The consequences of work-family conflict in families on the behavior of the child

The balancing act between work and family life can be a challenge that affects both the parents and other members of a family. This study investigates the effect of a parent’s experience of work-family conflict on the behavior of the child. Parental well-being and parenting act as mediators, as previous studies have suggested that this relationship does not run directly. Data from 969 children in the Pairfam database were analyzed using structural equation modelling. The results reveal that both directions of work-family conflict (WIF and FIW) have a negative impact on parental well-being, and only the specific negative spillover from family to work (FIW) influencing parenting performance. In addition, although good levels of well-being and adequate parenting have a positive influence on the behavior of children, the specific spillover effects from work responsibilities to the family (WIF) are apparently unrelated to parenting.


Introduction
Given that work and family are the two central spheres of life for adults, belonging to a family could be expected to come into conflict with participation in the labor market. Work-family conflict can be defined as a difficulty associated with fulfilling the roles of work and family in light of incompatible demands between the two roles (Greenhaus & Beutell 1985). This concern is not an exceptional phenomenon experienced by only a few individuals, and studies have indicated a significant increase in its prevalence in recent decades (Winslow 2005;OECD 2011). Although several studies have confirmed the negative impact of such conflict on aspects including mental health, physical health, and direct relationships on adults (for a review, see Allen et al. 2000;Amstad et al. 2011), few studies have addressed the effect of work-life conflict on the children involved.
The current study is one of the few to examine the effect of work-family conflict on the behavior of children, while accounting for the mediating effects of parental well-being and parenting. The study of these aspects is important, given that work life, family life, and the functioning of the child are three inextricably connected aspects that are decisive in determining our everyday life. Especially with regard to the functioning of the child, theoretical and empirical evidence suggests that individuals interact with their environments, which shape their development (Bronfenbrenner 1986). Scientific insights into the possible conflictual interaction between a parent's work and family lives, which could consequently affect the child involved emphasize the need for child and parental care at the levels of policy, institutions, and practice.
Previous studies on the consequences of work-family conflict have focused largely on the adults involved. A few studies have connected characteristics of work and family characteristics to the development of children. According to these authors, this effect is not direct, but operates through a linking mechanism, in which the main effect is mediated by parental characteristics (Strazdins et al. 2013;Bauer et al. 2012;Strazdins et al. 2006;Vahedi et al. 2018). Such general findings have been strengthened by more recent longitudinal research on the long-term consequences of these difficulties in combining work and family can have on all family members (Cho & Tay 2016;Huang, Wang, & Warrener 2010;Dittrich et al. 2018;Vahedi et al. 2018).
First, existing evidence suggests that the experience of work-family conflict has a negative impact on parental well-being and parenting performance. Difficulty combining work and family has been associated with decreased levels of mental health (Sharma, Dhar, & Tyagi 2016;Yang 2015;Fiksenbaum 2014;Aycan & Eskin 2005;Dinh et al. 2017), less satisfaction with life (Cho & Tay 2016;Aycan & Eskin 2005), greater irritability and less warmth in interactions with children (Cooklin et al. 2015;Bauer et al. 2012;Strazdins et al. 2006;Dinh et al. 2017), and less satisfaction with performance in the parental role (Aycan & Eskin 2005).
Second, child behavior is affected by the well-being and parenting performance of parents. Scientific literature provides clear evidence that a child's behavior is influenced by many aspects (e.g., biological foundation, direct relationships, housing environment) at a variety of levels (e.g., psychological, social, physical development). The immediate environment that is created and shared by the parents is of considerable importance to the development of the child. In this regard, previous studies have indicated that parents with mental health problems (e.g., depression, anxiety, and distress), lower overall well-being, and less warmth and sensitivity in parenting attitudes have a negative influence on the behavior of the child (Strazdins et al. 2013;Huang, Wang, & Warrener 2010;Keyser, Ahn, & Unick 2017;Dittrich et al. 2018;Ong et al. 2018;Dinh et al. 2017).
The aim of the current study is to provide new insight into the influence of workfamily conflict on the behavior of children. We demonstrate that this effect is mediated by parental well-being and parenting performance. A clear overview of these mediating pathways is crucial to understanding the complex, long-term processes that guide parentchild interactions in contemporary families. Furthermore, we distinguish between two types of conflict that can be experienced and examine possible differences in their relationships with mediators and outcomes. This study examines quantitative data from the Pairfam dataset, which were collected in Germany between 2015 and 2018, and which include detailed information about demographic aspects, intimate relationships, family dynamics, characteristics of work and family, and child development.

