1 Introduction

Concurrent with rising divorce rates, family law in Western countries altered radically during the past decades. What was in common of these legislative adaptations was their direction towards parental neutrality. With the decay of the institutional marriage and the increasing number of children born in consensual unions, parenthood became detached from the conjugal union (Van Krieken 2005; Jensen 2009). In several countries, the principle of joint parental responsibility was legally implemented. In Belgium, the legislator adopted this principle in 1995 and additionally promoted a legal recommendation for joint physical custody in 2006. Consequently, more and more Belgian children live a substantial (and equal) amount of time with their both parents after divorce.

Earlier research on post-divorce custody arrangements usually focused on children because the interest in the child is primordial in most divorce legislations (Amato 2000; Amato and Keith 1991; Brown 2006). Bauserman (2002) showed in his meta-analytic review that joint physical custody enables frequent contact between children and their both parents which can buffer detrimental effects for children caused by the divorce itself. However, also parents are supposed to be affected by the custody arrangement of their children. Because mothers used to be the main caregivers, gender neutral parenting arrangements have given fathers the opportunity to become more involved in their children’s lives, thereby increasing their levels of well-being (Amato and Gilbreth 1999; Amato et al. 2009; Bauserman 2012; Eggebeen and Knoester 2001). For divorced mothers, joint custody means sharing child-rearing tasks, which decreases their parental burden and increases their time to engage in personal leisure activities (Bauserman 2012; Botterman et al. 2013). Despite these implications for the parental life, earlier research on post-divorce custody arrangements has largely ignored the parental perspective.

This study wants to contribute to the current custody literature in different ways. Firstly, it focuses on the association between the custody arrangement and parents’ subjective well-being. Secondly, a social exchange framework is applied to assess this association, meaning that the relationship between the custody arrangement and parents’ well-being is conceptualized in terms of costs and rewards. More particularly, we investigate whether this relationship is mediated by the amount of parental leisure time and the level of parental involvement. Thirdly, our data is collected in Flanders (the northern part of Belgium), a region that represents an excellent case study for studying custody arrangements. Belgium has one of the highest divorce rates in Europe and is a pioneering country with regard to the legal recommendation of joint physical custody, which has led to a relatively high incidence of joint physical custody arrangements (Sodermans et al. 2013b). The ‘Divorce in Flanders’ survey provides us with excellent data to study consequences of post-divorce living arrangements for the well-being of parents. We focus on three different custody types: joint, sole and no custody parents.

Five sections follow this introduction. The second section starts by conceptualizing subjective well-being, the main outcome variable, as a multi-dimensional concept. This is followed by a theoretical explanation via the social exchange theory to understand the association between custody arrangements and parents’ well-being. In the methods section, we describe the Divorce in Flanders dataset and the operationalization of the variables. The next section describes the results of the structural equation models. Finally, the conclusion section summarizes the research results, highlights recommendations for future research and discusses the limitations of the current study.

2 Subjective Well-being: A Multi-Dimensional Concept

Subjective well-being (hereafter: well-being) is an umbrella concept that covers “a broad category of phenomena that includes people’s emotional responses, domain satisfactions, and global judgments of life satisfaction” (Diener et al. 1999, p. 277). Well-being relates to broader life domains like physical health, economic success and social relationships (Diener 2012; Lyubomirsky et al. 2005; Pressman and Cohen 2005). Well-being is important for the general welfare of societies. Like trust, it is seen as having several positive outcomes for the social, economic and political system within societies and communities. In this way, it forms the basis of a well-functioning, stable and flourishing democracy (Fukuyama 1995).

Well-being can be defined by an affective and an evaluative component. The first relates to how people feel on a day-to-day basis, the latter denotes how satisfied people are with their life (Diener 2012; Ryan and Deci 2001). The two elements of well-being are considered to be mutually interwoven (Amato 2005). The affective component of well-being is shaped by both positive and negative feelings and emotions (Diener 1994). The negative indicator of emotional well-being addressed in this study is feelings of depression. It represents a good overall indicator of poor mental health, correlates highly with other negative feelings of anxiety and anger, and is closely linked to feelings of guilt, shame and loneliness (Kalmijn 2009; Ross and Mirowsky 1999). The positive indicator of emotional well-being addressed in this study is self-esteem. This indicator reflects the positive feelings someone has about himself or is the perception one has of his own self-efficacy (Amato 2005). The evaluative component of well-being is life satisfaction. It refers to the satisfaction someone has regarding his or her life in general, reflecting how happy and fulfilled someone feels (Gilman et al. 2000) and is one of the most stable indicators of well-being, not influenced by short-term events and physical states (Levin et al. 2012).

