Keywords

Children’s postseparation family arrangements have radically changed due to the increasing legal and social reliance on shared physical custody (SPC) in a great number of Western countries in the last decade. While such changes have been widely debated in the public sphere, particularly because Australia and some European countries have adopted explicit legal regulations about the implementation of SPC (Schweine 2018), only recently have researchers begun to describe their prevalence and examine their consequences.

Most of the research so far reports that the effects of SPC on children’s well-being are overall positive, but also that parents in SPC arrangements tend to have a higher socioeconomic status than those in sole-custody arrangements. As a consequence, SPC seems to increase inequalities among children of divorce and separation given that children from better-off parents can partially counterbalance the negative effects of parental separation through SPC, while the latter remains relatively inaccessible for disadvantaged children.

Nevertheless, the contribution of SPC to social inequalities among children depends on whether such a living arrangement remain selective or is spread across social groups, and whether, in the latter case, the positive effects recorded among better off children are also observable among children from different social backgrounds. Research has not yet engaged with the diffusion of SPC across social groups and the consequent disparities among children of different social origin. In this Chapter we contribute to the literature by addressing the question on the evolution of shared physical custody and of its social gradient among children of disrupted family.

As the prevalence of SPC arrangements increases in the population, one scenario is that SPC families become more socially heterogeneous and the initial selectivity of those experiencing SPC gradually disappears as the process of social diffusion goes along (diffusion hypothesis). If this is the case, the inequality created by SPC among children of separated families would only be a transitional phenomenon. Children from any background would benefit from the positive effects of SPC over lone parent custody (Sodermans et al. 2013). A second scenario is that the social diffusion of SPC interacts with a parallel increase in the selectivity related to the risk of union break-ups. Recent evidence shows that family instability in most western countries is higher among disadvantaged families and that there has been a reversal of the social gradient of separations (Garriga and Cortina 2017; Härkönen and Dronkers 2006; Kennedy and Thomson 2010). Relatively advantaged parents tend to lead stable and married family lives, while less advantaged ones experience more instances of family dissolution, lone parenthood, and complex family arrangements. In a pivotal work, McLanahan (2004) defined such trends as “diverging destinies”. In a context of diverging destinies and class stratification of family developments, the diffusion of SPC among disadvantaged children will not be able to compensate for their increased risk of family disruptions, and social inequality among children of separated families may persist or even increase despite the increase in SPC. The diverging destinies hypothesis offers a theoretical background for a competitive hypothesis to the diffusion of SPC across social strata. As in the case of family stability, better off parents would remain more likely to adopt SPC so that the higher the prevalence of SPC, the higher the inequality among children will be (diverging destinies hypothesis).

At present, the few scholars that have examined the diffusion of SPC across social groups have produced mixed evidence in support of both the diffusion hypothesis and the diverging destinies hypothesis (Brown and Cook 2012; Cancian et al. 2014; Meyer et al. 2017; Sodermans et al. 2013).

The aim of this study is to expand existing research on the diffusion and social selectivity of SPC, with a focus on two country contexts, Spain and Sweden, that had been understudied as far as SPC is concerned. These countries represent compelling contrasting cases for studying social inequalities related to children’s living arrangements. While both countries have socially stratified family patterns together with a relatively high percentage of SPC families compared to the European average (Solsona and Spijker 2016), they display different levels of generosity of their welfare states and diverging gender ideologies, characteristics that are related to the patterns of children’s postseparation living arrangements (Grunow et al. 2018). Such differences speak in favour of more heterogeneity among SPC families in Sweden than in Spain given that the former context is characterized by a higher level of gender equality. We draw on the 2006 and 2014 comparative cross-sectional data from the Spanish and Swedish Health Behaviour in School-Aged Children (HBSC), to examine whether and to what extent parental socioeconomic status relates to children’s living arrangements in separated families as well as whether and the extent to which this association differs between countries and changes between 2006 and 2014, a period in which the prevalence of SPC greatly increases in both.

1 Competitive Hypotheses on the Social Stratification of Shared Physical Custody

Low prevalence of SPC is correlated with its concentration among upper-class parents, who adopt alternative custody arrangements after separation or divorce. Pioneer studies on SPC have concluded that the higher resources averagely available to more advantaged parents explain social differences in its adoption: higher information levels make them more likely to be aware of the possibility of SPC arrangements when this may be relatively uncommon (Donnelly and Finkelhor 1993); higher financial resources make them able to sustain double residences for their children as well as to engage in the necessary legal procedures to established it under the best conditions; more progressive gender attitudes makes them more likely to have more equal shares of parental responsibilities during union (Sodermans et al. 2013); and higher psychological resources translate into lower levels of interparental conflict which favours adopting SPC (Kitterød and Wiik 2017).

