Consequences of Divorce-Based Father Absence During Childhood for Young Adult Well-Being and Romantic Relationships

Objective: This research explores the implications of father absence due to divorce on young adults’ well-being and romantic relationships. Background: Studies have demonstrated the negative implications of father absence, a common consequence of divorce, on children’s development. However, previous research has not systematically compared complete versus partial father absence. Method: Young adults who, as children, experienced complete (n = 38) or partial (n = 41) father absence were compared with 40 participants in a father-presence control group. Results: Compared with those in the control group, young adults in the partial father-absence

group reported higher psychopathology and maternal overprotection, and lower maternal care, romantic intimacy, commitment, and passion. Under elevated maternal care, the partial-absence group reported lower dyadic adjustment and consolidated sense of identity. Participants in the complete-absence group reported higher self-criticism and maternal overprotection and lower maternal care than controls. Conclusion: Partial father absence might have particularly pernicious consequences for young adults' well-being. Implications: Clinical and public policy implications are discussed.
Divorce constitutes a severe family stress or (Hetherington & Stanley-Hagan, 1999) that is often associated with maladjustment in children of all ages (Amato, 2001(Amato, , 2010Amato & Dorius, 2010;Bernardi & Radl, 2014;Cummings & Davies, 2010;Harold & Leve, 2012;Hetherington et al., 1998). One of the major consequences of divorce for children is the departure of one parent, most commonly the father, from the household (Ahrons & Tanner, 2003;Braver et al., 1993). Although public policies and case law recommend continuing contact with both Family Relations (2020) parents after divorce, this contact may be difficult to sustain. Consequently, divorce often entails either a complete or partial absence of one of the parents, usually the father, from the child's life.
Father absence, also labeled "father hunger" (Herzog, 1982), pertains to the emotional and psychological longing that a person has for a father who has been physically, emotionally, or psychologically distant in that person's life (Erickson, 1996(Erickson, , 1998Perrin et al., 2009). Father absence due to divorce can have negative implications for child development, including increased risk of psychopathology (Culpin et al., 2013;East et al., 2006;Erickson, 1998;Kenny & Schreiner, 2009;McMunn et al., 2001), derailed interpersonal and romantic adjustment (Fergusson et al., 2014;Gruenert & Galligan, 2007;Wineburgh, 2000), and self-concept deficits (Pagura et al., 2006;Phares, 1999). The ongoing potential negative implications of father absence on intra-and interpersonal difficulties might be even more significant among individuals who experienced father absence as young children (age 0-6 years) compared with those who experienced it later (Beaty, 1995).
Nevertheless, research in this field has not systematically compared complete versus partial father absence (e.g., fathers not living with their children but being present in their lives to some extent). The latter comparison is particularly important in light of recommendations made by policymakers that the child should be in continuing contact with both parents, whereby contact is deemed important at any level of intensity, despite the fact that contact arrangements are often highly complicated and stressful for divorced parents (Target et al., 2017).
However, research has shown that despite the common agreement regarding the importance of the frequency of contact, frequency by itself has not generally been a good predictor of child outcomes (Amato & Gilbreth, 1999). This may be, in part, because fathers vary considerably in the quality of parenting they provide. This is evidenced by a meta-analysis of nonresident fathering and child well-being, which has shown that nonresident father involvement can potentially have positive effects on children; however, the quality of such involvement is more important than the quantity. This finding suggests that the types of activities in which nonresident fathers are involved and the quality of the time spent matter to their children's social and emotional well-being (Adamsons & Johnson, 2013). Moreover, it has been similarly suggested that while an increase in contact may be beneficial in general, its effects are dependent on the context of the contact and the quality of the parental involvement (Amato et al., 2009). Accordingly, if the contact occurs within the context of a hostile interparental relationship, contact might be problematic for the child's well-being. Also, specifically among younger children, the consistency of the schedule can be an important predictor of positive adjustment (for a review, see Kelly, 2007).
Based on the literature, it would appear that both quality and quantity are important to examine in relation to father presence. In agreement, Amato et al. (2009) have argued that despite research pointing to the importance of focusing on the quality of the relationship between nonresident fathers and their children, it is nevertheless important to continue studying the frequency of contact of nonresident fathers with their children. Given that between 1976 and 2002, the level of nonresident father involvement in the United States increased significantly (Amato et al., 2009), there is a growing need for studies in this area of research.