3 Theory and Hypotheses

We rely on the social exchange theory to explain the hypothetical association between custody arrangements and parents’ well-being. The theory has been used abundantly in the divorce literature to predict marital stability and satisfaction (e.g. Levinger 1979). The starting point of the theory is that “people implicitly ‘compute’ the estimated psychological ‘profit’ (i.e. the psychological rewards less the costs) of a potential or actual relationship when deciding whether to enter into it or to remain in it” (Braver et al. 1993, p. 10). In this way, it resembles the definition of trust, which is described as the willingness of individuals to place their well-being in the hands of other persons, taking a risk by placing faith in others (Delgado-Márquez et al. 2012; Wilson and Eckel 2011). The decision to trust involves the evaluation of costs and benefits. Therefore, the social exchange theory that will be applied to our research question could ideally be added with the trust concept, which could be a prerequisite to well-being. However, as trust is not within the scope of this article, we will not further explain how trust could be placed within our theoretical model, but will come back to it within the discussion of the results.

The profits can be divided into three types: affectional (interpersonal), material (tangible) and symbolic (moral) profits (Levinger 1979). Two decades ago, this social exchange theory was used by Braver et al. (1993) to predict the behavior of the non-custodial parent in relationship to the parent–child bond. After divorce, the non-custodial parent (usually the father) decided whether or not to maintain the relationship with the child based on an evaluation of the situation in terms of costs and rewards, i.e. “computing” the profit of maintaining a relationship. Three types of profits can be distinguished: affectional profits are related to the interpersonal value the parent attaches to the continuation of the parent–child relationship, material profits are the tangible investments in time or money related to maintaining a close relationship with the child and symbolic profits are related to the value given to the satisfaction derived from continuing the parental role.

We use the social exchange theory to assess the association between the custody arrangement of children and parents’ well-being. On the one hand, parenthood is rewarding because of the enjoyment of having a solid bond with the child and the fulfillment of the parental role. On the other hand, caring for children can be demanding because of the time needed to invest in child care (and hence the time lost for personal leisure activities). We expect that parents will implicitly “compute” the profit of a certain custody arrangement, which will in turn affect their level of well-being. We elaborate the work of Braver et al. (1993) by not only researching the non-custodial parent. We examine no custody parents, sole custody parents and joint custody parents. In this way, we examine all possible groups of parents, evaluating the social exchange theory for divorced fathers and mothers separately.

As stated before, divorced parents implicitly “compute” profits, according to their specific custody arrangement, which are expected to influence their well-being. In this study, we distinguish between two types of profits: parental involvement and personal leisure time. We start with discussing parental involvement, which can be classified within the group of affectional and/or symbolic profits. Earlier research efforts showed that frequent involvement with both parents after divorce enabled children to have a solid relationship with both parents, especially leading to a better father–child relationship (Amato and Rezac 1994; Gunnoe and Braver 2001; Spruijt and Duindam 2009). Moreover, the parental role is one of the highest valued social roles in society (Rittenour and Colaner 2012), leading to a higher profit when the satisfaction with this role is higher and to a higher level of subjective well-being altogether (Rogers and White 1998). The continuation of the parental role and the parent–child relationship after divorce may act as an important interpersonal resource in coping with the disrupting life event of the divorce itself (Amato 2000; Eggebeen and Knoester 2001; Madden-Derdich and Leonard 2000). To maintain a stable bond and to enact the parental role after divorce, parents need to have regular interaction with their children (Amato and Gilbreth 1999; Fabricius et al. 2012; King and Sobolewski 2006). Within the traditional custody arrangement of sole mother custody, the father–child bond usually erodes quickly when fathers do not reside on a regular basis with their children (Swiss and Le Bourdais 2009). Hence, especially fathers would prefer to have at least joint custody, increasing their relationship with their children and subsequently their level of well-being.