With the growing support of SPC in family law (legal presumption for SPC) and with the generalisation of the principle that SPC arrangements have to be considered as the preferred option in cases of separation and divorce debated in tribunals, scholars argue that we should observe a gradual democratisation of SPC across social strata (Sodermans et al. 2013). The trend towards the diffusion of SPC among lower classes would be reinforced by legislative changes favouring SPC, such as granting judges the ability to impose SPC in cases of custodial disagreement, given that low- and middle-class parents tend to have more disagreement on custodial issues than upper-class parents do.

Beside changes in family law, other important social transformations support the diffusion hypothesis, that is of a decrease in social inequality in the adoption of SPC rather than solo custody (Hypothesis 1a). First, the growing employment rates of lower class women (McLanahan 2004), who may now find SPC more attractive than before because of its higher compatibility with paid work. Evidence shows that when the mother is in the labour force rather than inactive during union (Juby et al. 2005), were she to separate from her partner, SPC would be more likely to be chosen. In addition, disadvantaged mothers who had not been working during partnership, have been increasingly incentivised to get into paid employment after separation, by the growing emphasis on activation and welfare to work policies in many European countries as the preferred way to contrast poverty risk among lone parents (Nieuwenhuis 2017). Second, the diffusion of less traditional gender patterns across social groups may also have affected the diffusion of SPC. For this reason, increases in SPC prevalence can be partially explained by what Hetherington and Kelly (2002) call “divorce-activated fathers”: men who take more active roles in their children’s lives after divorce than they did before. Such activation is supported by the results of a qualitative study from Sweden in which mothers in SPC arrangements reported that the fathers had stronger relationships with their children and improved parenting styles after the dissolution of their unions; these mothers also reported that the care obligations had become more gender-neutral (Fransson et al. 2016). Among non-separated parents, recent trends show that fathers’ involvement has also increased in the general population including in the least advantaged (Dotti Sani and Treas 2016; Sullivan 2010; Sullivan et al. 2014), which may have a positive impact on the increase of SPC among this group. Since, on average, fathers who show little engagement with their children are less likely to engage in SPC after separation than those who are more engaged (Juby et al. 2005; Poortman and van Gaalen 2017; Westphal et al. 2014).

In contrast to the diffusion hypothesis, the competing diverging destinies hypothesis would state that the stratification of family break-ups is likely to increase and this increase translates into an unchanged distribution of sole custody over social classes (Hypothesis 1b). In support of such a hypothesis are arguments related to the increasing economic and labor market inequality among parents and to the socially stratified father’s involvement with their children after separation. First, research in most Western countries has shown that children with lower socioeconomic backgrounds are at a greater risk of living in separated families than other children and this trend has been related to an increase in economic inequality among parents (Cherlin 2018). The arguments go as follows: the increasing polarization between low- and high-paying jobs, especially during the economic recession, and the unfavourable conditions associated to low-paying jobs (precarious jobs, temporary contracts, and unpredictable work schedules) produce ever- growing differences in economic and labour-market conditions between parents of different socioeconomic backgrounds (Smyth et al. 2014). Since financial resources are important to enter a SPC arrangement, economic and labour market inequalities may result in an increasing polarization of post-separation child custody arrangements. Second, since fathers’ involvement is an important predictor of SPC, if this is differentially distributed across social groups it may support a diverging destinies hypothesis. There is evidence that the socioeconomic gradient of father’s involvement is still present and has even increased in some cases to the advantage of higher status children. (McLanahan 2004; Dotti Sani and Treas 2016; Sullivan 2010; Sullivan et al. 2014). In part this is due to the growing popularity of the intensive parenting ideology among relatively advantaged fathers, sometimes even indicating a class status marker (Kalil 2015; Lareau 2003). Since higher involvement before separation is likely to translate into higher involvement after separation, children having more involved fathers are more likely to have parents (and judges) in favour of post separation SPC arrangements. In addition, regardless of father’s level of involvement before separation, fathers with a lower socioeconomic background tend to reduce their involvement after separation (Grätz 2017). Such fathers are more likely than those of higher socioeconomic status to experience additional children with multiple partners after separation and a consequently greater family complexity (Manning et al. 2014; Thomson et al. 2014). Parents’ transitions into new partnerships and new parenting roles is generally associated with reduced father involvement with children from previous unions (Berger et al. 2012; Tach et al. 2010). Between the 1980s and the 2000s, the likelihood of bearing children in multiple partnerships and the socioeconomic differences of this demographic behaviour increased in Norway, Sweden, the United States, and Australia (Thomson et al. 2014).