Indeed, even though reaching a global agreement about shared parenting policy has been elusive and difficult to attain, there is a large consensus among professionals that children of separated and divorced parents do best when they have stable, healthy, and continuing contact with both parents (Pruett & DiFonzo, 2014). Nevertheless, in many cases, divorce still entails a complete absence of the father from the child's life. Therefore, it is important to evaluate whether partial fathering (i.e., "partial father absence" as described herein)-that is, a variable amount of contact with the father-would have different outcomes for children's well-being than the complete absence of father-child contact (i.e., a discontinuation of any contact with the father following the parents' divorce). Such a comparison is even more important given the findings discussed herein suggesting that the quality of father involvement as well as its context may be more important than the quantity of contact (Adamsons & Johnson, 2013;Amato et al., 2009).
Another important limitation of previous research is the relative neglect of the implications of divorce-related father absence in young adulthood (Culpin et al., 2013;Rohner & Veneziano, 2001). This developmental period (ages 18-33 years; Arnett, 2000) involves tremendous opportunities for personal growth in various domains of development-educational, employment, interpersonal relationships, and sometimes parenthood (Salmela-Aro & Nurmi, 1997). It is also a critical period in which different elements of the developmental self (e.g., ego identity formation, self-criticism) are being manifested (Staples & Smarr, 1991). However, young adulthood also represents a developmentally challenging transition to adulthood, which might culminate in psychopathology (American College Health Association, 2006;Blanco et al., 2008;Hunt & Eisenberg, 2010); this might be particularly true when there is a history of father absence, given a growing body of literature suggesting that circumstances in the family of origin have consequences for the quality of early parent-child relationships that persist well into young adulthood (Amato & Sobolewski, 2001;Sobolewski & Amato, 2007).
Notably, despite our focus on the implications of father absence for young adults' well-being, it is also important to take into account that the father-child relationship does not exist within a void or as a standalone relationship. In fact, there is a vast consensus in both early and contemporary psychoanalytic views that the father-child relationship exists in the context-and even the approval-of the maternal figure (e.g., Target & Fonagy, 2002). Studies that have attempted to examine the potentially unique contributions of both father-child and mother-child relationships, as well as their interaction terms, to children's well-being have produced mixed results.
For example, Amato (1994) demonstrated that for children living with both parents, closeness to fathers was associated with more life satisfaction, happiness, and less distress in early adulthood, regardless of the quality of the mother-child relationship. However, the association between father closeness and satisfaction was moderated by family disruption (divorce): It was significantly weaker when offspring experienced parental divorce than when offspring grew up in continuously intact two-parent families. In addition, the two-way interaction terms between closeness to fathers and closeness to mothers were not significant in predicting children's outcomes (Amato, 1994).
In a general sample of divorced parents, it has been shown that father and mother warmth were both independently associated with lower child externalizing problems (Sandler et al., 2008). However, the associations between mother and child warmth and child internalizing problems varied as a function of interparental conflict and level of warmth with the other parent, suggesting that even when there is low conflict between the parents, the lack of a warm relationship with one parent might spill over on to the other parent, so that children are less able to benefit from the positive relationship with that parent (Sandler et al., 2008). According to another study, the associations between positive maternal and paternal parenting and child mental health problems were similarly moderated by the quality of parenting provided by the other parent, and also by the number of overnights children spent with parents, but not by the level of interparental conflict (Sandler et al., 2013). More specifically, when parenting by the other parent and number of overnights were considered together, only the number of overnights moderated the associations between parenting and child behavior problems, suggesting that the level of contact might also be an important factor to consider (Sandler et al., 2013).
In addition, in a study conducted among young adults, it was found that in cases of parental divorce, the young people did not report higher levels of subjective well-being if they were close to both parents than if they were close to only one parent and that divorce (but not marital conflict) was associated with an increased likelihood of having a close relationship with only one parent, usually the mother (Sobolewski & Amato, 2007). Interestingly, a significant interaction between maternal-child relationship and father absence has been found in predicting behavioral difficulties among adolescents, with a strong mother-adolescent relationship serving to protect adolescents in homes from which the father was completely absent from the risk of peer problem behavior (Mason et al., 1994).