Theoretical background
In investigations of the consequences of work-family conflict on the behavior of children, scholars have argued that this relationship does not operate directly (Strazdins et al. 2013;Bauer et al. 2012;Strazdins et al. 2006;Vahedi et al. 2018). The conceptual model tested in the current study is illustrated in path diagram presented in Figure 1, which depicts workfamily conflict as affecting the child's behavior through parental well-being and parenting.
Work-family conflict can occur when an individual is confronted with a clash between the expectations relating to the roles of work and family. A definition of work-family conflict has been provided by Greenhaus and Beutell (1985: 55): "a form of inter-role conflict in which the role pressures from the work and family domains are mutually incompatible in some respect." Within this general conceptualization, scholars investigating the workfamily relationship have drawn a distinction between two directions of conflict (Frone, Russell, & Cooper 1992;Grzywacz & Marks 2000). First, work-interfering-with-family (WIF) involves a spillover from work responsibilities to the family, e.g., when a mother is late picking up her children after school due to attending a meeting at work. Second, family-interfering-with-work (FIW) occurs when family responsibilities interfere with workrelated activities, e.g., when a mother cancels a meeting at work to be able to pick up her children from school. In general, people report more WIF than FIW (Burke & Greenglass 2001;Frone, Russell, & Cooper 1992;Gutek, Searle, & Klepa 1991;Eby et al. 2005), and several researchers have observed that, although they are strongly positively related, their unique determinants and consequences render them distinct types of conflict. For example, evidence suggests that work characteristics (e.g., working hours, work support) are more strongly related to WIF, whereas family characteristics (e.g., number of children, family involvement) are more strongly associated with FIW (Byron 2005;Eby et al. 2005;Ford, Heinen, & Langkamer 2007;Amstad et al. 2011). More specifically, with regard to the possible outcomes of experiencing work-family conflict, the meta-study by Amstad et al. (2011) reports consistent results indicating that both directions of conflict are more strongly related to within-domain consequences than they are to cross-domain consequences, as characteristics of the work or family domain are more salient to the direction of conflict to which they apply. Given that the mediators in the current research are personal (parental well-being and child behavior) and family characteristics (parenting behavior), we expect stronger relationships between FIW and the mediators, especially for parenting behavior.
In turn, parental well-being and parenting are regarded as factors that affect the behavior of children. An individual's well-being can be defined as a subjective evaluation of general psychological, social, cognitive, and physical health, which can be associated with various outcomes, both negative ones (e.g., internalizing and externalizing problems) and positive ones (e.g., life satisfaction and self-esteem) (Diener & Diener 2009;Bastaits, Ponnet, & Mortelmans 2014). As such, a person's subjective well-being is not reflected by any single dimension. This is acknowledged in the Pairfam database, which includes several valid scales addressing this issue (Thönnissen et al. 2019). The current study focuses on the psychological well-being of a parent. Various studies have supported the interrelationship between well-being and parenting, with findings indicating a stronger effect from well-being on parenting (Jackson et al. 2000;McCarty & McMahon 2003;Conger et al. 2002). According to Darling and Steinberg (1993), parenting behaviors are a key element of the parenting style applied. They define a parenting style as "a constellation of attitudes towards the child that are communicated to the child and that, taken together, create an emotional climate in which the parent's behaviors are expressed" (Darling & Steinberg 1993: 488). They further distinguish two central parental behaviors: support (i.e., affection and responsiveness) and control (i.e. supervision) (Baumrind 1991).
In theoretical terms, the pathways toward the behavior of a child can be explained by socio-ecological theories, which are rooted in developmental psychology. The essence of these theories is the dynamic interaction and interrelationship between the child and the environment. According to Bronfenbrenner's ecological system theory (1986), understanding a child's development requires taking into account the ecological system surrounding the child. At the center is the child's own biological and psychological predisposition, which is surrounded by four interrelated systems: the microsystem, the mesosystem, the exosystem, and the macrosystem. The microsystem is composed of the direct social environment (e.g., family, peers, and school). The mesosystem consists of the interactions between the direct interpersonal relationships of the microsystem. The exosystem comprises the broader social environment, which affects the child by determining the conditions and settings in which the child interacts (e.g., politics, media, and social services). The macrosystem consists of the attitudes, ideologies, and values of the culture. This theoretical perspective clarifies the transfer of shared influential contexts and social capital between parents and children through the well-being and parenting skills of the parent (Coleman 1988). The socio-ecological system in which the parent participates (e.g., work and family) is strongly related to the parent's own functioning. Evidence suggests that this system interacts with the child's system, and that it can affect the child's development. Acknowledging the role stress associated with work-family conflict and the bi-directional interactions between the child and the family environment, a child's behavior is regarded as being in-directly influenced by the parents' experiences of work-family conflict, through the mediation of parental well-being and parenting.