As mentioned above, only a few researchers have focused on the evolution of SPC among socioeconomic groups. These studies’ results provide mixed support for both the diffusion and diverging destinies hypotheses. Sodermans et al. (2013) used Flemish data from three cohorts of legally divorced couples from 1971 through 2010 to analyse differences in custody arrangements both before and after two significant legal reforms: the introduction of SPC in 1995 and its adoption as the favoured model in 2006. Sodermans et al. (2013) also showed that among divorced cohorts in which SPC was relatively uncommon, SPC was largely restricted to highly educated parents, and that when SPC became more frequent, it increased significantly among average-educated parents. However, in Flanders, the expansion of SPC remained relatively uncommon among parents with low educational levels. Cancian et al. (2014) analysed court records of divorces in the U.S. state of Wisconsin finalized between 1987 and 2008, showing that in the periods 1993–1998 and 2003–2008, parents with higher total incomes were more likely to have SPC than those with lower incomes. Between 1993 and 1998, the association between SPC and income was particularly strong. Between 2003 and 2008, SPC was still more likely to occur in high-income families but the difference was smaller than in the previous decade. Using the same data, Brown and Cook (2012) found a greater prevalence of SPC in the 2000s than in the 1990s among all income categories except the lowest. Later, Meyer et al. (2017) showed that high-income families were more likely than low-income families to adopt SPC in all time periods and also identified a widening gap in custody outcomes between low- and high-income groups.

All in all, the literature seems to suggest that, on the one hand, there is a tendency towards the diffusion of SPC while on the other, the diffusion process seems to slow down among the most vulnerable families; those with the lowest socioeconomic status. In other words, the diffusion process has positively affected a substantial number of separated families; nonetheless, SPC often remains inaccessible for the least advantaged group.

2 Limitations of Existing Research on SPC Social Stratification

Despite the growing relevance of SPC and the open questions about its diffusion across social groups, there are at least three important limitations in the still scant research on SPC social stratification and its development: (a) the focus is on married couples rather than cohabiters; (b) no comparisons between various types of separated families and intact families exist; and (c) the lack of comparative studies hinders identification of the economic and institutional contexts’ moderating role in the relationship between custody and children’s outcomes.

The lack of consideration for cohabiters is problematic for two reasons. Firstly, excluding children from cohabiting couples excludes a substantial number of children who have experienced parental separation (Meyer et al. 2017; Sodermans et al. 2013). Cohabitation, which is increasing in all Western countries, remains more likely to end in separation than marriage (Kiernan 2004). Secondly, childbearing within cohabitation is more common among parents of low socioeconomic status than among those of high socioeconomic status; consequently, divorced parents tend to have higher socioeconomic status than formerly cohabiting parents (Castro-Martín and Seiz 2014; Garriga et al. 2015; Kennedy and Thomson 2010; Perelli-Harris et al. 2010). For these reasons, excluding cohabiting couples from the analysis results in the exclusion of a disproportionate number of disadvantaged couples, who are the likeliest to cohabit and then separate. Therefore, studying the social diffusion of SPC only among children of divorce while excluding children of separated cohabiters risks creating the misleading perception that SPC is diffused equally across socioeconomic statuses.

The second limitation in the studies on SPC’s diffusion across social strata is that their authors have compared the characteristics of couples who have adopted a SPC arrangement to those who have chosen sole custody, without considering couples who do not separate. This lack of a comparison with two-parent families implies not considering important findings from the diverging destinies literature, in which scholars have noted the growing polarization of separation risks and a related worsening of socioeconomic conditions among those who separate irrespective of if they are a lone parent or have SPC. Thus, to understand the dynamics of each type of separated family, it is necessary to compare these groups with intact families.

The third problematic limitation in the SPC literature is its lack of comparative studies across welfare contexts. Regarding parental divorce’s effects on children, researchers have explored the extent to which these effects diverge across countries with different levels of generosity towards families with children (Hampden-Thompson 2013; Hampden-Thompson and Pong 2005). However, no scholars have investigated the correlation between the welfare state and the effect of parental socioeconomic status on SPC arrangements.

A welfare state’s generosity and ideology can moderate the relationship between parents’ socioeconomic status and the probability of shared custody in at least two ways. First, given that parents with SPC must have sufficient resources if they are to accommodate their children in separate households (Melli and Brown 1994), SPC is often more expensive than exclusive custody (in absolute terms). SPC entails double the housing costs and higher related expenditures, including for utilities, household furnishings, play and study spaces, and toys and play equipment. These costs are not significantly reduced when a child spends considerable time with both parents. The economic well-being of mothers with SPC decreases more than that of mothers with sole custody, despite the latter being economically worse off in absolute terms (Bartfeld et al. 2012). In fact, evidence indicates that most fathers in SPC arrangements do not pay child support, which is not the case for noncustodial fathers. For example, in Sweden, 79.7% of sole-custody parents receive child support, as compared to only 10.1% of those who have equal physical custody (Statistics Sweden 2014). Thus, less advantaged mothers may prefer to have sole custody (and receive child support) rather than engage in SPC arrangements. Although a proportion of less advantaged fathers may prefer SPC for purely economic reasons (e.g., to pay less child support), for many of them, the costs of SPC - such as providing extra rooms for children - are greater than the savings in child-support payments due to SPC. This may be why most separated parents, who have relatively few resources adopt sole-custody rather than SPC arrangements. Therefore, it is reasonable to assume that welfare states’ generosity can moderate the social gradient of custody arrangements by providing parents of low socioeconomic status with enough resources to maintain separate households that are sufficiently equipped to share child custody. Thus, less advantaged parents in countries with more generous family policies have more chances to engage in SPC than do those in countries where such policies are scarce.

