Several studies have documented that high parental support and control enhances the well-being of children (Baumrind 1991; Lamborn et al. 1991; Milevsky et al. 2007), even after a parental divorce. These studies mainly concentrated on parenting of the mother (e.g., Benson et al. 2008; Forgatch and DeGarmo 1997; Lengua et al. 2000; Wood et al. 2004). Previous research on divorced fathers primarily examined the economic role of the father or the contact between father and child and the association of these aspects with children’s well-being (e.g., Amato et al. 2009; King and Heard 1999). However, growing evidence suggests that qualitative parenting (i.e., the parenting dimensions support and control) of the father is also beneficial to children’s well-being (e.g., King and Sobolewski 2006). Recently, the focus of some researchers has shifted to the parenting of non-residential fathers (Booth et al. 2010; Flouri 2006; King and Sobolewski 2006; Stewart 2003). While this shift is an important one, research on the parenting dimensions of fathers in different family structures, especially fathers in joint-custody, and the association with children’s well-being has remained underdeveloped.

The current study seeks to extend previous literature on the parenting of divorced fathers and the association with children’s self-esteem. In this research, we first compared differences in the parenting of non-residential fathers, fathers in joint custody and married fathers. Second, we conducted multivariate regression models to assess the association between children’s well-being and parenting of fathers in different family structures, controlled for the parenting of the mother. Using children’s reports on the parenting of fathers instead of parental reports, we also sought to give a unique insight on how children experience the parenting of their fathers. So, the present study aimed to provide a better understanding of how parenting of fathers in different family structures is associated differently with children’s well-being, including fathers in joint custody, controlling the multivariate models for parenting of the mother and using children’s reports on the parenting of both fathers and mothers.

Differences in Parenting Dimensions of Fathers in Different Family Structures

According to Thomson et al. (1994), parents offer their children two key resources: money and time. Whereas money is used to supply food, clothing and shelter for children and provide material goods and experiences that foster positive child development, time gives parents the opportunity to demonstrate support and control to their children. Support and control are the two dimensions of the parenting construct of Baumrind (1991). Support (also termed responsiveness) refers to the amount of affection and warmth that parents show and provide for their children. Control (also termed demandingness) refers to the extent to which parents supervise their children. The present study is focused on the parental support and control of different fathers, spending a different amount of time with their children.

Divorce is associated with a decline in parental resources (King and Sobolewski 2006; McLanahan and Sandefur 1994; Thomson et al. 1994). From a risk and resilience perspective, Degarmo and Forgatch (1999) have stated that divorce and the subsequent transitions and changes in the life course of fathers and children (e.g., formation of a new stepfamily) can function as stressors that lead to disruptions in the parenting of fathers. This decline in father support and control after divorce is confirmed in previous literature (Bronte-Tinkew et al. 2010; Hilton and Devall 1998; Sirvanli-Ozen 2005). After a divorce, fathers are more likely to spend less time with their children and the support and control they provided diminishes.

Still, the support and control provided by divorced fathers who spend a different amount of time with their children is less extensively investigated. After a divorce, most fathers become either non-residential fathers (i.e., fathers who spend less than 34 % of the time with their children) or fathers in joint custody (i.e., fathers who spend least 34 % and at most 67 % of the time with their children) (Melli 1999; Smyth and Moloney 2008). Since parental resources will diminish if fathers spend less time with their children (King and Sobolewski 2006; McLanahan and Sandefur 1994; Thomson et al. 1994), differences in parenting between non-residential fathers and fathers in joint custody should be found. Still, empirical studies that address the parenting of fathers in joint custody arrangements are scarce. To our knowledge, only one early study by Bowman and Ahrons (1985) investigated the differences of parenting between non-residential fathers and fathers in joint custody, indicating that fathers in joint custody were more involved in parenting than non-residential fathers. Although the importance of investigating parenting of divorced fathers in different custodial arrangements is stressed (Pasley and Braver 2004), empirical research remains underdeveloped. Therefore, the current study compares the parenting dimensions provided by fathers who spend a different amount of time with their children.

Parenting Dimensions of Fathers in Different Family Structures and Children’s Well-Being

Theoretical Frameworks on the Link Between Parenting Dimensions of Fathers and Children’s Well-Being

Two theoretical frameworks shape the context of the current study: the family system perspective and the social capital theory. On the one hand, the family system perspective offers a framework to investigate the link between fathers and children. The key idea of this perspective is that a family is a complex, multilateral and integrated structure in which family members are necessarily interdependent (Cox and Paley 1997; Minuchin 1974). As a consequence, actions of one family member might have an impact on another member. In this respect, the divorce-stress-adjustment perspective of Amato (2000) is embedded in the family system perspective and defines a parental divorce as a stressful process with consequences for children’s well-being that can be mediated by effective parenting (high support and high control). Indeed, previous research has demonstrated that higher support and control on the part of the non-resident father after a parental divorce have a positive influence on the well-being of the child (Bronte-Tinkew et al. 2010; Campana et al. 2008; Carlson 2006; Dunlop et al. 2001; King and Sobolewski 2006; Lansford 2009).