Figure 1:
Path diagram of the conceptual model

Empirical evidence
One important question concerns the existence of a connection between work-family conflict and the behavior of the child. Understanding the development and behavior of children requires considering the direct (and broader) contexts within which they are embedded. It has been stated that the work and family contexts of the parents contribute to the environment they share with their children. This environment influences the mental and physical health of a child through family interactions (Roeters, Van der Lippe, & Kluwer 2010;Crouter & Bumpus 2001). For example, if both parents are employed and experience stress in balancing the demands of work and home, this might create a less healthy family food environment, which they share with their children. Such an environment could be characterized by less frequent family meals, lower consumption of fruits and vegetables, and more frequent fast-food intake, ultimately resulting in less healthy conditions for the child (Bauer et al. 2012). In addition, with regard to the mental development of children, empirical evidence has acknowledged the interrelationship between children and their parents. According to the results of a study by Strazdins et al. (2006), parents with nonstandard work schedules report more difficulties with family functioning, more depressive symptoms, and more ineffective parenting, all of which are associated with emotional and behavioral difficulties in children. In another study, Strazdins et al. (2013) specifically investigate risks to the mental health of young children (in terms of emotional and behav-ioral problems) whose parents experience conflict between the roles of work and family. Based on these results and the stated theoretical framework, our proposed model links the work-family conflict experienced by parents to the behavior of their children. In light of the findings discussed above, we hypothesize that higher levels of work-family conflict (both WIF and FIW) experienced by a parent are negatively related to the behavior of the child (H1). Second, the effect of work-family conflict on the behavior of the child is likely to be mediated by the well-being of the parents. Researchers have consistently reported that the experience of work-family conflict has a negative impact on an individual's well-being. Within the broad concept of well-being, several different and related dimensions have been addressed as consequences of work-family conflict. Elevated stress levels, lower psychological health and well-being, less satisfaction with life, subjective well-being, and poorer mental health (Sharma, Dhar, & Tyagi 2016;Yang 2015;Cho & Tay 2016;Aycan & Eskin 2005;Amstad et al. 2011) are associated with work-family conflict. These results thus suggest that conflict between the roles of work and family affect an individual's personal balance and resources, which can cause stress and ultimately decrease well-being. From a psycho-social perspective, studies have examined the association between the functioning of parents and children within the shared home context. Huang, Wang, and Warrener (2010) report that stressful home situations in which the mother has experienced domestic violence can have a long-term impact on the external behavioral problems of the child, through the mediator of maternal mental health. Similarly, in a study focusing more specifically on the effect of maternal depression, Keyser, Ahn, and Unick (2017) report more behavioral problems in infants whose mothers were depressed. Dittrich et al. (2018) report similar results, but emphasize the long-term effects of maternal depression on the child's quality of life. In their study, a maternal history of depression decreased the child's quality of life, with maternal stress and sensitivity acting as mediators. Building upon these empirical and theoretical arguments, we hypothesize that parental well-being mediates the relationship between work-family conflict (both WIF and FIW) and the behavior of the child (H2).
Third, work-family conflict is assumed to be indirectly linked to a child's behavior through parenting performance. Children raised in families struggling to balance work and family obligations may feel spillover effects from elevated stress levels and the limited time and energy that their parents have for demonstrating good and appropriate parenting techniques. Although a vast volume of literature has investigated the negative consequences of work-family conflict, fewer studies have focused specifically on the effects of such conflict on parenting. In a study by Cooklin et al. (2015), parents experiencing workfamily conflict responded with significantly less warmth and affection, in addition to being more irritable in their interactions with their children. In terms of subjective satisfaction with parental role performance, it has been suggested that parents who experience conflict between their work and family roles are less satisfied with their performance in the parental role (Aycan & Eskin 2005). The experience of such conflict could thus be expected to undermine their parenting capacity, consequently affecting the functioning and behavior of the child through the shared home environment. Previous studies have focused especially on malfunctioning families, reporting that parenting practices have direct and long-term effects on the behavior of children (Huang, Wang, & Warrener 2010; Keyser, Ahn, & Unick 2017). The importance of good parenting to the general development of a child has also been highlighted by its effect on various child-related outcomes, including socio-emotional behavior, physical health, and psychological well-being. These results have been found across different cultures and age groups (Ong et al. 2018;Dittrich et al. 2018;Roman et al. 2015). Taking these findings into account, we hypothesize that parenting mediates the relationship between work-family conflict (both WIF and FIW) and the behavior of the child (H3).