Another possible mechanism behind the welfare state’s influence on SPC’s diffusion across social strata is the existence and degree of generosity of policies that promote gender equality. Several pieces of evidence support this statement. Firstly, as mentioned, fathers who show little engagement with their children are, on average, less likely to engage in SPC after separation than those who are more engaged (Juby et al. 2005; Poortman and van Gaalen 2017; Westphal et al. 2014) and, among partnered fathers, those with a lower socioeconomic status tend to be less involved in their children’s lives than those with a higher status (Dotti Sani and Treas 2016). Secondly, some scholars have found that in countries where family policies encourage both mothers and fathers to engage in paid work and to care for their children, fathers’ involvement is more homogeneously spread across the social strata than in countries where such policies do not exist (Gracia and Ghysels 2017; Sayer et al. 2004). A possible explanation of this finding is that less advantaged fathers have less time due to the competing demands of paid work and child-care and, in countries where the state eases external time demands through policies, fathers at all education levels may have more time to spend with their children. In addition to that, the fact that most mothers from all educational levels work in these countries may force fathers to take care of their children. For all of these reasons, if father’s involvement is one of the key predictors of SPC and is socially stratified and policies are able to reduce the social gradient of father’s involvement, it is reasonable to hypothesize that in countries with generous gender-equality policies, socioeconomic status may have less influence on the SPC arrangement than in countries where such policies are non-existent or scarce.

Overall, in order to overcome the limitations presented here, our study does not only include previously married couples but all children including those from previously cohabiting unions. We compare SPC families not only with lone mother families but also with two –parent families. Moreover, we use a comparative perspective to study the evolution of inequality in SPC, comparing Sweden and Spain.

3 Similarities and Differences Between Spain and Sweden as Contexts for SPC

The concurrent evolution of families, separation and SPC in Spain has been very different than in Sweden. The latter is considered as a model of the second demographic transition. Historically, Spain has been regarded as a European country with a particularly great emphasis on the traditional family. Nonetheless, the family situations in these countries are now surprisingly similar, especially concerning the prevalence of divorce, separation, and SPC.

Since the 1960s, Sweden has had one of the highest divorce rates among Western societies. In 1960, the crude divorce rate (the number of divorces per 1000 married women) in Sweden was 1.2 (compared to France, for example, which had a much lower level of 0.7); since the 1980s, the crude divorce rate in Sweden has hovered around 2.4. The trend of less educated couples divorcing more often than relatively highly educated ones also started earlier in Sweden than in most other countries. This tendency emerged around 1980 and became firmly established by 1990 (Hoem 1997). By contrast, Spain was one of the last European countries to enact divorce (in 1981). During the 1980s and 1990s, the number of divorces in Spain was low, as it was in other Mediterranean countries. For example, the 1990 divorce rates in Spain and Italy were 0.5 and 0.6, respectively (Eurostat 2021). However, starting in the 2000s, Spain displayed a unique family development path among Southern European countries, as it moved rapidly from that traditional model to a Nordic family model (Moreno and Marie-Klose 2013). In 2005, Spain’s divorce law changed from one of the most restrictive in the world to one of the most liberal (Garriga et al. 2015); by 2014, the divorce rates in Spain and Sweden were far closer, at 2.2 and 2.9, respectively. However, Spanish society has not only witnessed a rapid increase in divorce and lone-parenthood rates; starting in the 2000s, it has also experienced a reversal of the educational gradient for these phenomena from positive to negative (Garriga and Cortina 2017). Traditionally, divorce and sole parenthood were more common among highly educated Spaniards, but today, they are more common among those with low education levels.

Spain and Sweden currently have among the highest rates of SPC within the European context (Solsona and Spijker 2016), even though SPC has evolved very differently in these two countries. Among European countries, Sweden has the highest rate of children in SPC (Bjarnason and Arnarsson 2011). The SPC rate there was 1% in the mid-1980s, increasing to 4% at the beginning of the 1990s; it reached 28% in 2006 and 35% in 2014 (Statistics Sweden 2014)Footnote 1. Swedish legislation has promoted voluntary agreements between parents on custody and contact issues after a union dissolution (Heimer and Palme 2016). Since 1992, a legal presumption of shared custody of children has existed for both cohabiting and married parents; thus, this is the default option unless the parents or a court decide otherwise. In a custody dispute, the courts can decide to apply either shared legal or physical custody against the will of one of the parents as long as the parents do not have severe problems that impede cooperation (Schiratzki 2008). Importantly, however, the Swedish legislation on family issues does not have a presumption of SPC, only legal custody. The legal situation and evolution of SPC in Spain is quite different than in Sweden. Until the divorce reform of 2005 this custody arrangement was only briefly mentioned. However, 5 years later, SPC became the default judicial recommendation in some Spanish autonomous communities and from 2013 the doctrine of the Constitutional Court is that SPC must not be exceptional but normal and even desirable. Following these legal changes, the SPC rate of all custody sentences rose from 10% in 2007 to 34% in 2018 (Spanish National Institute of Statistics 2019).