Another pathway through which the custody arrangement may influence parents’ well-being relates to the availability of free or leisure time, which can be classified within the group of material profits. Because parenting time interferes with personal leisure time, sharing the parental tasks after divorce may create more time for the social, work and relational life (Degarmo et al. 2008; Eggebeen and Knoester 2001; Gunnoe and Braver 2001). In particular for engaging in social activities, the burden of lone parenthood forms a barrier after divorce (Kalmijn and Broese van Groenou 2005). Earlier research showed a clear advantage of joint physical custody, compared to sole custody, of an increased availability of leisure time (Breivik and Olweus 2006; Kalmijn and Broese van Groenou 2005; Lee 2002). Most research already demonstrated this for divorced mothers, yet not often for divorced fathers. The reason was that the group of sole custody fathers in empirical studies was often too small to examine. Research on time use showed that mothers reduced their personal leisure time when they needed to spend time with their children and that they were less able than fathers to protect some leisure time for themselves (Beck and Arnold 2009; Shaw 2001). Because parental tasks can be shared, joint custody mothers are, compared to sole custody mothers, half of their time ‘childless’ and free to engage in outdoor social activities. In other words, joint custody mothers are assumed to have a more active social life than full-time residential mothers, because of their extended leisure time and decreased childcare tasks. Joint custody parents are able to organize a life ‘at two speeds’; their ‘built-in break’ permits them to save social activities for the amount of time that their children are with the other parent. This in turn may have a positive effect on mothers’ levels of subjective well-being. We thus expect that especially being a sole custody parent impedes the social life after divorce, leading to a lower level of subjective well-being. Having a satisfying social life is linked positively to someone’s subjective well-being (Gilman et al. 2000; Kroll 2010; Shapiro and Keyes 2007). Leisure time activities, such as doing sports or going out with friends, widen one’s social network and generate social resources. Spending time in these less organized and often unplanned activities is as important as participating in more formal associations (Putnam 2000; Seippel 2006). They provide people with resources, social support and information (Almond and Verba 1963) and they are important for an individual’s mental and physical health (Berkman et al. 2000).

Based on the social exchange theory, we expect that the association between custody arrangement and parents’ subjective well-being can be predicted by the perceived rewards and costs (defined here as the amount of parental involvement and the availability of leisure time) attached to each specific custody arrangement. Sole custody parents’ well-being may benefit from the satisfaction of continuing the parental role, but could be impaired by the lack of available leisure time. No custody parents’ well-being may benefit from having more leisure time for a profound social life, but may be weakened by the absence of a close bond with the child. Joint custody parents are expected to have maximal rewards and minimal costs which brings their estimated outcome or profit (defined here as subjective well-being) to the highest level. On the one hand, they live a substantial amount of time with their children, which may help them to maintain a close bond with their children. On the other hand, they have a considerable amount of ‘childless time’, in which they can engage in personal leisure activities. Therefore, we expect that joint physical custody parents have (a) higher levels of well-being than no custody parents because of their higher level of parental involvement and (b) higher levels of well-being than sole custody parents because of their extended amount of leisure time.

Because sole mother custody used to be the main post-divorce model for a long period of time, our research questions contain an important gender dimension. The shift from sole mother to joint physical custody, which receives particular attention in this study, has changed the post-divorce lives of mothers and fathers differently. Divorced mothers experienced a shift from being the sole custody parent to being half-time responsible for the children. Fathers experienced a transformation from being hardly to being regularly involved in their children’s lives. Therefore, gender specific hypotheses are highly recommended, in which joint physical custody mothers are compared with sole custody mothers and joint physical custody fathers with no custody fathers. Based on this reasoning, we formulate the following hypotheses:

  1. 1.

    Joint custody fathers show higher father involvement than no custody fathers.

  2. 2.

    Higher father involvement is related to higher paternal subjective well-being.

  3. 3.

    Joint custody fathers show higher levels of well-being than no custody fathers because of their higher levels of father involvement.

  4. 4.

    Joint custody mothers engage more often in leisure time activities than sole custody mothers.

  5. 5.

    A more active social life (in terms of leisure activities) is related to higher maternal subjective well-being.

  6. 6.

    Joint custody mothers show higher levels of well-being than sole custody mothers because of their more active social life.