On the other hand, the mechanisms through which parenting of fathers and children’s well-being are positively associated, can be explained by the social capital theory (Coleman 1988). Social capital exists in the relationship between the parent and the child and has been measured as the quantity (i.e., contact) as well as the quality (i.e., parenting) of the parental involvement (Furstenberg and Hughes 1995). Previous research has revealed that the quality of the involvement, i.e., parental support and control, is important for children’s well-being (e.g., Baumrind 1991; King and Sobolewski 2006). Nevertheless, maintaining contact between the parent and the child is a necessary condition for the transfer of social capital. After a parental divorce, contact between the father and the child declines and former married fathers become fathers in joint custody or non-residential fathers. As the contact diminishes, the transfer of social capital as qualitative parental involvement also might decline with consequences for children’s well-being. Still, the majority of the empirical research on the parenting of divorced fathers and its link with children’s well-being has been limited to non-residential fathers (e.g., Carlson 2006; Hawkins et al. 2007) and have not taken into account the quantitative aspect of the social capital theory. Examining the link between the qualitative aspect of parenting and children’s well-being, the current study considers the quantitative aspect of social capital. It should therefore provide some additional insight in children’s well-being after divorce, especially for children of fathers in joint custody.

Self-Esteem, a Positive Indicator of Children’s Well-Being

The family system perspective and the social capital theory clearly indicate the association between the parenting dimensions of the father and the well-being of the child. Studies investigating this association have applied different definitions of children’s well-being as well as different measures for children’s well-being. Previous research mainly has used internalizing or externalizing problem behavior as a measure of children’s well-being (for an overview, see: Amato 2000; Hetherington and Stanley-Hagen 1999; Lansford 2009). Well-being can thus be defined as a lack of internalizing and externalizing problems. Nevertheless, as observed by Ben-Arieh (2005), the absence of internalizing and externalizing problems does not necessarily indicate a person’s feelings of happiness and self-worth. Consequently, a positive indicator of children’s well-being should be employed.

Such a positive indicator of well-being, highly correlated with happiness (Lyubomirsky et al. 2006), is self-esteem, which assesses a person’s feelings of self-acceptance and self-worth (Rosenberg 1965). It can be described as the level of satisfaction with one’s own behavior and one’s self (Diener and Diener 1995). As described by the interactionist Cooley (1902) in early sociology, individual feelings towards the self and an evaluation of the self are shaped through reflected appraisals of significant others in a context of social interaction (i.e., looking glass self). Self-esteem is constituted during childhood and adolescence, in close relationship with significant others, like parents (Birkeland et al. 2012; Crocker and Park 2004; Zakeri and Karimpour 2011). Research by Zakeri and Karimpour (2011) has indicated that high levels of support and high levels of control were associated with higher self-esteem among adolescents. Other research has reported the same result for support but the opposite result for control (Plunkett et al. 2007; Siffert et al. 2012). A recent study by Garcia and Gracia (2009) on parenting styles and self-esteem of adolescents has demonstrated that adolescents with indulgent parents (who only provide support and no control) had the highest self-esteem. It is important to note that in the ongoing literature concerning self-esteem, researchers debate on whether they should study self-esteem as a trait or as a pursuit and its costs. This pursuit of self-esteem is associated with goals and motivations to obtain high self-esteem rather than the evaluation of the self (Crocker and Park 2004; DuBois and Flay 2004). Bearing this in mind, previous research consistently has shown that a high level of self-esteem, or rather the avoidance of low self-esteem, is important for a person’s well-being throughout the life span (DuBois and Flay 2004). As a consequence, the current study concentrates on self-esteem as a positive indicator of children’s well-being.

The Current Study

Guided by the parental resource theory (Thomson et al. 1994), the first aim of the present study was to draw the comparison between the support and control provided by non-resident fathers, fathers in joint custody and married fathers. Since these three types of fathers have an unequal distribution of the resource “time with children” (McLanahan and Sandefur 1994; Thomson et al. 1994), we expected them to differ in their parenting. We hypothesized that married fathers provided the highest levels of support and control, subsequently followed by fathers in joint custody and non-residential fathers provided the lowest levels of support and control (hypothesis1).