Despite the high rates of SPC in both Spain and Sweden, the existing differences in these welfare states could affect the diffusion of SPC across socioeconomic strata. Firstly, as mentioned, as compared to mothers who have sole custody, mothers with SPC receive less child support from fathers and are therefore more dependent on the labour market and on state financial support. On the one hand, work–family reconciliation policies are necessary for lone mothers to obtain and sustain employment as they have to reconcile work and family without a partner who can share these responsibilities. Sweden’s reconciliation policies are more developed than Spain’s (Crespi 2007). This could explain why the employment rate among lone mothers is higher in Sweden (74.8%) than in Spain (58.9%; OECD Family Database 2014). Additionally, in Sweden, lone mothers receive more financial support from the state than those in Spain do. For example, in Sweden, the state’s contribution to single parents’ income, in the form of cash transfers, is around 40%, but in Spain, it is less than 10% (Bradshaw et al. 2018). The differences between the policies of the Swedish and the Spanish Welfare State can be found not only in their generosity but also in their ideology. In Sweden, child custody laws are the result of policymakers’ desire to increase gender equality in family life. Since the 1970s, policymakers have worked to implement family policies, such as individual taxation of married couples and gender-neutral parental leave (Schiratzki 2008). These laws and policies are meant to enforce married and cohabiting fathers’ care obligations (Bergman and Hobson 2002). In Spain, custody laws have not been accompanied by other family policies oriented towards increasing gender equality. It is therefore reasonable to think that separated families of low socioeconomic status will enjoy better economic conditions and greater father involvement in Sweden than in Spain and therefore that SPC should be more evenly spread across social strata in Sweden than in Spain, despite the increasing SPC rates in both countries (Hypothesis 2).

4 Data and Methods

We drew on Spanish and Swedish cross-sectional survey data from the 2006 and 2014 HBSC surveys, which the World Health Organization carries out every 4 years (Currie et al. 2008; Inchley et al. 2016). The HSBC includes a sample of adolescents, aged 11–16 years, from more than 40 countries across Europe and North America. Its self-completed questionnaires are administered in schools.

The Swedish sampling relied on a two-step cluster design. First, schools were randomly selected; then, for each chosen school, the students in one class were randomly selected to answer the questionnaire. For the Spanish sample, schools were selected using multistage stratified random sampling, taking into account age, region (i.e., autonomous community), school site (rural or urban), and type of school (public or private). The basic sample unit for the Spanish sample was also a class.

The response rates for the Spanish survey were 82%Footnote 2 in 2006 (Simões et al. 2012) and 83% in 2014 (Simões et al. 2018); the rates for the Swedish survey were 85% in 2006 (Swedish National Institute of Public Health 2006) and 69.4% in 2014 (Public Health Agency of Sweden 2015). The final Spanish samples comprised 15,559 adolescents in 2006 and 13,828 in 2014; the final Swedish samples comprised 4000 adolescents in 2006 and 6994 in 2014. We included dummy variables for the missing cases of the independent variables.

In the models, we used indicators of family type, parents’ socioeconomic status, and control variables; these indicators are shown in Table 9.1 and are described below in more detail.

Table 9.1 Descriptive percentages of the Spanish and Swedish 2006 and 2014 samples
Full size table

4.1 Family Structure

The family-structure variable groups adolescents into three categories: those in (a) two-parent families, (b) SPC families, or (c) one-parent (lone-motherFootnote 3) families. The questions about family structure on the 2006 and 2014 Swedish surveys and on the 2006 Spanish survey were identical:

  • “With whom do you live in the home that you live all or most of the time?” (mother, father, or another person)

  • “Do you have another home?” (yes or no)

  • “How often do you stay in the second home?” (half the time, regularly but less than half the time, sometimes, or almost never)

  • “With whom do you live in the second home?” (mother, father, or another person)

The children who lived with both parents in one household were defined as living in a two-parent family. Those who lived approximately half the time with one parent and half the time with the other parent in separate homes were defined as having an SPC arrangement. Finally, those who lived only or primarily with their mother were defined as having a lone-mother arrangement. The 2014 Spanish survey also includes a question with several descriptions of family situations as part of the family-structure variable. Adolescents who selected the choice I live with my mother and father were considered to be in a two-parent family. Those who selected I live half of the time with my mother and half of the time my father were classified as being in a SPC family. Those who indicated that I live with my mother and stay with my father less than half the time were considered to be in a lone-mother family.

Data on parental occupation can be difficult to collect from young people because they often do not know or are not willing to reveal such information, which results in a high nonresponse rate (Currie et al. 1997, 2008; Molcho et al. 2007; Wardle et al. 2002). Given adolescents’ difficulties in answering direct questions about their families’ socioeconomic status, we instead measured perceived family affluence by means of the Family Affluence Scale II (FAS-II). The FAS-IIFootnote 4 includes items that reflect a family’s material resources, patterns of consumption, and purchasing power (Currie et al. 2008). The resulting score is based on four items:

  1. 1.