4 Method

4.1 Data

The Divorce in Flanders survey (Mortelmans et al. 2011) sampled marriages from the Belgian National Register. These marriages were first marriages of both spouses, concluded in the period 1971–2008, between two Belgians of the opposite sex, who were between 18 and 40 years old at the time of marriage, and living in Flanders (the northern part of Belgium). The sample was drawn proportionally towards marriage year, but disproportionally towards marriage status to obtain a distribution of 2/3 dissolved and 1/3 intact marriages. Divorced respondents could have been remarried in the meanwhile. Respondents were interviewed by means of face-to-face Computer Assisted Personal Interviews between 2009 and 2010. From each marriage (either dissolved or still intact), one child was randomly chosen among all mutual biological (or adoptive) children. All questions with regard to children, including the custody status, related to this child. Our subsample included all divorced parents (N = 1,506) with a child that was maximum 21 years old and still living with (at least) one of the parents. For 49 % of the dissolved marriages, both mother and father participated in the study.

4.2 Measurement

Subjective well-being is a multidimensional latent construct, measured via three observable well-being dimensions: absence of depressive feelings, self-esteem and life satisfaction. This well-being index is efficient and concise (O’Hare et al. 2012). The eight-item CES-D scale (Radloff 1977) is used to measure feelings of depression and reversed to measure well-being instead of ill-being. This scale is often used and shows good psychometric properties (Van de Velde et al. 2011). Respondents indicate how often they felt or behaved in a certain way, e.g. felt lonely, slept bad, felt depressed. The four answer categories range from (almost) never to (almost) always. Self-esteem is measured via the ten-item Rosenberg self-esteem scale (Rosenberg et al. 1989). Examples of items are “I feel that I am a person of worth, at least on an equal plane with others” and “I feel that I have a number of good qualities”. The five answer categories range from (almost) never to (almost) always. Life satisfaction is measured via the Cantril ladder (Cantril 1965). On a scale from 0 to 10, respondents are asked how they feel in general, if they are happy and content. The extremes are defined by the respondent himself: 0 means the worst possible life quality and ten means the best possible life quality one can imagine.

Parents’ custody status is measured with the residential calendar (Sodermans et al. 2012) and refers to the actual living arrangement of the child at the time of the interview. The proportion of time the child lives with each parent is calculated and a variable with three categories is constructed: sole custody (more than 66 % of the time residential parent), joint custody (between 33 and 66 % of the time residential parent) and no custody (<33 % of the time residential parent). These cut-off criteria are commonly used (e.g. Fabricius et al. 2012; Melli 1999; Melli et al. 1997; Smyth et al. 2008). To enable comparisons between mothers and fathers, joint physical custody serves as the reference group.

Parental involvement is measured via three indicators. First, a single-item question measures the quality of the parent–child relationship. Parents indicate on a five-point scale “How good or how bad is your relationship with your child?” Answer categories range from very bad to very good. The majority of parents (91.2 %) reports a good to very good relationship with their child. The second and third indicator are positive and negative communication, measured by openness in communication and problems with communication subscales of the parents–adolescent communication scale (Barnes and Olson 1986). Answers on a seven-point scale range from totally disagree to totally agree.

The availability of leisure time is measured via three different forms of participation in social activities: doing sports, participating in cultural events, and going out (Kalmijn and Broese van Groenou 2005; Scheepers et al. 2002). All questions have seven-point answering scales ranging from never until very frequently (i.e. daily).

We control for confounding factors that might be correlated with both the custody arrangement and the level of well-being. Highly educated and employed parents are more likely to have joint custody (Juby et al. 2005; Spruijt and Duindam 2009) and higher levels of well-being (Diener et al. 1999). Education is divided into three categories. Low educated parents only finished lower secondary education (on average obtained at the age of 15), average educated parents obtained a higher secondary education degree (on average obtained at the age of 18), and highly educated parents obtained a higher educational or university degree (on average obtained at the age of 22). We differentiate between three employment types: full-time employed (working 95 % or more), part-time employed (working 25–94 %) and unemployed (<25 %). For those with a missing value (13 mothers and 19 fathers), an additional dummy variable is included. The presence of a new inhabiting partner and the age of inhabiting children can influence both the decision to share custody, as well as the adjustment after divorce (Degarmo et al. 2008; Van Ingen 2008). The presence of a co-residing new partner and children below 12 years old are included as two dummy variables. Also the time elapsed since the divorce is important to consider. Recent divorcees are more likely to choose a joint custody arrangement, yet, may have a lower well-being because the recovering process is still ongoing. The relationship between age and well-being is unclear (Diener et al. 1999). Finally, conflict between ex-partners is an important stressor that influences well-being (Degarmo et al. 2008). The association between parental conflict and custody is, however, not straightforward (Breivik and Olweus 2006). Parental conflict is measured via the frequency of having severe disagreements with the ex-spouse, classified in three categories: often, sometimes or never. A separate category is created when there is no contact between ex-partners.

