and social stigma (cf., Amato and James 2010;Dronkers and Härkönen 2008). The primary reason to include multiple countries here is to pool data and reach a sufficient number of cases to be able to perform detailed analyses of complex and less common family types. A secondary reason, however, is to explore whether variations in wellbeing across nontraditional dual-household family types are country specific, or more general for at least Western Europe. Our results show some heterogeneity, but in broad terms the relationship between living arrangements and adolescent wellbeing appears rather similar across countries. It is not within the scope of this chapter to formally test all possible interactions between family type and country.

Data
We use data from the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU), funded by New Opportunities for Research Funding Agency Co-operation in Europe (NORFACE) (Kalter et al. 2014). The overarching goal of the CILS4EU project is to focus on second generation integration. To achieve this goal, a two-step cluster sampling procedure was adopted. First, schools in the four countries were selected, over-sampling schools with a high proportion of immigrant youth, then all pupils in around two classes within each school were invited to participate in the study. In total, 958 school classes in 480 schools participated and the individual pupil participation rate within school classes was generally high but varied between countries (England: 80.5%, Germany: 80.9, Sweden: 86.1, and the Netherlands: 91.1%). During lesson time in the school year 2010/2011, a total number of 18,716 Dutch (n = 4363), English (n = 4315), German (n = 5013), and Swedish (n = 5025) pupils, aged around 14, answered a 45-min self-completion questionnaire on, e.g., family context, educational achievement and aspiration, relations to parents, feelings, attitudes, and beliefs, health, and leisure time activities. The project has a longitudinal design, i.e., pupils were followed over time via repeated questionnaires the following school years, but we only use data from wave 1. 1 Survey data are available at www.gesis.org (ZA5353 data file).

Adolescents in Dual-Household Families
Whereas previously many children lived entirely with one parent (the mother) and spent little (if any) time in the other parent's household following parental divorce or separation, many children today spend a substantial amount of time in both households. An increasing share spends equal time living with their mother and their father (alternate living). The CILS4EU questionnaire acknowledges this new situation by asking the respondent whether s/he lives with both biological parents in one home and, if not, the reason for this. Respondents are also asked who lives in the household (e.g. biological, step-, foster parent, siblings, grandparents, other family members). A unique feature of these data is that they include information on the family context for two households, i.e., pupils are asked whether they live in another home on a regular basis and, if so, how much of the time they live there and who lives there. Based on this information, we are able to identify 15 family categories derived directly from the survey categories: intact family (i), single mother (ii), single father (iii), repartnered mother (iv), and repartnered father (v) (family type i-v: respondent living in only one household), single mother and single father (vi), single mother and repartnered father (vii), repartnered mother and single father (viii), repartnered mother and repartnered father (ix) (family type vi-ix: respondent living in two households but not alternate living), alternate living: single mother and single father (x), alternate living: single mother and repartnered father (xi), alternate living: single father and repartnered mother (xii), alternate living: repartnered mother and repartnered father (xiii) (family type x-xiii: alternate living is defined as living "about half the time" in the second home), foster family (xiv), and other (xv) (e.g., adolescents who have experienced a parental divorce or separation and live alone or with others, not including biological, step-or foster parents) (see Table 16.1 for overview). We excluded three groups of respondents. First, 376 respondents in one German federal state were not asked any specific questions on household members. Second, 936 respondents who have not experienced parental divorce or separation and live alone or with others, not including biological, step-or foster parents. Third, 1100 respondents who were living in non-intact families for other reasons than parental divorce or separation or because their parents had never lived together (e.g., deceased parent(s), parent(s) living/working abroad, or other reason, but also some cases with missing information on reason for living in non-intact family). In total, then, we include a maximum number of 16,304 (18,716-376-936-1100) cases in our analyses. In Table 16.1, we display the family types and how they are distributed by country. Bearing in mind, then, that we only include respondents who either live with both their biological parents or in non-intact families because their parents are divorced/separated or never lived together, we do find a rather striking similarity between the four included countries. The overall proportion of 14 year olds living with both their biological parents is 71% and variation around this figure is small, ranging from 68% for England to 75% for the Netherlands. The most common family type for adolescents not living with both their biological parents is still living with a single mother only, around 10% in our sample, but here we find some variation between countries. Whereas this is common in England and Germany, 12%, it is less common in the Netherlands and Sweden, 7-8%. In these countries, it is instead more common that adolescents from non-intact families live in two households. This is most clearly demonstrated when studying family types where adolescents spend an equal amount of time with their mother and their father in two different households. If we add the four alternate living categories together for Sweden, we find that almost 9% (3.3 + 1.6 + 1.6 + 2.1), or almost every three out of ten (8.6/29.8) of all adolescents from non-intact families live in such an arrangement.