Nevertheless, no study to date has examined the role of the mother-child relationship on subsequent child well-being by differentiating between families in which there is complete father absence and families with partial father absence. Given the aforementioned findings, it is important to further assess potential interactions between father absence and levels of mother-child bonding (e.g., care, overprotection) in predicting young adults' well-being.
Moreover, while previous studies have examined the implications of parental warmth for offspring's subsequent well-being, to the best of our knowledge, no study has examined the role of the mother-child relationships specifically on the child's romantic relationships in later life.
This study extends the current line of inquiry in its field in three ways. First, we examined the implications of divorce-based father absence occurring before age 6 years for young adults' well-being and interpersonal relationships. We explicitly distinguish between complete father absence (i.e., a discontinuation of any contact with the father following the parents' divorce) and partial father absence (a variable amount of contact with the father) and compare both types of father absence with father presence during childhood. Second, we focused on the implications of divorce-related father absence on young adulthood, a critical developmental period that has been relatively neglected in the research in this field. Finally, we examined a potential interaction between maternal bonding and father absence in predicting young adults' well-being and romantic relationship satisfaction.

Hypotheses
We hypothesized that young adults belonging to either the complete or partial divorce-based father-absence groups will exhibit greater psychopathology, romantic maladjustment, and self-concept deficits in comparison to the father-presence group. Self-concept deficits pertain to low levels of a consolidated sense of identity and high levels of self-criticism, a dimension demonstrated to be strongly implicated in psychopathology (Blatt, 1995;Shahar, 2015).
We also examined a potential interaction between maternal bonding and father absence in predicting young adults' well-being and romantic relationship satisfaction. We hypothesized that the expected differences between complete and partial father absence, and between both father-absence groups and the father-presence group, will be particularly pronounced in low vs. high levels of maternal care.

Participants and Procedure
Participants were 119 young adult undergraduates (73 females; age range: 22-32 years; M = 24.64, SD = 2.24). All were involved in a significant, meaningful, and stable romantic relationship lasting more than 3 months. Participants were age-matched, F(2, 116) = .077, MSE = .39, ns, into three groups: (a) complete father absence, n = 38, 24 females, M age = 24.68 years, SD age = 2.55; (b) partial father absence, n = 41, 28 females, M age = 24.89 years, SD age = 2.92; and (c) a father-presence control group, n = 40, 21 females, M age = 24.53 years, SD age = 1.28. According to the inclusion criteria for the study, the complete father-absence group included participants reporting not having any contact with their father, whereas the partial father-absence group consisted of participants reporting diverse frequencies of communication with their noncustodial father, ranging from contact on a weekly basis to once every 2 years. In the partial father-absence group, six participants reported contacting their father on a weekly basis, nine participants reported a monthly basis, six participants reported contacting their father once every 2 to 3 months, seven participants reported contacting their father once a year, two participants reported contact once every 2 years, and 11 participants reported a varied amount of contact with their father during their childhood.
In both groups, father absence resulted from divorce occurring when the participant was age 6 years or younger (M = 3.23, SD = 1.87, range: 0-6.5 years), with all participants subsequently living with their custodial mother. 1 Participants were contacted through advertisements placed on bulletin boards posted at universities and colleges nationwide. They completed a consent form, filled out self-report questionnaires (described later in the article; included other factors not pertinent to the current report) in the presence of a research assistant, and were then debriefed and financially compensated with $15. Participants came from different areas in Israel, including urban and rural regions (e.g., cities, kibbutzim [collective settlements]). The sample of 119 participants was highly diverse 1 We also requested that participants report whether their mother had a significant partner after the divorce and whether this person had been a significant part of their lives. However, we did not have enough statistical power to examine interactions between father-absence status and the presence or absence of a postdivorce significant partner. This should be targeted in future research. in terms of ethnicity and religious and socioeconomic status; 71% (84) of the participants were born in Israel, 19% (23) in former Soviet Union (USSR) countries, and 3% (4) in other countries; the remaining eight (7%) participants did not complete this question. The study was approved by the Ethics Committee of the Department of Psychology.