In Table 1, the descriptive statistics of all variables are reported. Exact wordings of scale items are reported in the “Appendix”.

Table 1 Descriptive measurements
Full size table

4.3 Analytic Strategy

To measure the two mediating paths from custody arrangement to well-being, we conduct structural equation modeling (SEM), using Mplus (version 5.21) (Wu and Zumbo 2008). We proceed in two steps. First, measurement models are constructed and tested. Confirmatory factor analyses are performed to measure the latent constructs. Regarding the dependent variable well-being, we construct the latent variables of depression and self-esteem and subsequently perform a second-order confirmatory factor analysis on these two latent variables plus life satisfaction. To test the parental involvement hypothesis, we construct an openness in communication and a problems in communication scale. To test the leisure time hypothesis, we construct the latent concept leisure time, based on the three leisure activities. We perform a simultaneous confirmatory factor analysis in which all these latent constructs are tested simultaneously, the so-called null model. Second, structural models are built and tested in which relations between custody status, the mediating variables and well-being are specified.

Because a considerable number of interviewed mothers and fathers belong to, and thus are nested within the same dissolved marriage, single-group SEM could lead to biased results. The dyadic data structure is therefore approached via multi-group SEM. This imposes the same measurement model for mothers and fathers, making gender comparisons possible (Meuleman and Billiet 2009).

As we use several categorical variables, we opt for a robust weighted least square (WLSMV) estimation method instead of a maximum likelihood (ML) estimation method. WLSMV corrects for not following the assumptions of continuous items and normality of data. It is considered the most preferable estimation technique for ordinal variables in SEM (Byrne 2011; Muthén 2007). WLMSV does not start from the analysis of the sample variance–covariance matrix, but is based on the analysis of the correlation matrix. The evaluations of the models are based on several fit indices. Modification indices and parameter estimates are assessed to modify our model. Because of the size of the sample, we do not rely on the χ2 and its p value. Instead, we study the more appropriate Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Weighted Root Mean Square Residual (WRMR). The CFI and the TLI indices are two comparative fit indices that take on values between 0 and 1. Values of .90 and higher indicate a good model fit (Byrne 2011). The RMSEA index penalizes for poor model parsimony and takes the error of approximation in the population into account. Values smaller than .08 suggest an adequate model fit and values smaller than .05 a good model fit (Browne and Cudeck 1993). The WRMR index assesses the difference between the implied correlation matrix and the correlation matrix in the sample. It is designed especially for confirmatory factor analysis with ordinal items (Byrne 2011). Values smaller than .95 are regarded as indicators of a good model fit (Yu 2002). However, the WRMR is an experimental fit measure and does not always behave well (Muthén and Muthén 2010). Therefore, we approach this fit index with caution.

5 Results

5.1 Measurement Model

Figure 1 shows the final measurement model in which all latent variables are examined simultaneously via confirmatory factor analysis. The index of well-being is constructed via a second-order confirmatory factor analysis. Based on the separate confirmatory factor analyses, modifications are made. Error correlations are included and these are allowed to differ between fathers and mothers. The threshold of the third item in the depression scale is also allowed to differ between fathers and mothers. Measurement equivalence between fathers and mothers is reached for all latent constructs. Subjective well-being is a clear index, composed out of self-esteem, life satisfaction and absence of feelings of depression. For fathers, self-esteem is mostly related to the higher order composite of well-being, while for mothers, well-being is more equally determined by self-esteem and feelings of depression. Life satisfaction, which is regularly used as a single measurement of well-being, relates strongly to well-being as a composite index. Correlations between the latent variables are also examined. Perceived openness in communication with the child correlates positively with well-being (for mothers: r = .28, p < .0001; for fathers: r = .24, p < .0001), whereas problems in the parent–child communication correlate negatively with well-being (for mothers: r = −.39, p < .0001; for fathers: r = −.34, p < .0001). Both dimensions of communication are negatively related to each other (for mothers: r = −.42, p < .0001; for fathers: r = −.27, p < .0001). Leisure time, composed out of the three leisure activities, relates positively with well-being (for mothers: r = .41, p < .0001; for fathers: r = .34; p < .0001). This measurement model forms the null model in which no causal relations are specified and is the reference model to which every structural model is compared to. The fit indices of the null model indicate that the model fit is sufficient (CFI = .947, TLI = .973, RMSEA = .068). In a next step, we proceed with this measurement model and specify causal relations between the latent concepts.