The second aim of the present study was to investigate the association between the parental support and control of the fathers and children’s self-esteem, differentiating between fathers with a different amount of contact: married fathers, fathers in joint custody and non-residential fathers. First, from a family system perspective, the association between parenting of the father and children’s self-esteem is controlled for those different types of fathers. Based on previous literature (Bronte-Tinkew et al. 2010; Campana et al. 2008; Carlson 2006; Dunlop et al. 2001; King and Sobolewski 2006; Lansford 2009; Plunkett et al. 2007), we expected that higher levels of parental support as well as higher levels of parental control might be associated with a higher self-esteem (hypothesis 2). Second, we investigated the social capital hypothesis that parenting dimensions of fathers are not equally beneficial for children’s self-esteem when the amount of contact between the father and the child is different. This was a more explorative part of the study since, to our knowledge, previous studies had not compared the association between parenting of the father and children’s self-esteem, differentiating between married fathers, fathers in joint custody and non-residential fathers. We expected differences by family structure in the link between children’s self-esteem and parenting dimensions of the father, hypothesizing that the association between support and control of married fathers and children’s self-esteem would be the largest, subsequently followed by the support and control of fathers in joint custody and the association with children’s self-esteem, the association between support and control of non-residential fathers and children’s self-esteem would be the smallest (hypothesis 3).

Method

Sample and Procedure

Our analyses were based on data from the “Divorce in Flanders” study (Flanders is the Dutch speaking part of Belgium), which applied a multi-actor design, including information on both currently and formerly married partners, as well as on their children of 10 years of age or older, their parents, and (for formerly married partners) their new partners. In order to contact the currently and formerly married partners and after approval of the privacy committee, their adressess were randomly selected from the Belgian National Register (1/3 addresses of currently married partners, 2/3 addresses of formerly married partners), so the data were representative within the group of married or divorced partners. Data collection started in October 2009 and ended in December 2010. The final dataset contained information on 6,470 current and former partners, 1,257 residential children of minimum 10 years old, 320 non-residential children of minimum 18 years old, 2,157 parents, and 1,837 new partners. Divorced couples were overrepresented in the dataset (N = 6,004). Only partners who had been married after 1971 and who could have been divorced only once were included (although partners who had been married twice were included).

The current research used a subsample of the “Divorce in Flanders” dataset. The analytic sample contained data only from one partner of a current or former couple and a residential child of minimum 10 years old. The current or former parent was always the biological or adoptive parent of the residential child. In case of multiple children older than 10 years, one child was randomly selected. Both the child and the parent were questioned in face-to-face interviews (Computer Assisted Personal Interviews). First, the parent was questioned and asked for permission to question the child. Both were interviewed in their current household. Although the questionnaire was adapted to suit the child’s age (i.e., 10–13 years, 14–17 years, and 18 years or older), each child was asked the same questions. The parents reported on the custody arrangement and background information whereas the child reported on self-esteem and parenting dimensions of both fathers and mothers. Using children’s reports on parenting corresponds with the recent tendency of considering children as active agents (Ben-Arieh 2005). Previous research by Pasley and Braver (2004) has indicated that parental responses on parenting might be biased, due to social desirability. They found that using fathers’ reports might overestimate the support and control of fathers, whereas research using mothers’ reports might underestimate both parenting dimensions. Madden-Derdich and Leonard (2002) found similar results for parenting behaviors. Self-reports from children may provide a unique perspective on parenting dimensions, as the feeling of being controlled or criticized is a highly subjective experience (Aunola et al. 2000; Barbe, 1996).

A number of restrictions were made for our subsample. First, in line with the studies in the meta-analysis on parenting and children’s well-being of McLeod et al. (2007), only children between 10 and 18 years old who still had contact with both parents were included. Since the parenting of both fathers and mothers are at the core of this research, children from families in which contact between a parent and the child has been broken down, were beyond the scope of this study (n = 24). We also excluded children from whom information on the custody arrangement was missing (n = 14). Second, we included data from only one randomly selected parent (not necessarily the father), given the previously noted bias of father responses. More specifically, comparing the children of participating and non-participating fathers, we found that children with participating fathers considered their fathers more supportive and controlling than children with non-participating fathers did. In order to include children with non-participating fathers in the study, we used data from only one parent (mother or father). The parents provided background information about themselves as well as about their current or former partners. Our analytic sample included 628 children and their parents: 224 children with married parents and 404 children with divorced parents.

Measures

Children’s Self-Esteem

The Rosenberg Self-Esteem Scale (Rosenberg 1965) assesses a person’s feelings of self-acceptance and self-worth. Children rated ten items along a four-point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree. Consistent with the procedure developed by Marsh (1996), confirmatory factor analysis revealed a single factor. Factor loadings ranged from 0.54 to 0.75. The range in those factor loadings indicated that each item contributed differently to the latent concept of self-esteem. Therefore, factor scores based on those factor loadings were preferable above simple mean or sum scores and were used in the following analyses. Internal consistency (Cronbach’s alpha) of this scale was 0.85. Other recent studies on adolescents using the Rosenberg Self-esteem Scale have found comparable internal consistency for self-esteem (e.g., Asgeirsdottir et al. 2010; Birkeland et al. 2012; Kim and Sherraden 2011; Moksnes et al. 2010). Moreover, our internal consistency was equal to the internal consistency of this scale used in another sample of Dutch speaking adolescents (Bos et al. 2010).