Measures
Psychopathology. We used the Brief Symptom Inventory (BSI; Derogatis & Melisaratos, 1983;Derogatis & Spencer, 1982), a questionnaire consisting of 53 items rated on a 5-point Likert scale assessing a wide range of self-reported psychopathological symptoms pertaining to the past month. The authors documented high test-retest and internal consistency reliabilities, together with strong evidence of convergent, discriminate, and construct validity. A global BSI score was computed by averaging the 53 items (in this sample Cronbach's α = .96).
Romantic relationship satisfaction. The Dyadic Adjustment Scale-4 (DAS-4; Sabourin et al., 2005), a brief four-item scale developed and validated based on the original 32-item (DAS-32) scale, was used to assess romantic relationship satisfaction. Individuals are asked to estimate their relationship satisfaction (e.g., "In general, how often do you think things between you and your partner are going well?" and "Do you confide in your mate?"). Authors reported the DAS-4 to be informative at all levels of couple satisfaction and as effective as the 32-item version of the DAS in predicting couple separation, less affected by socially desirable responses, and less time consuming to complete. The underlying latent construct measured by the DAS-4 was very stable over a 2-year period (Sabourin et al., 2005); in this sample, Cronbach's α = .71.
Perceived romantic relationships (intimacy, passion, commitment). Perceived romantic relationships were assessed using the Triangular Love Scale (TLS; Sternberg, 1997), an extensively used, validated, 45-item self-report measure rated on a 9-point Likert scale (Whitley, 1993). The TLS is based on Sternberg's triangular theory of love, according to which love comprises an emotional component, intimacy (i.e., feelings of closeness, connectedness, and bonding in loving relationships); a motivational component, passion (i.e., the drives that lead to romance, physical attraction, sexual consummation, and related phenomena in loving relationships); and a cognitive component, decision/commitment (i.e., the decision that one loves another person and a commitment to maintain that love). In the present sample, Cronbach's α = .93 for intimacy, .89 for passion, and .92 for commitment.
Self-criticism. Self-criticism was assessed using the DEQ-SC6, a validated six-item measure of self-criticism (for details, see Rudich et al., 2008), based on the original 66-item scale of the Depressive Experiences Questionnaire (DEQ; Blatt et al., 1976). The DEQ is a self-report measurement designed to measure the role of personality in depression. Individuals are asked to indicate, on a 7-point Likert scale, the extent to which they agree with each item. The DEQ-SC6 was used in previous studies, evincing satisfactory internal consistency and reliability (Lassri et al., 2013;Rudich et al., 2008;Soffer et al., 2008). In this sample, Cronbach's α = .74.
Identity. The Adolescent Ego-Identity Scale (AEIS; Tzuriel, 1984Tzuriel, , 1992, a validated 38-item scale developed based on Erikson's ego-identity theory (Erikson, 1963(Erikson, , 1968, was used to assess identity. The scale assesses the level of ego-identity formation/cohesion among young adults and adolescents. Individuals are asked to indicate on a 5-point Likert scale how well an item describes their personal feelings or opinions about themselves (1 = not true at all, 4 = very true). The AEIS produces a global score, computed by averaging the 38 items reflecting the status of the participant's ego-identity cohesion. In the present study, Cronbach's α = .86.
Maternal care and overprotection. We used the Parental Bonding Instrument (PBI; Parker et al., 1979), a 25-item scale designed to measure key parental styles (i.e., behaviors and attitudes) contributing to the development of bonding experiences during childhood, as retrospectively perceived by the individual. Individuals are asked to rate how closely each description corresponds to their experience with their parents (mothers and fathers are rated separately) during their first 16 years on a 4-point Likert scale (0 = very unlike, 3 = very like). Two dimensions derive from the PBI items: care (12 items) and overprotection (13 items). The PBI has been found to have good reliability and validity (Parker et al., 1979;Parker, 1990), with split-half reliability coefficients of .88 for care and .74 for overprotection. In addition, the PBI has been shown to be stable over a 20-year period in a nonclinical population and resistant to the effects of life experiences and mood states In addition, it has been shown to be stable over a 20-year time period, with retest coefficients in the range of .64 to .83 for maternal care and .74 to .82 for paternal care. Similarly, maternal overprotection coefficients were in the range of .67 to .77, and paternal overprotection scores .59 to .78 (Wilhelm et al., 2005). Internal consistency in this sample was Cronbach's α = .90 for paternal care, α = .75 for paternal overprotection, α = .86 for maternal care, and α = .86 for maternal overprotection. 2

Analytic Strategy
Before testing our hypotheses, we computed means and standard deviations of the variables in each of the three groups, as well as correlations among the study variables in the entire sample. Skewness and kurtosis for all study variables were calculated and fell within the acceptable range.