Emotional and Psychological Wellbeing
We are mainly interested in studying how family and alternate living arrangements are associated with adolescent's emotional and psychological status, and we use three indicators measuring this. First, Internalizing problems is based on four questions, "How often are each of these statements true about you?": "I feel very worried", "I feel anxious", "I feel depressed", and "I feel worthless". Responses range from "never true" (0) to "often true" (3). We use these questions to form an index, ranging from 0 to 12, with a Cronbach's alpha of .784. Self-esteem is measured by four indicators: "How much do you agree or disagree with each of the statements?" "I have a lot of good qualities", "I have a lot to be proud of", "I like myself just the way I am", and "I think things will go well for me in the future". Responses range from "strongly disagree" (1) to "strongly agree" (5). We use these questions to form an index, ranging from 4 to 20, with a Cronbach's alpha of .814. Finally, life satisfaction is measured by the question "On a scale of 1-10 where 1 is very unsatisfied and 10 is very satisfied, how satisfied are you with your life in general?" Descriptive statistics for these variables, by country, are presented in Table 16.2. On average, adolescents in the four countries seem to do rather well. Their self-esteem and life satisfaction leans towards the higher end of the scale whereas the presence of internalizing problems is relatively low. English and German adolescents are slightly more likely to report internalizing problems and they express lower life satisfaction than their peers in the Netherlands and Sweden whereas German and Swedish adolescents exhibit a relatively high self-esteem. Although these intercountry differences are interesting (and often statistically significant), the implications of these differences are not within the scope of this chapter.

Control Variables
Besides providing evidence on unconditional differences in wellbeing across country and family types, we also include a series of variables to examine if differences in wellbeing are a result of child characteristics, context, and socio-economic resources rather than family type. As control variables we include survey country (dummy variables for England, Germany, the Netherlands, and Sweden), gender (boy/girl), immigrant status (four dummy variables: respondent born in survey country and both parents born in survey country (i), respondent and one parent born in survey country, one parent born in another country (ii), respondent born in survey country, both parents born in another country (iii), and respondent born in another country (iv)), parents' highest occupational status (according to the 2008 International Standard Classification of Occupations (ISCO-08) and converted into the interval-scale ISEI-08 (Ganzeboom and Treiman 1996;Ganzeboom 2010)) 2 , and immigrant stratum (indicator of proportion of immigrants in respondent's school: 0-9.9, 10-29.9, 30-59.9, and 60-100%). Moreover, we control for family cohesion [summated index of five indicators, e.g. "How often is each of the following true about your home? We feel very close to each other", ranging from "always" (1) to "never" (4)], parental closeness [summated index of seven indicators, e.g. "How much do you agree or disagree with each of these statements? My parents show me that they love me", ranging from "strongly agree" (1) to "strongly disagree" (5)], parental monitoring [summated index of three indicators, e.g. "How much do you agree or disagree with each of these statements? I always need to tell my parents exactly where I am and what I am doing when I am not at home", ranging from "strongly agree" (1) to "strongly disagree" (5)], and parental school encouragement [summated index of three indicators, e.g. "How much do you agree or disagree with each of these statements? My parents show a lot of interest in my grades and my achievement in school", ranging from "strongly agree" (1) to "strongly disagree" (5)].

Results
For all analyses we use OLS regression. We first present results from univariate unconditional OLS models, and then from models where we condition on the control variables discussed above. All parameter estimates are expressed as standardized coefficients expressing parameter sizes in increases in one standard deviation.

Univariate Results
The results from the unconditional analyses do indeed suggest that salient differences in child wellbeing exist across family structures. The findings are presented in Fig. 16.1 along the dashed lines. In the figure, results for the reference category, i.e., respondents in intact families, living with both their biological parents, are set to 0. The plotted symbols show the standardized deviation from the reference category for all other family types. Let us take an example: in the first line of the left panel of Fig. 16.1, for adolescents living with a single mother in one household only, we see that adolescents in both the full sample and in all four countries separately report more internalizing problems than adolescents in intact families, but the association is insignificant for the UK. If we study the other panels, where outcomes are selfesteem and life satisfaction, we again find that respondents living with only single mothers fare less well than respondents in intact families; the former express less self-esteem and lower life satisfaction (although for self-esteem the estimate for the Netherlands is insignificant). In fact, this seems to be the case for most other family types too, regardless of outcome variable and regardless of survey country. Thus, our first conclusion aligns well with most previous research: adolescents in nonintact families generally report lower emotional and psychological wellbeing than their peers in intact families. An exception, however, is adolescents in alternate living who report levels of internalizing problems close to the reference category.