Family Structure

Children of married parents were assigned to the group of married families (n = 224). Children of divorced parents (n = 404) were assigned either to the group of fathers in joint custody or to the group of non-residential fathers. To assign a child of divorced parents to either one of those family structures, the randomly selected, divorced parent completed two custody calendars. On the first calendar, the responding parent indicated the days and nights the child stayed with him or her. On the second calendar, the responding parent indicated the days and nights the child stayed with the other parent. Based on this calendar, a custody arrangement was assigned to each child. We used the distribution method of custody arrangement, as applied in other research (Melli 1999; Smyth and Moloney 2008): parents were considered having joint custody if the child stayed with one parent for at least 34 % and at most 67 % of all nights and stayed the rest of the nights with the other parent. A residential parent was defined as a parent with whom the child stayed for more than 67 % of all nights; a parent with whom the child stayed for less than 34 % of all nights was considered a non-residential parent. This division yielded four custody arrangements: children in joint custody (n = 138), children with a residential mother and a non-residential father (n = 225), children with a non-residential mother and a residential father (n = 31) and children for whom both the mother and the father were non-residential (n = 10). The last two groups were excluded from our analyses because of the small number of cases.

Parenting Dimensions of Fathers and Mothers

To measure parental support and control, two subscales of the Parenting Style Inventory II (PSI-II) were used (Darling and Toyokawa 1997). This is consistent with previous research on parenting dimensions and styles, although this previous research used simple mean or sum scores (Carlo et al. 2007; Nijhof and Engels 2007). In order to examine the unique and independent link between fathers’ parental support and control and children’s self-esteem, we controlled our analyses for the parental support and control of the mother, as suggested by Pleck (2010), whereas most research on parenting dimensions of fathers had concentrated solely on the father (e.g. Stewart 2003). As a consequence, children completed the PSI-II for their fathers as well as for their mothers. For both subscales, children rated five items along a five-point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree. The following are two examples of items from the support subscale: “I can count on my mother/father to help me out if I have a problem” and “My mother/father and I do things that are fun together.” The following are two examples of items from the control subscale: “If I don’t behave myself, my mother/father will punish me,” and “My mother/father points out ways I could do better.” Measurement equivalence was found for the PSI-II scale between children of married fathers, children of fathers in joint custody and children of non-residential fathers, since we controlled for configural and metric invariance (∆χ2 = 52.11; ∆df = 42; p = 0.14).

Factor analysis was conducted on the five support items for mothers and fathers separately, resulting in a single factor. For support of the mother, all items were included, with factor loadings ranging from 0.60 to 0.82. For support of the father, all items were included, with factor loadings ranging from 0.58 to 0.82. The range in those factor loadings indicated that each item contributed differently to the latent concept of support. Therefore, factor scores based on those factor loadings were preferable above simple mean or sum scores and used in the subsequent analyses. The internal consistency (Cronbach’s alpha) of the support subscale was 0.77 for mothers and 0.80 for fathers. Other studies using the PSI-II have found comparable internal consistency for support (Carlo et al. 2007; Hardy et al. 2008; Nijhof and Engels 2007).

Factor analysis was conducted on the five control items for mothers and fathers separately, resulting in a single factor. Still, within this factor, three of the items (two of mother’s control and one for father’s control) had factor loadings lower than 0.4, and were therefore excluded. The other item for father’s control also was excluded, as its factor loading was substantially lower than those of the other items (0.49 versus 0.74). For control of the mother, the factor loadings for the remaining three items ranged from 0.68 to 0.83. For control of the father, the factor loadings for the remaining three items (which were the same as for mothers) ranged from 0.74 to 0.80. The range in those factor loadings indicated that each item contributed differently to the latent concept of control. Therefore, factor scores based on those factor loadings were preferable above simple mean or sum scores and used in the subsequent analyses. The internal consistency (Cronbach’s alpha) of the control subscale was 0.67 for mothers and 0.73 for fathers. This is comparable to the internal consistency for control that has been reported in other studies using the PSI-II (Carlo et al. 2007; Hardy et al. 2008; Nijhof and Engels 2007).