We tested our hypothesis regarding group differences using two sets of multiple regression analyses. In Set 1, the two father-absence groups were compared with the father-presence group via two dummy-coded variables (Aiken & West, 1991). The father-presence group served as a reference group, coded as 0 on both dummy variables. In one dummy variable, the partial father-absence group was coded as 1, and in the other, the complete father-absence group was coded as 1. The outcomes were BSI-psychopathology, DEQ-SC6 self-criticism, DAS-romantic relationships, TLI-perception of romantic relationships, AEIS-identity, and PBI-maternal care and overprotection.
Set 2 was similar to Set 1, except for the identity of the dummy variable. We compared the complete versus partial-absence groups by having the complete-absence group serve as a reference group, coded as 0 on both dummy variables. In one dummy variable, the partial father-absence group was coded as 1, and in the other, the father-presence group was coded as 1. A statistically significant effect of the first dummy variable (i.e., partial father absence = 1) in the presence of the other dummy variable (i.e., father presence = 1) amounts to a comparison between the complete and partial father-absence groups (Aiken & West, 1991).
We tested the hypothesis regarding the moderating effect of maternal care on the link between father absence (complete or partial) and young adults' well-being using two sets of multiple regression analyses. The outcomes were the same as in the preceding analyses, with the exception of maternal care. The predictors were the previously described two dummy variables, as well as PBI maternal care and two interaction terms involving maternal care and the dummy variables. In Set 1, the dummy-coded variables enabled a comparison between complete father absence and father presence, and between partial father absence and father presence. In Set 2, the dummy-coded variables enabled a comparison between complete and partial father absence. Statistically significant interactions were probed based on recommendations provided by Aiken and West (1991). Specifically, the effects of the dummy variables on the outcomes were examined under high versus low levels of maternal care, where high versus low levels correspond, respectively, to 1 SD above and below the centered mean of maternal care.

Descriptive Statistics
In Table 1, we present means and SD of all the study variables for each of the three study groups. In Table 2, we present zero-order correlations among the study variables in the entire sample. Maternal care was positively correlated with all romantic relationship measures, including satisfaction, intimacy, commitment, and passion, as well as with identity consolidation, and was negatively correlated with levels of self-criticism, psychopathology, and maternal overprotection. Maternal overprotection was positively correlated with levels of psychopathology and self-criticism, negatively correlated with identity, and marginally significantly correlated with less passion (p = .056). Note. AEIS = Adolescent Ego-Identity Scale; BSI = Brief Symptoms Inventory; DAS = Dyadic Adjustment Scale; DEQ-SC6 = Depressive Experience Questionnaire, Self-Criticism Six-Items Scale; PBI = Parental Bonding Inventory; TLI = Triangular Love Inventory. Note. AEIS = Adolescent Ego-Identity Scale; BSI = Brief Symptoms Inventory; DAS = Dyadic Adjustment Scale; DEQ-SC6 = Depressive Experience Questionnaire, Self-Criticism Six-Items Scale; PBI = Parental Bonding Inventory; TLI = Triangular Love Inventory. ns = nonsignificant; * p = < .05. * * p = < .01. * * * p = < .001.

Group Differences
In Table 3, we present the results of Set 1 findings deriving from the multiple regression analyses, in which each of the father-absence groups was compared with the father-presence control group. As shown in the table, the partial father-absence group differed from the father-presence group on almost all outcomes (note: the difference for DAS romantic relationship satisfaction was marginally significant, p = .05). The exception is DEQ self-criticism, for which a marginally significant (p = .05) difference was found between the complete father-absence and father-presence groups. In addition, the complete father-absence group differed from the father-presence group in DAS romantic relationship satisfaction and in maternal care and overprotection. The direction of relationships in all statistically significant comparisons was such that the father-absence groups demonstrated greater maladjustment than the father-presence group. Finally (and not shown in Table 3), for TLI intimacy, there was a statistically significant sex effect (b = .33, = .18, t = 2.00, p < .05), with females reporting greater intimacy than males.
The Set 2 analysis revealed no differences between the complete and partial father-absence groups, hence these analyses are not presented.