Fig. 1
figure 1

Final measurement model for two groups solution. Entries are standardized factor loadings (one-directional arrows). Reversed items are presented by °. Error terms are not reported here, but are available upon request

Full size image

5.2 Structural Model

5.2.1 Direct Effect of Custody on Subjective Well-being

Figure 2 presents the final structural model in which we specify the direct relations between the observed variables and the latent constructs. We model direct effects of custody arrangements on parents’ well-being, parental involvement and leisure time, controlled for the confounding factors. The direct effects from the mediating variables (parental involvement and leisure time) on well-being are also reported. The structural model fits our data reasonably well. The CFI and TLI value are both higher than the cut-off value of .900 (CFI = .925, TLI = .946). The RMSEA value is .068, which indicates an adequate model fit. The direct effects of the control variables on well-being are reported separately in Table 2.

Fig. 2
figure 2

Final structural model for two groups solution: direct effects of custody type on subjective well-being. Entries are standardized coefficients. Dotted arrows denote non-significant effects, full arrows denote significant effects (*p < .05; **p < .01; ***p < .001.). The short, dashed arrows symbolize error variances of the endogeneous variables. Reference group of custody type is joint physical custody

Full size image
Table 2 Direct effects of control variables on subjective well-being
Full size table

In a first step, we look at the direct effects of the custody arrangement on the mediating variables (parental involvement and leisure time), presented in Fig. 2. The level of parental involvement is related with the custody status of fathers. The quality of the parent–child relationship is not reported differently by sole and joint custody parents, but is lower for no custody parents. Regarding communication, different patterns emerge for mothers and fathers. Openness in communication is only related to custody status for mothers, problems in communication is only related to custody status for fathers. Sole custody mothers have more open communication with their child than joint custody mothers, who in turn communicate more openly with their child than do no custody mothers. Sole custody fathers report more negative communication with their child than do joint custody fathers who in turn report more negative communication than do no custody fathers. Our results show evidence for hypothesis 1: joint custody fathers are more involved with their children than no custody fathers.

The availability of parental leisure time is related with the custody status of mothers, not with those of fathers. Mothers who live predominantly together with their child engage less frequently in leisure activities than mothers who share parental tasks with their ex-husband. Joint and no custody mothers are not different regarding their frequency of having leisure activities. Our results show evidence for hypothesis 4: joint custody mothers have a more active social life than sole custody mothers.

In a second step, we look at the direct effects of the custody arrangement, parental involvement and leisure time on well-being, also presented in Fig. 2. There are no direct effects of the custody arrangement on the well-being of divorced mothers and fathers. Hence, we have to reject hypothesis 3 and 6. As formulated in hypothesis 5, more time spent on leisure activities is positively linked with well-being. Hypothesis 2 can be confirmed partially: more openly communication between parent and child is positively linked with well-being while problems in communication are negatively linked with well-being. The quality of the parent–child relationship is not related to well-being. No gender differences exist regarding the effects of the mediating factors on well-being.

In a third step, we discuss the direct effects of the confounding control variables on well-being (Table 2). The socioeconomic status variables influence strongly the well-being of divorced mothers and fathers. Being highly educated and having a job increases divorced parents’ well-being profoundly. While there is no difference in well-being between part-time and full-time working mothers, fathers who only work part-time have a lower well-being than fathers working full-time. Other indicators that point to the relational life of divorcees also affect the level of well-being. While a new inhabiting partner increases divorcees’ well-being, conflict with the ex-spouse influences well-being negatively. It has to be noted that for mothers, few conflicts with the ex-spouse decrease their well-being, compared to mothers who report an average level of conflicts with the ex-spouse. This may implicate that having some conflicts with the ex-spouse is considered beneficial for the relationship between ex-partners and mothers’ well-being. The other indicators regarding age and time since divorce do not influence the well-being of divorced parents.