Control Variables

From the divorce-stress-adjustment perspective (Amato 2000), it became clear that individual characteristics of parents and children also were associated with children’s well-being. Previous research has indicated that, amongst others, child gender and age, duration since divorce, presence of a new partner, parents’ age and educational level of the parent can moderate between parental divorce and children’s well-being (Amato 2000; Goodman and Pickens 2001; Hetherington and Stanley-Hagen 1999; Lansford 2009). So, we controlled for characteristics of both the child and the parents (mother and father). For child characteristics, we included the child’s gender (50 % boys and 50 % girls) and age (M = 14.23, SD = 2.53). For parental characteristics, we included mother’s age (M = 42.86, SD = 4.12), father’s age (M = 44.44, SD = 4.24), mother’s level of education (12 % had completed lower secondary education or less, 41 % had completed higher secondary education, and 47 % had completed higher education), father’s level of education (17 % had completed lower secondary education or less, 46 % had completed higher secondary education, and 37 % had completed higher education), the presence of a new partner in the household of divorced mothers (50 % had no new partner in the household, and 50 % had a new partner in the household) and the presence of a new partner in the household of divorced fathers (39 % had no new partner in the household, and 61 % had a new partner in the household). Duration since divorce, defined as the amount of time that had passed since both parents had started to live separately in different homes (M = 7.89, SD = 3.78), was included as well. These characteristics were reported by the parent. Descriptive statistics for all control variables by family structure are shown in Table 1, indicating that the three groups of family structure did not significantly differ for education of the mother and the presence of a new partner of the mother. For all the other control variables, significant differences between the three groups of family structure were found.

Table 1 Descriptive statistics of control variables by family structure
Full size table

Analytic Strategy

In order to examine the first aim and hypothesis, variance analyses were performed to compare support and control of married fathers, fathers in joint-custody arrangements and non-residential fathers. We used the non-parametric test of Kruskall and Wallis, because the data were not normally distributed. Ordinary-least-squares (OLS) regression was conducted to investigate the second aim. Before conducting the OLS regression, all assumptions of the regression model (e.g., multicollinearity, homoscedasticity) were tested, and the assumptions were confirmed. First, base regression models that only included parental support and control of the father were estimated. Second, total regression models, including parental support and control of the father as well as all the other independent variables, were estimated. Hypothesis 2 is investigated with an OLS regression of parenting dimensions of the father on children’s self-esteem, controlled for family structure. Hypothesis 3 is examined using as interaction-effect between family structures and parenting dimensions of the father (in order to statistically compare the expected differences by family structure) as well as estimating separated regression models for each family structure. For both hypothesis 2 and 3 the coefficients of the base models remained quite stable in the total models. Therefore, only the regression coefficients of the total models were reported, although the explained variance of both the base models and the total models was reported. Missing values on dependent, independent and control variables were excluded from the analyses using listwise deletion.

Results

Comparing Support and Control of Fathers Between Different Family Structures

In order to investigate the first aim and hypothesis, we compared the parental support and control of married fathers, fathers in joint custody and non-residential fathers. Mean factor scores for father’s support and control are presented in Table 2. Contrary to what we hypothesized, the support and control provided by fathers in joint custody did not differ from the support and control provided by married fathers. In line with the first hypothesis, we found that the support and control of non-residential fathers differed significantly from those of married fathers and fathers in joint custody. Children reported non-residential fathers as less supportive and less controlling than married fathers and fathers in joint custody.

Table 2 Parental support and control of fathers by family structure
Full size table

Associations Between Fathers’ Parenting Dimensions and Children’s Self-Esteem

In order to examine the second hypothesis, a total regression model that reports the unique contribution of fathers’ parental support and control on children’s self-esteem, controlled for family structure, is shown in Table 3. The results indicated that, as expected, the parental support of fathers affected the self-esteem of children, even after controlling for mothers’ parenting dimensions and family structure. As expected, children reported higher self-esteem when their fathers were more supportive. Fathers’ control had no significant association with children’s self-esteem, so hypothesis 2 is partially confirmed. In line with previous literature (Amato 2000; Goodman and Pickens 2001; Lansford 2009), our results on the control variables indicated that children with a supportive mother reported higher self-esteem, boys reported higher self-esteem than girls, and children of more highly educated fathers reported higher self-esteem. No other significant results were found. When it comes to the goodness of fit (GFI), this total model explained approximately 21 % of the variance in children’s self-esteem; the base model indicated that the parenting dimensions of the father explained approximately 10 % of the variance in children’s self-esteem. The parental support and control provided by fathers can thus be considered as important predictors of children’s self-esteem.