The pattern of this interaction, presented in Figure 1, was probed based on Aiken and West (1991). Under high levels of maternal care (1 SD above the mean), the dummy predictor comparing partial father absence versus father presence was statistically significant (unstandardized simple slope = -2.73, SE = .94, t = -2.91, β = -.42, p < .01). The direction of the relationship, as shown in Figure 1, suggested that under high maternal care, young adults in the father-presence group reported higher DAS romantic relationship satisfaction than those in the partial father-absence group. Under low levels of maternal care (1 SD below the mean), the relationship was reversed, although it was nonsignificant (unstandardized simple slope = 1.38, SE = 1.32, t = 1.04, β = .21, p = .30).
AEIS identity was predicted by PBI maternal care, b = .04, SE = .01, t = 2.54, β = .65, p < .05, and the maternal care × partial father absence versus father presence interaction, b = .04, SE = .02, t = -2.17, β = -.41, p < .05. The model accounted for 15% of the variance of the outcome, F(6, 112) = 3.40, p < .01. When the pattern of this interaction, presented in Figure 2, was probed based on Aiken and West (1991), we found that under high levels of maternal care, the dummy predictor comparing partial father absence versus father presence was statistically significant (unstandardized simple slope = -.40, SE = .13, t = -3.07, β = -.44, p < .01). The direction of the relationship suggests that under high maternal care, young adults in the father-presence group reported a more consolidated identity than those in the partial father-absence group. Under low levels of maternal care (1 SD below the mean), the relationship was reversed, although it was nonsignificant (unstandardized simple slope = .14, SE = .18, t = .76, β = .15, p = .44). Set 2: Comparing complete versus partial father absence. Of all the outcome variables, only DAS romantic relationship satisfaction yielded pertinent findings. Specifically, this outcome was predicted by maternal care (b = .21, SE = .07, t = 2.84, β = .39, p < .01), and the maternal care × complete versus partial father absence interaction (b = -.24, SE = .10, t = -2.39, β = -.29, p < .05). However, there were no group differences under high or low maternal care (1 SD above and below the centered mean). Hence, we did not interpret this interaction. The model accounted for 15% of the variance of the outcome, F(6, 112) = 3.49, p < .01.

Discussion
We investigated the consequences of divorce-based complete versus partial father absence before age 6 years on young adults' well-being. The results, based on 119 young adults who were currently involved in a romantic relationship, highlight the deleterious associations between noncustodial father absence and a range of outcomes. Specifically, compared with young adults growing up with both parents, young adults whose fathers were partially absent before age 6 years reported greater psychopathology; perception of lower romantic intimacy, commitment, and passion; and a more negative retrospective recall of maternal bonding in terms of both care and overprotection. Under high levels of retrospectively recalled maternal care, young adults whose father had been partially absent before age 6 years reported less dyadic adjustment (relationship satisfaction) and consolidated identity than young adults in the father-presence group. The complete father-absence group demonstrated higher subsequent self-criticism and retrospectively reported maternal overprotection, and lower retrospectively recalled maternal care, than the father-presence group. Strikingly, there were no differences between the partial versus complete father-absence groups in terms of young adults' well-being. In general, our findings add to mounting evidence signifying the longitudinal associations between divorce and children's developmental outcomes, adjustment, and well-being (Adamsons & Johnson, 2013;Amato, 2001Amato, , 2010Amato & Gilbreth, 1999;Amato & Keith, 1991;Bartoszuk & Pittman, 2010;Culpin et al., 2013;East et al., 2006;Erickson, 1998;Gruenert & Galligan, 2007;Kenny & Schreiner, 2009;Levy-Shiff, 2018;McMunn et al., 2001;Pagura et al., 2006;Wineburgh, 2000). Previous findings suggest that children's subjective well-being was highest when they were close to both continuously married parents who had a low-conflict relationship compared with other conditions, such as divorce or high parental conflict, suggesting that divorce appeared to negate the advantage of having two close parent-child ties (Sobolewski & Amato, 2007).