    “Does your family own a car, van, or truck?” (no [0]; yes, one [1]; or yes, two or more [2])

  2. 2.

    “Do you have your own bedroom for yourself?” (no [0] or yes [1])

  3. 3.

    “During the past twelve months, how many times did you travel away on holiday (vacation) with your family?” (not at all [0], once [1], twice [2], or more than twice [3])

  4. 4.

    “How many computers does your family own?” (none [0], one [1], two [2], or more than two [3]).

A score of 3 on the third and fourth questions (on vacations and computers) was considered as 2. Therefore, the total FAS-II score has values ranging from 0 to 7. Scores of 0 through 3 are low, those of 4 or 5 are mid, and those of 6 or 7 are high (Due et al. 2009). We merged the low and mid categories because not enough cases of the former existed among SPC families to perform an analysis. For this reason, we coded the final variable as low or mid (1) or high (0).

The children’s gender was coded as boy (0) or girl (1). We used two questions (“In which country was your mother born?” and “In which country was your father born?”) to measure foreign background, giving the value of 0 if both parents were born in the country of the sample (Spain or Sweden) and 1 if at least one parent was not born in that countryFootnote 5. We also controlled for the children’s age group (11–12, 13–14, or 15–16).

We analysed the data using multinomial logistic regressions with three categories of family type (two-parent household, sole custody, or SPC) as the dependent variable. This statistical methodological approach is justified by the choice of a dependent variable with three nominal (unordered) categories. Initially, we analysed each country separately; we then combined the data for both countries to analyse the potential country-interaction effects. To adjust for the sampling design, we weighted the results—with the exception of those comparing Sweden in 2006 and 2014, which we did not weight because the 2006 Swedish sample lacked weights.

To test our hypothesis about the evolution of SPC among socioeconomic groups (Hypothesis 1), we use separated multinomial logistic regression models for Spain and Sweden. For each country, we perform a model for 2006 data and another model for 2014 data in order to observe to what extent the magnitude of the effect of family affluence on custody arrangements has changed over time. Then, to test if the observed change in the coefficients of family affluence is significant or not, we pool data for both survey years and add the main effect of survey year and an interaction term between family affluence and survey year to the model. Finally, to test differences in the relationship between family affluence and custody arrangements between Sweden and Spain (Hypothesis 2), we pool 2014 Swedish and Spanish data and perform a multinomial logistic model that includes family affluence, survey year and an interaction term between family affluence and survey year alongside control variables.

5 Results

5.1 Descriptive Results

Tables 9.2 and 9.3 present the categories of family type according to family affluence for Spain and Sweden, respectively. For Spain in 2006, the prevalence of two-parent families was greater among highly affluent families than among less affluent ones. These differences were larger in 2014. In 2006, among separated families, the most affluent ones had a higher percentage of SPC than the least affluent ones. These differences also existed in 2014, albeit to a lesser extent.

Table 9.2 Prevalence of family arrangements in Spain by family affluence (percentages)
Full size table
Table 9.3 Prevalence of family arrangements in Sweden by family affluence (percentages)
Full size table

In Sweden, the differences in the prevalence of two-parent families based on family affluence were greater than they were in Spain for both years. Highly affluent families, relative to less affluent families, were more likely to have the two-parent setup. In 2006, the percentage of SPC families was twice as high for very affluent families than for less affluent ones. In 2014, this difference was smaller but was still very large. In 2006 and 2014, among families without two parents, Sweden has a higher proportion of SPC families than Spain did for all levels of family affluence.

Nevertheless, these descriptive results do not rule out the possibility that the differences observed between custody arrangements and family affluence could be explained by differences in socio-demographics. Hence, we performed multinomial logistic regressions to test the effect of family affluence while controlling for these other characteristics.

5.2 Associations Between Adolescents’ Living Arrangements and Family Affluence in 2014 and 2006

Spain

Table 9.4 shows that in 2006, adolescents in less affluent families had a lower probability of living in a SPC family than in a two-parent family (Model 1); however, the effect of family affluence was not significant in 2014 (Model 2). The interaction between year and family affluence was positive and significant (Model 3), indicating a trend towards reduced socioeconomic differences between these family types. By contrast, Models 4 and 5 show that adolescents with low or mid family affluence had a higher probability of living in lone-mother families than in two-parent families for both 2006 and 2014. The positive and significant interaction of Model 6 indicates that socioeconomic differences between these family types increased during this period. Models 7 and 8 show that adolescents from less well-off families were more likely to live in lone-mother families than in SPC families for both 2006 and 2014. Last, the effect of family affluence on the probability of living in a lone-mother family (as opposed to in a SPC family) decreased between these years because the interaction between year and FAS-II score was negative and significant (Model 9).