5.2.2 Indirect Effect of Custody on Subjective Well-being

The well-being of joint custody mothers is positively influenced through more leisure time activities and negatively influenced through less open communication with their child, compared to sole custody mothers. The well-being of joint custody fathers is negatively influenced through more negative communication with the child, when compared to no custody fathers. Hence, joint physical custody rewards mothers with more leisure time which they have to pay off by reduced closeness in the relationship with their children. For fathers it means higher involvement in their children’s lives, but this extended parenting also brings communication difficulties. The total sum of all these counteracting forces for the well-being of fathers and mothers is however not significantly different from zero (Table 3).

Table 3 Indirect effects of custody type on subjective well-being through mediating variables
Full size table

6 Conclusion

We examined the association between custody arrangements and divorced parents’ well-being. Two explaining mechanisms were explored: the parental involvement and the availability of leisure time. We especially focused on parents with joint physical custody, the residential arrangement that was recently put forward by the Belgian legislator.

Our study found no direct association between custody status and well-being of divorced parents. However, the results confirmed some mediating paths, suggesting that there are specific costs and rewards related to the shift towards joint physical custody, which are different for mothers and fathers. No custody parents (living <33 % with their child) report a worse parent–child relationship than joint and sole custody parents, who are not differing from each other. Hence, a legislation in favor of gender neutral parenting arrangements, instead of traditional sole mother custody, may be positive for the relationships between children and both parents, especially between children and their father. Joint physical custody facilitates continuity in the post-divorce father–child relationship and does not hinder the mother–child relationship. Our result could be interesting for socialization theory, that treats both parents as important agents of socialization for children, because our findings may yield important implications for the father–son and the mother–daughter relationship (Grusec and Hastings 2007). Previous research showed that the relationship with the same-sex parent has a stronger impact on adolescents’ well-being than the relationship with the opposite-sex parent (King 2006; Videon 2002). Joint physical custody might in this regard be a good strategy to approximate two-parent families, limiting the loss of parental resources following divorce (Breivik and Olweus 2006). It could help to eliminate some of the stress experienced by families of divorce (Lowery and Settle 1985) and minimize changes in both structural and functional characteristics of the family.

Though the quality of the parent–child relationship was perceived equally by mothers with sole and joint physical custody, the latter reported less open communication with their child and this indirectly influenced their well-being negatively. Joint physical custody fathers had a better relationship with their child, but also more negative communication than no custody fathers. Spending more time with the child and taking up parental responsibilities on a daily base seems to offers extended opportunities for conflicts between fathers and their children. Moreover, this problematic parent–child communication was negatively associated with well-being of fathers. Problematic parent–child communication was not related to the custody status of mothers. According to Beck and Beck-Gernsheim (2002), mothers are still the ‘heart of the family’ and perform the more expressive functions, like listening to children, and talking about emotions. Fathers may fulfil more controlling functions, like setting boundaries, which could explain their higher communication difficulties with their resident children. This gender difference can be linked to the role identity theory of Maurer and Pleck (2006). Caregiving, although increasingly taken up by fathers, is still seen as a traditional female task. The role of the father remains somewhat ambiguous which causes fathers to have more difficulties in taking up the parenting role, in particular after divorce. Mothers, even those who are not living with their children, have less difficulties than fathers with continuing their parental role after divorce (Degarmo et al. 2008; Furstenberg and Nord 1985; Hetherington and Stanley-Hagan 1999). Future research should give more attention to these issues.

Mothers who share the care have more time for leisure activities than sole custody mothers and this is positively related with their well-being. Dividing parental care with the ex-partner appears to make extra room for personal outdoor activities. Herewith, we may confirm the statement of Bauserman (2012) that more equally divided custody arrangements caused a liberating effect for women. Divorced fathers’ leisure activities were not influenced by their custody status. This finding could be important for future research about the organization of post-separation childcare by mothers and fathers. Today, a considerable share of divorced fathers can see their children grow up, instead of spending time with them only during the weekends. Surprisingly, the social life of these fathers seems not to differ from that of non-residential fathers. Thus, while joint custody seems to bring more personal time for mothers, it does not seem to cut it back for fathers. Two possible explanations could explain this finding. First, only full-time parenthood impedes parents’ social life, whereas co-parents are able to organize a life ‘at two speeds’. Their ‘built-in break’ permits them to save social activities for the week that the children are with the other parent. A second explanation may be that fathers are less inclined than mothers to readjust and reorganize their life according to the presence of children (Shaw 2001) or rely more easily on informal help from a new partner or parents, as expressed by Jappens and Van Bavel (2012). Fathers may have few reservations about obtaining external help with the children (holiday camps, babysits, formal childcare) while the same assistance may induce feelings of ambivalence and guilt among mothers. This would be an interesting topic of future research.