Nevertheless, despite the substantial emerging evidence of the importance of the paternal figure for children's development (Amato, 2010;Amato & Dorius, 2010;Target & Fonagy, 2002), for many years the significance of the paternal figure was virtually absent from both researchers' and practitioners' attention. This is presumably the result of a tendency to regard parenting as synonymous with motherhood, in line with a societal norm viewing the mother as the child's main caregiver, regardless of the fact that many fathers participate in caring for their children and positively influence their children's psychological well-being (East et al., 2006). This marginalization has resulted in a need to further evaluate the implications of fathers' presence and, importantly, absence on children's development. Sobolewski and Amato (2007) addressed this issue by suggesting that many nonresident fathers may act as if they are "visitors" rather than parents, a dynamic that often promotes father-child relationships that are mainly entertaining in nature. Given that relying on fun activities may tend to be somewhat superficial, it is difficult for children and their nonresident fathers to maintain bonds characterized by depth and mutual satisfaction. Therefore, it might be that fathers in the partial absence group, despite having more contact with their children compared with the compete father-absence group, had difficulties establishing close and significant relationships with their children-the type of bonds that contribute to healthy development but are predicated upon stable relationships (Main, 2000;Sobolewski & Amato, 2007). Thus, whereas children who are close to two continuously married parents who have a low-conflict relationship are potentially able to turn to both parents for protection and emotional and stress regulation, children with an unstable relationship with their father are less likely to be able to access that parent to resolve distress, and the father may even be the origin of their distress, given the instability and inconsistency of the contact arrangement.
Our results indicating no differences between the partial versus complete father-absence groups in terms of young adults' well-being support this understanding and are in agreement with previous studies proposing that the nature of father-child contact is more important than simply its quantity, especially in regard to stability (Adamsons & Johnson, 2013;Amato et al., 2009;Sandler et al., 2013). Thus, it is possible that in some cases of partial father absence, despite there being more father-child contact than if the father were completely absent, the higher level of father-child contact might not be linked to better outcomes in the child's well-being compared with the complete father-absence group, presumably as the father-child contact lacks consistency and stability. It is also possible that in some cases, having more father-child contact (as in the case of the partial father-absence group vs. the complete father-absence group) may also mean having higher levels of exposure to interparental conflict, given the need for the divorced parents to discuss contact arrangements. As previously suggested, although an increase in contact may be generally beneficial for the child, it might be problematic if the contact occurs in the context of a hostile interparental relationship (Amato et al., 2009). Consistently, it has been found that among boys from divorced families, children's contact with nonresident parents increased children's behavior problems when interparental conflict was high (Amato & Rezac, 1994). Moreover, having partial contact with the father may be associated with additional instability in the child's life, given the child's exposure to potentially meaningful events, such as further transitions in the father's marital status or new partner(s) and/or children living in the father's household; these factors would not add instability for children whose father was completely absent.
We also found that maternal care moderates the link between partial father absence and young adults' romantic relationship satisfaction and sense of identity. This is consistent with previous research attesting to such a moderating effect of maternal relationships (Amato, 1994;Mason et al., 1994;Sandler et al., 2008Sandler et al., , 2013. Two issues concerning this moderating effect should be highlighted, however. First, this effect was found for two of the seven outcomes examined, and thus future replications are warranted. Second, the pattern of this moderating effect was somewhat different from what we had hypothesized. Namely, we expected group differences under low but not high levels of maternal care. We actually found that young adults in the father-presence group reported greater romantic satisfaction and consolidated identity than young adults in the partial father-absence group under high maternal care, whereas under low maternal care, group differences were nonsignificant (see Figures 1 and 2). This pattern suggests that maternal care might have a different meaning for young adults belonging to the partial father-absence group compared with the father-presence group-namely, a risk-related meaning for the former group and a protective meaning for the latter group. Such a pattern is consistent with interactive-synergistic models of vulnerability (e.g., Shahar et al., 2004), according to which the meaning of one risk or protective factor is dependent on the presence of another.
According to Sobolewski and Amato (2007), in cases of divorce or when the parents' relationship is characterized by high conflict, children may experience difficulty feeling emotionally close to both parents without concurrently feeling a sense of disloyalty. The stress felt in these circumstances may lead to a tendency to side with and identify with one parent while emotionally detaching from the other parent. Sobolewski and Amato (2007) proposed that the psychological cost of feeling divided between two parents in dispute may overshadow the benefits of being close to both parents, as exemplified by their findings that children in high-conflict families and divorced families had similar levels of well-being if they were close to both parents or to one parent only. On the basis of this understanding, it might be possible that for individuals in the partial father-absence group (i.e., those who were still in contact with their father but not necessarily on a consistent and stable basis), having a caring and warm mother may actually interfere with their ability to form a consolidated identity and to feel satisfied by their romantic relationships. That is, having high maternal care while also trying to maintain a relationship with their father might be linked to difficult feelings of having to choose between two parents. The psychological cost of such feelings may put them at greater risk of exhibiting difficulties in both romantic relationships and identity consolidation.