Controlling for Observable Characteristics
As we have demonstrated, adolescents' family type and living arrangements are associated with their emotional and psychological wellbeing. This is not to say, however, that we have established a causal link, i.e. that different family types and living arrangements cause variation in outcomes. Why not? There are at least two possible reasons for this. First, adolescents with a certain emotional and psychological status may select into certain family types. For example, if adolescents with emotional and psychological problems are more likely to see their parents divorce it could indicate a reversed causal order of events. In other words, family arrangements do not affect the wellbeing of adolescents, instead the wellbeing of adolescents affects which type of family they live in. Second, the association between family type and adolescent wellbeing may be spurious. A certain family type may be associated with other, observable or non-observable, characteristics, e.g., socioeconomic conditions, parental conflict, child time with parents and so on, and it may be these characteristics, rather than family type per se, that affect adolescent wellbeing. In other words, although we find an association between family type and adolescent wellbeing there may not be a causal effect. The first of these alternative possibilities, i.e., selection, we are not able to test here as we have only access to cross-sectional data, but the second alternative we can partly account for by controlling for some potentially important observable characteristics. As mentioned, we control for adolescent's gender, survey country, immigrant status, parents' highest occupational status, immigrant stratum, family cohesion, parental closeness, parental monitoring, and parental school encouragement. The results from these analyses are presented across the full lines in Fig. 16.1. By comparing estimates from dashed and full lines, we can estimate how much of the initial differences between family types that can be explained by observable characteristics. We find that the direction of almost all deviations from the reference category remains after adding these controls but in some instances the difference becomes statistically non-significant. For example, this is the case for the repartnered mother, no father family type regarding internalizing problems. After controlling for the background characteristics listed above, adolescents in this family type still exhibit more internalizing problems but the difference in relation to young people in intact families is no longer statistically significant. For most family types, however, and for all outcomes, the distance to the reference category diminishes after controls are added but the statistical significance remains. This suggests that the lower wellbeing among 14 years old in "alternative" family types to some extent could be explained by other factors than their living arrangements. Is the remaining difference caused by family type? Not necessarily, it could potentially be explained by other factors that we were not able to control for here and/or by selection. However, even after controlling for confounding factors, children in these family types are still doing less well in terms of wellbeing.
Adolescents in alternate living generally do well but we find some slight indications of support for an impact of symmetry: alternate living adolescents report lower levels of self-esteem and life satisfaction when their parents do not have symmetric households. Across all three outcomes, adolescents in alternate living between two single parent households report wellbeing at the same levels as peers from intact families. The same goes for alternate living with two repartnered parents, except a borderline significant negative estimate for lower life satisfaction for Dutch adolescents. Estimates for asymmetric alternate living are more dispersed, with several significant parameters, indicating that some negative impact of asymmetry may take place. In general, however, adolescents in alternate living are reporting better wellbeing than peers in other forms of dual-and single-household arrangements, even after controlling for a number of variables. For other non-intact family types, we find no indication that family type symmetry is better than asymmetry for adolescent wellbeing. Adolescents with two single or two repartnered parents do not exhibit better wellbeing than adolescents with one single and one repartnered parent.
In the univariate models, we generally find that adolescents in reconstituted families, living in only one household, exhibit less wellbeing than young males and females living with a single parent only. When we compare the results for single mother families with the results for remarried mothers, and do the corresponding comparison for father families, we find that the comparison falls to the advantage of single parent families in every case except for one, life satisfaction for mother families where we find no difference. These differences disappear, however, in the models where we add controls. More detailed analyses (not displayed) show that this is mainly because adolescents in repartnered families exhibit lower levels of family cohesion and parental closeness.
Going from living with a single parent only, whether the mother or the father, to living in two households, where both parents are single or either or both of them is repartnered, does not seem to improve adolescent wellbeing. In fact, the opposite often seems to be the case. For example, adolescents living with a single father or mother in one household only, exhibit fewer internalizing problems and higher life esteem than adolescents living in two households with two single parents or with one single and one repartnered parent.
Finally, for inter-country variation, we need to be aware that these results are based on much smaller numbers than the aggregate and therefore less reliable. Bearing this in mind, we may tentatively conclude that the conclusions drawn for the aggregate generally seem to be valid also for each of the four countries included, i.e. country differences are for the most part small, and overall patterns are rather similar, although some variation does exist.