Table 3 Overall influence of father’s parenting dimensions on children’s self-esteem (N = 530)
Full size table

Although there were no differences in children’s self-esteem according to family structure (married fathers, fathers in joint-custody arrangements, or non-residential fathers), fathers’ parental support and control of married fathers, non-residential fathers and fathers in joint custody could affect children’s self-esteem differently. The third hypothesis on the differences in the association between fathers’ parental support and control and children’s self-esteem according to family structure was tested by investigating an interaction-effect as well as separate total regression models for each group of family structure. The interaction-effect between family structure and parental support of the father as well as the interaction-effect between family structure and parental control of the father did not indicate significant differences according to family structure. Still, parental support and control of the father could contribute differently to the explained variance of children’s self-esteem in each family structure. So, results for the separated total regression models are shown in Table 4.

Table 4 Regression models of father’s parenting dimensions on children’s self-esteem for each family structure separately
Full size table

Models 2a, 2b, and 2c showed that, regardless of family structure, having a supportive father was associated strongly with children’s self-esteem. This was in line with the non-significant interaction-effect and did not confirm the third hypothesis. In joint-custody arrangements, having a supportive father appeared to be even more important than having a supportive mother. The results of Model 2b suggested that having a supportive mother is not associated significantly with children’s self-esteem in any family structure. As in Model 1, father’s control was not associated significantly with children’s self-esteem. More than 24 % of the variance in children’s self-esteem was explained in Model 2a and Model 2c. The explained variance for children in joint-custody arrangements (Model 2b) was 15 %. The differences in explained variance for family structure could not be attributed to the father’s parental support and control, as the adjusted R² of the base models indicated that father’s parental support and control accounted for approximately 9 % of the variance in children’s self-esteem in each family structure.

As shown in Table 4, significant results for control variables were largely in line with previous literature (Amato 2000; Goodman and Pickens 2001; Lansford 2009). Regarding children of married parents (Model 2a) and children with non-resident fathers (Model 2c), girls reported significantly lower self-esteem than boys. This association was not found for children in joint-custody arrangements (Model 2b). Furthermore, self-esteem was significantly lower for children with residential mothers who were living with new partners. The results further revealed that married fathers with higher levels of education were associated positively with the self-esteem of their children. The educational level of the mother was only associated significantly with the self-esteem of children in joint-custody arrangements. Children in joint-custody arrangements with more highly educated mothers reported higher self-esteem than children in joint-custody arrangements with less highly educated mothers. Finally, the results revealed a significant influence of child’s age for the children of married parents. Older children of married partners reported lower self-esteem than younger children of married partners.

Discussion

With European divorce rates rising and more children experiencing a parental divorce (Eurostat New Cronos 2011), increased attention is needed in order to understand the parenting of the father after divorce, especially fathers in joint custody arrangements. Most previous studies have focused on either the mother’s parenting or on the economic role of the father after divorce (e.g., Amato et al. 2009; Benson et al. 2008). Research on fathers’ parenting after divorce tended to concentrate on non-residential fathers and to use mothers’ or fathers’ reports to measure fathers’ parenting (Booth et al. 2010; Flouri 2006; King and Sobolewski 2006; Pasley and Braver 2004; Plunkett et al. 2007; Stewart 2003). The current study has compared the parenting dimensions of married fathers, non-resident fathers and fathers in joint-custody as reported by children, as well as the association between those dimensions and children’s self-esteem, adding to our knowledge on differences in parenting of fathers in different family structures and their association with children’s self-esteem. Our results contribute to the literature including fathers in joint-custody arrangements instead of focusing only on non-residential fathers, examining a positive indicator of children’s well-being, controlling for the parenting dimensions of the mother when investigating in the association between parenting dimensions of the father and children’s self-esteem, and using children’s reports.

With regard to the first aim of the current study on the comparison of parental support and control of fathers in different family structures, our analyses have revealed mixed results. In line with the first hypothesis, the findings of this study indicate that, as expected, non-residential fathers are less supportive and controlling than married fathers and fathers in joint custody, which may be due to the decline in parental resources after a divorce (McLanahan and Sandefur 1994; Thomson et al. 1994). Still, contrary to the first hypothesis, parental support and control provided by fathers in joint custody does not differ from that provided by married fathers, although it does differ from that provided by non-residential fathers. One possible explanation is that divorced fathers tend to be distressed about the loss of contact with their children (Jacobs 1983). This distress can function as a stressor for fathers’ parenting (Degarmo and Forgatch 1999). Given that fathers in joint-custody arrangements have continued the contact with their children (or have experienced a lesser degree of diminished contact), thus experiencing less distress and therefore less stress, they might be more supportive and controlling than non-residential fathers, who see their children less after divorce. Another possibility is that the continuation of contact with the child might result in the continuation of the same parental behavior of the fathers in joint-custody arrangements and therefore resemble the same levels of parental support and control as married fathers. In contrast, non-residential fathers might engage more in leisure activities with their children, and participate less in their children’s daily routines. This could be a compensation for the parental divorce and the diminished contact with their children (Stewart 1999). In conclusion, our findings indicate that the parenting of fathers in joint custody is more similar to the parenting of married fathers than to the parenting of non-residential fathers.