As discussed, Sandler and colleagues (2008) showed that while maternal warmth was independently related to lower child externalizing problems, this association differed as a function of interparental conflict and level of paternal warmth; they similarly proposed that a lack of warm child-parent relationship with one parent might influence the child's ability to benefit from the positive relationship with the other parent. Consistently, Sandler et al. (2013) also demonstrated that the associations between positive maternal and paternal parenting and child mental health difficulties were moderated not only by the quality of parenting provided by the other parent but also by the number of overnights children spent with parents, attesting to the importance of having adequate time with at least one parent who provides high-quality parenting.
Our findings also might suggest that when a father is completely absent or when a father is fully present in children's lives, the children can benefit from having a warm relationship with their mother because they are not conflicted about their loyalties to each parent. This interpretation is in accordance with additional findings suggesting that a strong mother-adolescent relationship could serve as a protective factor from the risk of peer problem behavior among adolescents in homes from which the father is absent (Mason et al., 1994). According to Sobolewski and Amato (2007), children in divorced families face a dual risk. On the one hand, trying to be in a close relationship with both parents may prove beneficial, but at the risk of feeling divided and disloyal. On the other hand, having a close relationship with one parent only may result in avoiding the stress of trying to be loyal to both parents, but at the risk of losing the benefits associated with the other parent. The advantages and risks associated with these options might counterbalance one another, leaving these children without any clear benefit in regard to their subjective well-being (Sobolewski & Amato, 2007). This might be the psychological dynamic that is underlying the specific difficulties of children in families from which the father is (partially) absent, as depicted in our study.

Limitations
Limitations of this study should be noted. The first is the cross-sectional design, which limits causal inference. Nevertheless, although father absence or presence came about before the young adults' reports on well-being and romantic relationships, previous research attests to the accuracy and reliability of adults' retrospective recall of their parents' behavior (Brewin et al., 1992). Second, in the current study, we focused on the implications of noncustodial fathers' absence. Future studies might benefit from also including individuals who grew up with both parents as custodial (despite the loss of the father from the household), as well as collecting data regarding the levels of interparental conflict to which the child had been exposed postdivorce. Third, due to a large number of missing values in the paternal PBI (father care and overprotection scores), we were unable to assess the quality of the child's relationship with the father. Similarly, there is lack of information regarding the stability of this relationship. Both aspects should be considered in future studies. Fourth, participants had to be currently involved in a meaningful romantic relationship of at least 3 months' duration, which limits the generalizability of our findings to other populations. In addition, given that no data were collected regarding the length of the relationship, we were unable to assess whether it had any potential confounding effect on our results; further studies should address this important issue. Fifth, the sample was relatively small and predominantly female, heterosexual, and consisted of well-functioning college students. However, this population has been understudied in the literature on father absence. Nevertheless, further research in larger, more heterogeneous samples is needed. Finally, given our relatively small sample, we did not have enough statistical power to examine whether participants' mothers had a significant partner after the divorce and whether this person had been a significant part of participants' lives. This important aspect should also be targeted in future research.

Implications
To the extent that our findings are replicated and extended, we suggest that they bear some relevant clinical and public policy implications. First, the findings attest to the need for routine assessments of children experiencing divorce-based father absence, particularly partial father absence. Similarly, for young adults presenting for treatment due to psychological distress, self-concept maladjustment, or romantic difficulties, it may be useful for the clinician to inquire about the possibility of their having experienced (partial) father absence during childhood and the pain associated with these experiences should be worked through. Additionally, from a public policy perspective, our study raises substantial concerns regarding current postseparation arrangements. Specifically, our findings are not consistent with extant recommendations that children of divorced parents who remain in maternal custody will maintain any amount of contact with their fathers. It appears that the consistency and stability of this contact is more important than its mere existence.