Table 9.4 Multinomial logistic regression of family arrangements in Spain (coefficients)
Full size table

Focusing on the control variables’ effects, Models 1 and 2 indicate no significant differences in the probability of living in a SPC family, relative to living in a two-parent family, based on gender, foreign background, or age of the child. However, control variables had notably different effects on the probability of living in a lone-mother family compared to living in a two-parent or SPC family. Adolescents with a foreign background had a higher probability of living in a lone-mother family than in a two-parent family for both years. Some demographic variables had different impacts in each year. In the 2014 data, older adolescents have a higher probability than younger adolescents of living in a lone-mother family than in a two-parent family or SPC. Similarly, girls were more likely to live in a lone-mother family than in a two-parent or SPC family in 2014. The effect of foreign background on the probability of living in a lone-mother family rather than in a SPC family was significant in 2014 but not in 2006.

Sweden

The multinomial models from Table 9.5 show that in Sweden, adolescents in SPC families were no more affluent than those in two-parent families in both 2006 and 2014; there were no significant between-year differences in the effects of family affluence (Model 3). Less affluent adolescents, however, had a higher probability of living in a lone-mother family than in a two-parent family in both years (Models 4 and 5). Model 6 shows that the interaction between FAS level and year was insignificant, which indicates that the effect of family affluence is stable over time. Adolescents in low- or mid-FAS families had a higher probability of living in a lone-mother family in both 2014 and 2006 (Models 7 and 8), and the differences between years were not significant (Model 9).

Table 9.5 Multinomial logistic regression of family arrangements in Sweden (coefficients)
Full size table

Models 1 and 2 show that in both 2006 and 2014, the probability of living in a SPC family rather than a two-parent family did not differ by gender or age of the child. Adolescents with foreign backgrounds had less risk of living in a SPC family rather than in a two-parent family in 2006, but there was no such difference in 2014. In both years, compared to younger adolescents, older ones had a higher probability of living in a lone-mother family than in a two-parent or SPC family. Girls were more likely than boys to live in a lone-mother family than in a two-parent family in 2014. Adolescents with a foreign background were more likely to live in a lone-mother family than in a SPC family in 2006, but there were no significant differences by foreign background in 2014.

Comparing Spain and Sweden

As shown in Model 1 of Table 9.6,in 2014, the interaction term between family affluence and country is significant which indicates that there were no significant differences by country in the probability of living in a SPC family rather than in a two-parent family (Model 1). Model 2 of Table 9.6 indicates that adolescents of low or mid affluence have similar chances of living in a lone-mother family relative to living in a two-parent family in both Sweden and Spain. In addition, Model 3 of Table 9.6 shows that family affluence’s effect on the probability of living in a lone-mother family rather than in a SPC family did not differ by country.

Table 9.6 Country-interaction models of multinomial logistic regression for family arrangements in 2014 (coefficients; weighted sample)
Full size table

6 Discussion

This study contributes to the emerging literature on the diffusion of SPC across social strata, by considering children from married and cohabiting unions and by comparing two contexts, Sweden and Spain, with different prevalences of SPC and with diverging social and gender policies. The results suggest that despite the fact that SPC correlates with positive outcomes for children, this effect is unevenly distributed across social strata both in Spain and Sweden. As already demonstrated in the literature, SPC arrangements are more frequent among parents with higher socioeconomic status and sole-custody arrangements among other parents. More original are the results related to the testing of the two competing hypotheses about the evolution of inequality in post-separation childcare arrangements; the diffusion hypothesis and the diverging destinies hypothesis. Our study clearly shows that the evolution of inequality in post-separation arrangements differ in the two countries. In Spain, we found evidence in favour of the diffusion hypothesis (Hypothesis 1a), with increases in the prevalence of SPC going hand in hand with the diffusion of SPC across social strata. By contrast, the Swedish data support the diverging destinies hypothesis (Hypothesis 1b), with increases in SPC producing no variation in its social stratification over time. This latter result was surprising as we had expected (Hypothesis 2) that SPC would be more widespread across social strata in Sweden than in Spain, given the greater degree of promotion of gender-equal parenting in Sweden and given the greater generosity of the Swedish welfare state towards broken-up families. However welfare state and gender regimes do not seems to make a difference in the association of family affluence and SPC.

How to explain such results? One possibility is to turn towards the countries’ legal systems and the way in which it shapes living arrangements opportunities and costs. In Sweden, there is no legal presumption of SPC, and most custody arrangements are decided by the parents outside of the court system and without the advice of a legal counsellor. This leads to the high social selectivity of such arrangements. In a report from Statistics Sweden (2014), only about 4% of respondents said that their decisions were based on “legal advice” (including court decisions and advice from privately hired divorce lawyers). An additional 7% applied advice from social services, whereas over three quarters of parents reported having decided on their own. In Spain, by contrast, some autonomous communities have a legal presumption in favour of SPC while all other cases are decided through a tribunal procedure and the Spanish courts seem much more frequently involved in custodial decisions than the Swedish courts are. The intervention of judges and a legal framework pushing for the implementation of SPC may be more central than the generosity of the welfare state in the accelerated expansion of SPC across social strata.