The second aim of the current study was to investigate the association between parental support and control of the father and children’s self-esteem. Results suggest that it is beneficial for children when their fathers are more supportive, controlled for family structure, which is in line with the second hypothesis. This contribution of divorced fathers to children’s self-esteem is independent of the parenting dimensions of mothers. Fathers’ control is not associated significantly with children’s self-esteem. Hence, this result does not support the second hypothesis, possibly due to the specific relationship between parental control and self-esteem, since we do not find any association with mothers’ control either. Whereas most research on children’s well-being and parental control found a positive relationship, these studies mainly have focused on internalizing or externalizing problems of children (e.g., Carlson 2006; Hawkins et al. 2007; Lansford 2009). Research on parental control and children’s self-esteem has revealed mixed results, for instance in a study by Zakeri and Karimpour 2011) a positive association between parental control and children’s self-esteem was found, whereas other scholars found a negative one (Plunkett et al. 2007; Siffert et al. 2012). These mixed results might be due to the measures of parental control. Zakeri and Karimpour (2011) measured control as psychological autonomy-granting as well as behavioral strictness-supervision, but only the former was related to children’s self-esteem. Plunkett et al. (2007) also used two measures of parental control (psychological control and punitiveness) and also only the former was related to children’s self-esteem. Another study by Kakihara et al. (2010) found that children’ self-esteem is only related negatively to control measured as rules, but not to control measured as restriction of freedom or coldness-rejection. So, in this study, the lack of a significant association between parental control (measured as supervision) and children’s self-esteem is in line with previous research on control measured as restriction of freedom, punitiveness or behavioral strictness-supervision (Kakihara et al. 2010; Plunkett et al. 2007; Zakeri and Karimpour 2011). Consequently, additional research is needed with regard to the role of parental control on children’s self-esteem, taking into consideration the different meanings and measures of the concept of parental control. Nevertheless, our findings with regard to father’s support sustain the family structure perspective, as they indicate that actions of the father are associated with feelings of the children, as well as the social capital theory, as there appears to be a positive relationship between parenting of the father and children’s well-being, even when controlled for the parenting of the mother.

Considering the differences in the association between fathers’ parenting dimensions and children’s self-esteem according to family structure, the third hypothesis is not confirmed. Interaction-effects as well as separate regression models all indicate that, in each family structure, having a supportive father is associated positively with children’s self-esteem. Similar to results of hypothesis 2, no significant association between control provided by the father and children’s self-esteem is found. Furthermore, comparing the explained variance of parenting dimensions of fathers in different family structures does not reveal differences with regard to the base line models. So, our findings do not provide support for this social capital hypothesis on contact between divorced fathers and children. Still, our results demonstrate that having a supportive father is associated positively with children’s self-esteem in different family structures, even after a parental divorce. This might give an indication of the protective role of parenting dimensions between family structure and children’s well-being, as described in the divorce-stress-adjustment perspective (Amato 2000). Although this perspective implies that the parental support and control of the mother are of primary importance in protecting the well-being of children, our findings indicate that the parental support and control of both non-residential fathers and fathers in joint-custody also are associated positively with children’s well-being. This adds to a growing body of research (Booth et al. 2010; King and Sobolewski 2006; Plunkett et al. 2007) suggesting that the involvement of divorced fathers does matter for children’s well-being. Instead of focusing on the decline in parenting of divorced fathers as a result of a decline in resources or the stressful transitions after divorce, future research should consider divorced fathers as fathers with the capacity to contribute to their children’s self-esteem.

It is important to note the limitations of the current study. First, it was based on cross-sectional data. We studied the parenting dimensions of fathers after divorce, as well as the impact of these dimensions on children’s self-esteem at a certain point in time. We thus provided a snapshot of fathers’ parenting dimensions, which are not static but dynamic concepts. Amato (2000) stressed that the parenting dimensions of fathers could have both a short-term and a long-term impact on children’s well-being. Other researchers have stated that having an involved father at Time 1 can benefit children’s well-being at Time 2 (Coley and Medeiros 2007). Furthermore, fathers’ parenting dimensions change through time, and some studies have indicated that they even decline over time after divorce (McLanahan and Sandefur 1994; Seltzer 2000). In contrast, other researchers have found no change in parenting after divorce (e.g., Strohschein 2007). Although we controlled for duration since divorce, we could examine only the short-term impact of fathers’ parenting dimensions. Future research might therefore benefit from the use of longitudinal data. Second, as suggested by Pleck (2010), reciprocal pathways of fathers’ parenting dimensions and children’s well-being should be investigated. A majority of the research on parenting after divorce and children’s well-being assumes that the parenting dimensions of fathers affect the well-being of children. A study by Hawkins et al. (2007) has demonstrated that children’s well-being also could affect the involvement of fathers. Possibly, fathers tend to be more involved when the well-being of their children is higher. Another recent study by Gault-Sherman (2011) also found a reciprocal relationship between parental attachment and adolescent delinquency. Third, shared method variance cannot be ruled out since we used children’s reports on both parenting and self-esteem. Nevertheless, these reports have offered a unique insight in children’s perception of their self-esteem and the parenting. Fourth, no information on separated parents who were never married was available in the “Divorce in Flanders” dataset. We acknowledge that in Western societies, the number of non-married parents is increasing. So, more insight on the parenting of those parents is necessary. Finally, because of the small number of cases (n = 31), we did not include residential fathers after divorce in our analysis. Although residential fathers after divorce are rare, their parenting dimensions might differ from those of other fathers, possibly because the circumstances under which fathers become residential fathers after divorce are associated with problematic mothers (Pryor 2004). Future research might benefit from focusing on divorced, residential fathers, their parenting dimensions, and the impact of these dimensions on children’s well-being.