This study had limitations that should be taken into consideration. Firstly, we study cross-sectional data and cannot make any causal interpretations. Happier parents may be more likely to choose for joint custody than emotional disturbed parents. It is confirmed in earlier research that joint custody parents differ from sole custody parents on a number of variables, like psychological functioning (Wolchik et al. 1985), socio-economic status (Donnelly and Finkelhor 1993) and parental involvement (Fabricius et al. 2012; Gunnoe and Braver 2001). Secondly, the group of no custody mothers/sole custody fathers would definitely deserve more attention in future research. No custody mothers are less inclined to participate in surveys because of their social undesirable status and there is also evidence that they have lower levels of well-being (Buchanan et al. 1992). This may lead to selective non-response. Thirdly, the quality of the parent–child relationship was measured with a single item question showing very little item variance. Perhaps, high social desirability was present as parents have a tendency to report more positive experiences/perceptions than children do (Sodermans et al. 2013c). Finally, some discrepancies between the answers of divorced mothers and fathers regarding the custody arrangement were possible. Divorced parents tend to overestimate the time that their children reside with them, causing a response bias. Furthermore, the selectivity of survey participation can produce this difference, as divorced parents with sole custody are more eager to participate than parents without custody.

Additional to these limitations, we also want to point out that we encourage further research that would overcome these limitations and furthermore, would add new fields of research. For instance, it would be interesting to include the concept of trust within the model, as a prerequisite of well-being. As we know that trust and social participation form a virtuous relationship (Putnam 2000), it would be interesting to see whether we could interchange the well-being concept of well-being by trust. Trust increases when social interaction increases and vice versa. People who trust are more likely to take higher risks when engaging in social relations and cooperate more in social networks (Gambetta 1988; Knack and Keefer 1997). This dynamic effect of trust would mean that custody would lead to not only better relationships but also more trust. This is still to be researched.

Despite these limitations, we believe that this study delivered an important contribution to the study of custody arrangements and family functioning. In a landscape where countries are increasingly modifying laws and policies to promote joint custody, studying living arrangements and well-being of divorced mothers and fathers is highly relevant. Our final conclusion would be that a shift towards shared care after divorce has both positive and negative effects for parents. For fathers it means higher involvement in their children’s lives, but this extended parenting also brings new difficulties. Joint physical custody is the very expression of the ‘changed nature of fatherhood’ in which fathers are increasingly willing and socially expected to be involved in caregiving (Maurer and Pleck 2006). For mothers, joint physical custody does not seem to touch their relationship quality with their children, although their bond may be somewhat less close. Moreover, it permits them to have a more active social life, which is rewarding for their general well-being. Hence, joint physical custody may be seen as a continuation of the individualization process and the modernization of the family. Beck and Beck-Gernsheim (2002) describe that female emancipation in the field of education and paid labour transformed the maternal biography from ‘living to others’ to ‘a life of one’s own’. Yet, they add that this individualization process is incomplete for women because many of them still take up the majority of domestic and child-related tasks leaving them with frustration and ambivalence. The legal recommendation of shared care practices pushes divorced couples in the direction of gender neutrality and is a continuation of the ‘detraditionalisation of gender roles’ (Van Krieken 2005).

The question arises whether children also benefit from joint physical custody. According to Australian evidence, Fehlberg et al. (2011) concluded that shared time parenting favors in the first place fathers, followed by mothers and in the least place the children. For them, commuting between two places means less stability in their post-divorce life. There is also evidence that legislating for joint physical custody may lead to more parental conflicts (Fehlberg et al. 2011; Singer 2008; Sodermans et al. 2013a). Jensen (2009) asks the somewhat provocative question whether the well-being of children is sacrificed for the sake of involved fatherhood and for a ‘fair’ share of parenting time. It is evident that future research should give more attention to the consequences for children because joint physical custody became the legally recommended custody option in the first place to serve their best interests.