Concluding Discussion
Between the rise in divorce rates, the increase in multiple partner fertility, and the changing roles of fathers in children's lives, children and adolescents today experience a diverse range of family constellations. In this chapter, we have defined and documented the prevalence of 15 types of family compositions and shown that they offer very different circumstances for children, at least in terms of wellbeing. Some of our results align well with what has previously been known: adolescents in intact families fare better than their peers in dissolved families (e.g., Amato 2010; Bernardi et al. 2013) and family reconstitution, i.e., step-families, does not improve the adolescent's situation (e.g., Coleman et al. 2000;Sweeney 2010). If anything, it deteriorates it even further. These results indicate that the economic resources that step-parents bring to the family, and the time they potentially free for the biological parent to spend with the child is not sufficient to improve these adolescent's emotional and psychological wellbeing. Either the children are unable to utilize these (potential) resources or the positive effect of the resources are trumped by negative effects associated with family reconstitution. One such possible effect being that family cohesion and adolescent's closeness with the biological parent is weaker, potentially caused by tension and rivalry with the step-parent. Another possible explanation is that loyalty to the other parent may make adolescents less likely to engage positively with a step-parent.
A new finding from this study is that adolescents also do not seem to benefit from changing from a single parent one-household family to a dual-household living arrangement, unless it is alternate living, i.e., the adolescent spending (about) the same time in both households. It remains an open question why only balanced time in both parents' households benefits the adolescent, but one possibility is that alternate living parents are still positively selected regarding, e.g., cooperation around the adolescent whereas this may not be the case for parents dividing the adolescent's living unequally between them. This result implies that a continuous interaction with both parents, even living with them, is not necessarily associated with an improved well-being in the child, regardless of whether these parents initiate a new relationship or not. As suggested by household disruption theory, alternate living arrangement likely occurs in families with less post-divorce conflict, which in turn likely correlates with higher levels of child wellbeing.
Finally, the conclusion that alternate living benefits the adolescent must be nuanced. Our results suggest that this is clearly the case when the two households are symmetrical, i.e., when both parents remain single or when both parents are repartnered. Under these circumstances, adolescents fare as well as their peers in intact families. When households are asymmetrical, however, we find a tendency that alternate living is not as beneficial for the adolescent, in particular if the mother is single and the father repartnered. For adolescents in other dual-household living arrangements, i.e. other than alternate living, we find no variation in well-being according to whether the households are symmetrical or asymmetrical.
Our analyses are not without limitations. First, they are based on cross-sectional data and we are unable to account for any selection. Second, although we were able to control for a number of key observables, we also miss some important controls, e.g., time since union dissolution, step-siblings, and precise time spent in each household (except for alternate living). Some reconstituted families may effectively be the result of family swapping, which could be more detrimental for the wellbeing of children from the previous relationship than a bringing-together-of-two-families type of reconstitution would be. We also lack information on financial resources and parental conflict. Still, our study is one of the first attempts to widen the scope and cover the dual-household context that is reality for so many children and adolescents today.
For future studies, another large question remains: Are the dimensions we use to understand families still adequate, or is there a need to rethink and reconsider them? We have extended on "traditional" family categories by combining two households and considering living arrangements as well. In doing so, we have added more detail by including the non-residential parent as well. Yet, we still rely on the traditional categories being a reliable measure of the experienced living conditions. For example, elements such as child's age at parents' divorce/dissolution, "family swapping" where parents focus more on their new family, age difference between focal child and half-and step-siblings, and geographical distance between parental households could all be (more) salient dimensions affecting child wellbeing above simply examining the composition of both households when children reside outside of a two-biological parents family. The inclusion and consideration of alternate living arrangements likely provides an important first step in updating our considerations on the actual family situation that children experience, yet a substantial amount of empirical and theoretical work is needed still if we truly are to reconsider how to define dual-household living arrangements in a way that captures children's actual lived experience in an accurate way.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.