Despite its limitations, the current study could have implications for public policy and debates concerning the role of the father after divorce. Discussions on the obligations and the rights of divorced fathers focus largely on the payment of child support after divorce (Menning 2006). Our findings suggest that beyond their ability to pay child support, divorced fathers have the capacity to play a unique role in the well-being of their children by being supportive. Moreover, further analysis after including the payment of child support in our final model has revealed that it has no significant impact, whereas the significant impact of father support remains. It might therefore be important for public policy to emphasize types of involvement on the part of divorced fathers other than child support and custody arrangements. Nowadays, divorcing parents are obliged to arrange the custody arrangement and the financial support for their children, either in court or in out-of-court negotiations. In many countries, including Belgium, a trend towards shared parental responsibility is noticeable (McIntosh 2009), especially in the current custody arrangements (e.g., joint custody) and the payment of child support. Still, public policy could go beyond this shared parenting concerning the custody or child support agreements by encouraging divorced fathers and mothers to also discuss and negotiate their parenting behavior and involvement with their children in a formal agreement, consistent with the formal parenting agreement of the Netherlands (De Rijksoverheid voor Nederland 2010). This formal parenting agreement includes the arrangements for custody of the child and the amount of child support, but also agreements on the parenting behavior and involvement, like which parent is to pay for extra costs (e.g., medical costs, hobbies, school trips), which parent is to assume caring tasks (e.g., staying at home when the child is sick, helping the child with homework, picking the child up from school), how the parents should communicate about the child, and which decisions about the child should be made together (e.g., school choice, punishment for coming home late). The latter is also included in the Australian legislation on shared parenting after divorce (McIntosh 2009). Such a formal parenting agreement would go beyond custody and child support arrangements and also includes guidelines on parenting and parental involvement. Therefore, it might lead parents to reflect more on the arrangements concerning the children, supply children with a more central role in the divorce process and encourage both mothers and fathers to remain or become supportive after divorce. Furthermore, such a formal parenting agreement could offer children an active role, if parents let them take part and take their wishes into consideration. It also is claimed that out-of-court negotiated agreements concerning the children are fairer, more acceptable to both parties, more durable, more likely to be fully implemented and lead to less conflict (Wasoff 2005). Nevertheless, we believe that legislation should only specify the framework of the formal parenting agreement during the divorce process and that it should not interfere with its content. Furthermore, divorced parents should be able to change this agreement in whole or in part as the child grows older, or as they experience other family transitions. Still, we should acknowledge that a formal parenting agreement might not be in the best interest of all divorced families (McIntosh 2009), so legislation should provide other possibilities.

In summary, the current study has made important contributions to the understanding of how formerly married fathers parent their children and how their parenting can benefit children’s well-being. Using a representative sample, our findings applied to a wide range of ever married fathers of adolescents, even beyond differences in custody arrangements. In particular, we have provided new evidence regarding the parenting dimensions of non-residential fathers as well as on fathers in joint-custody arrangements. We also compared fathers’ parenting dimensions after divorce to those of fathers who are still married, along with associations between these dimensions and children’s well-being after divorce. Controlling for the parenting dimensions of mothers, we have grasped the unique link of fathers’ parenting dimensions with children’s well-being. Even after controlling for mothers’ parenting dimensions, the importance of fathers’ parenting dimensions (especially support) for children’s well-being after divorce still remained, indicating that divorced fathers have the capacity to increase the well-being of their children in a positive manner. Finally, this study was based on children’s reports of fathers’ parenting dimensions, providing a unique insight in how children experience parenting. The current study has revealed that, even though divorced fathers are less supportive to and controlling of their children than married fathers, the children of divorced fathers benefit more when their fathers show higher support towards them.