Another possible explanation of the support for the diverging destinies hypothesis in Sweden and for the diffusion hypothesis in Spain is that our data are capturing different periods of the evolution of SPC in each country. The Spanish HBSC data were collected shortly after the moment in which SPC was included in the legislation, whereas the Swedish data was collected when this living arrangement was already relatively widespread. Since the prevalence of SPC is lower in Spain than in Sweden both in 2006 and 2014, the two contexts are at different stages of the phenomenon. It is well possible that social diffusion is typically related to the first years of strong implantation of the legal framework but it then slows down after having reached a critical threshold. In order to test for such possibility, one shall rather compare Spain in 2006 and 2014, with Sweden in the years in which SPC was at similar levels. Future research may test the diffusion hypothesis in Spain when prevalence rates are higher and similar to Swedish ones nowadays.

This research has limitations that need to be acknowledged in order to correctly interpret the results. First, while the HBSC is the only available data set that allows for comparative research of SPC families, its characteristics limit the analyses in several ways. The first limitation concerns the FAS indicator. Ideally, one would prefer to measure the socioeconomic status of adolescents in SPC by differentiating each parent’s household. Such households should be modelled separately because these children are living in at least two households with potentially different levels of resources. However, the FAS questionnaire of the HBSC implicitly assumes that all adolescents can only refer to one household in their responses (e.g., “Does your family own a car, van, or truck?”). This makes it impossible to determine whether the children are referring to one household or are pooling the resources of both. As an example, one of the questions from the FAS-II is: “during the past twelve months, how many times did you travel away on holiday (vacation) with your family?”. For adolescents from SPC households, we do not know exactly with whom they travelled: with their mother, their father, or both. For this reason, we cannot rule out the possibility that this study’s results would differ if we could have used FAS-II results for each household where the adolescents in SPC arrangements live.

Another limitation of this study’s data is the impossibility of distinguishing among the various categories of socioeconomic status; previous researchers have come to different conclusions regarding children at the low and middle income levels. Additionally, this study only has one subjective measure of socioeconomic status. Previous scholars have used objective indicators of parents’ socioeconomic status such as education or income, which parents answered. However, we used family affluence as perceived by the adolescent. This single subjective source may not comprehensively capture the socioeconomic conditions of the adolescent’s family. The HBSC database offers information on both the father’s and the mother’s occupation, but it has a high number of missing cases – especially among adolescents who live in lone-mother families – so we had to leave this indicator out. Moreover, despite evidence from the sociological literature that parents’ education is one of the most important indicators in terms of both children’s well-being and changes related to family structure (e.g., a higher rate of separation among less educated people), the HBSC’s international data file does not include questions regarding parents’ educational level, with the exception of the Spanish survey in 2014. The database also includes no information about income. Another limitation of the data is that SPC can be defined only as a perfectly equal share between parents (children living half time with each parent). Less restrictive forms of SPC (30–70 shares) may be less costly (economically and practically) for parents and therefore, the diverging results by social strata are rather intensified here. If equal sharing is likely to be more constraining for parents, still some form of sharing is qualitatively different from an arrangement where the child has only visiting rights toward the non-resident parent.

All in all, this study’s findings imply a need for more research on the prevalence of SPC across social strata comparing different countries; in addition, they show that cohabiting couples should be included in the sample. Future researchers should overcome this study’s acknowledged limitations by using both objective indicators (e.g., income, education, and occupation) and subjective ones to measure parents’ socioeconomic status (with responses from both parents and children). Scholars should also consider the differences in socioeconomic status between parents’ households for children in SPC arrangements. Furthermore, these future investigations should focus on explaining the differences between countries.

7 Conclusion

Despite this study’s limitations, its findings raise several important questions regarding the effects of inequality on children’s well-being. In Sweden, taking into account the results from both this study and from the diverging destinies literature, we affirm that children from lower socioeconomic strata (as compared to those from higher strata) have a greater probability of parental separation and a lesser probability of living in SPC. In other words, there are two unequal family patterns, both before and after separation. Each pattern may be reinforcing (or helping to create) diverging destinies for children under the second demographic transition.

On the other hand, in the Spanish case, the unequal prevalence of SPC across socioeconomic strata seems to be decreasing. This does not mean, however, that the benefits of SPC are equally diffused to all children. In fact, it means quite the opposite. There are reasons to suspect that SPC may be positive for children of high socioeconomic status but not for those of low socioeconomic status. For less advantaged parents, the conditions of SPC may not be ideal, as it is more expensive than sole custody; these parents also tend to have more conflict in their relationships than more advantaged parents do. The research on SPC’s effects on children’s well-being, thus far, has only focused on average effects. No researchers have yet checked the differential effects of SPC across socioeconomic strata. Conducting such studies is very important to determine the real effects that the diffusion of SPC has on less advantaged families.

The results of this study indicate that the current debates on the diverging destinies of children and on SPC’s diffusion and benefits for children are not separate; rather, they must be analysed together. This research is of crucial importance because several European countries are currently debating the possibility of adopting SPC as a default system.