Data
To test our hypotheses, we analyzed data from the Pairfam (Panel Analysis of Intimate Relationships and Family Dynamics) study. This large-scale, longitudinal, multi-actor study is sponsored by the German Research Foundation (DFG), and the database includes information on partnerships and family dynamics in Germany. Nine waves are currently available, with the first wave having been collected in 2008 and 2009. Subsequent waves have been added biannually. Each wave includes around 12,000 individuals ("anchor respondents") from three birth cohorts: 1971-73, 1981-83, and 1991-93. The anchor respondents are randomly selected from all individuals living in private households in Germany. The anchor respondent's partner is also asked to participate in the study and, beginning with the second wave, the parents and a child/children of the anchor respondent have been included as well. Participating children are between the ages of 8 and 15 years (children 15 years of age and older become anchor respondents in their own right).
For the current study, we selected a subsample of the dataset. This subsample contains data from 969 children included in the dataset of children (nboys = 496, ngirls = 473), along with 384 fathers and 585 mothers. Given our interest in the effects of factors on the behavior of the child -while accounting for the possible presence of reversed causalityinformation on work-family conflict, and both mediating variables were obtained from Wave 8, with data concerning the child's behavior taken from Wave 9. This allows us to rule out the possibility that the investigated child's behavior was affecting the parent's work-family conflict.

Analytical strategy
To test our model, we applied a framework of structural equation modeling (SEM) (Hatcher 1994;Bollen 1989). More specifically, we use a family of models within the SEM framework that allow for a higher-order factor structure. Two alternatives have been suggested in the literature: higher-order factor models and the bifactor model (Mulaik & Quartetti 1997). Given that the bifactor model assumes orthogonal associations between factors (an assumption that is too strict for our conceptual model), we opted to use a second-order hierarchical SEM model (Chen, West, & Sousa 2006). In this model, we start by including first (lower) order factors that together constitute a second (higher) order factor. In a preparatory phase, all first-order factors were tested according to exploratory factor analysis before being entered into the SEM model 1 , which was estimated in two steps. In the first step -the measurement model -we omitted all directional paths and focused on the first-order and second-order measurement of all latent concepts. We then estimated the structural model, which we used to test our hypotheses. The measurement model had a good fit (RMSEA = 0.0455; CFI = 0.86). It was not necessary to add any error covariances between first-order items to obtain an acceptable fit. The structural model also fit the data well (RMSEA = 0.049; CFI = 0.84).
The composition of all lower-order factors is discussed in the Measures subsection. We calculated the composite reliability of all lower-order and higher-order factors (comparable to Cronbach's alpha within the framework of exploratory factor analysis) (Hatcher 1994). Of the higher-order factors, only Parenting (composite reliability = 0.27) failed to achieve the lower threshold of 0.70. The other two higher-order factors had high and significant factor loadings, with composite reliability values of 0.76 (Parental well-being) and 0.75 (Child behavior). The fact that the higher-order Parenting factor failed to reach the 0.70 threshold can be explained in part by the fact that it is composed of only two lower-order factors. For this reason, and because the model fit the data very well, we decided to retain this factor in the final model.
In line with the proposed conceptual model and associated hypotheses, the mediating effects are tested in the structural model in three steps. First, we test only the direct effect of both types of work-family conflict on the behavior of the child. Second, we add the direct effect between the two mediators. Third, we test parental well-being and parenting as mediators in the full model.

Measures
WIF and FIW. The Pairfam Work-Family Conflict scale was developed specifically for the Pairfam study, based on the instrument developed by Carlson, Kacmar, and Williams (2000). This scale assesses an individual's subjective evaluation of the conflict that they experience between work and family in both directions. Individuals rate four items relating to WIF (comp. reliability = 0.78) and four items relating to FIW (comp. reliability = 0.72) along a five-point scale ranging from "Not at all" to "Absolutely." The respondents' ratings indicated the extent to which specific statements applied to them. The following is an example of an item measuring WIF: "My work prevents me from doing things with my friends, partner, and family more than I'd like." The following is an example of an item measuring FIW: "Because I am often under stress in my private life, I have problems concentrating on my work." A higher-order "Work-family conflict" factor comprising these two factors is not included in this model, since the two types of work-family conflict are theoretically regarded as two separated concepts. The correlation between the two factors was 0.45.
Parental well-being. The higher-order Parental well-being factor consists of four subscales proposed by the Pairfam dataset: Self-esteem, Depressiveness (reversed), Overload (reversed), and Activity. Self-esteem was originally measured according to a short version of the Rosenberg Self-Esteem Scale (2015), which consists of three items. After the exploratory 1 Output not shown, but available from the first author on request.
factor analysis, we could retain only two items with five response categories ranging from 1 (Not at all) to 5 (Absolutely). The following is an example item: "All in all, I am pleased with myself." The measurement model yielded high and significant factor loadings, and the composite reliability of this subscale was 0.90. The measurement for Depressiveness was derived from the State-Trait-Depression scales (Spaderna, Schmukle, & Krohne 2002), and it originally consisted of 10 items. The items for this factor (Depressive symptoms) were recoded into a positively worded factor. The exploratory factor analyses pointed to a scale consisting of six items to be included in the SEM model and achieved a composite reliability of 0.85. Respondents answered all items on a four-point Likert scale, ranging from "Almost never" to "Almost always." The following is an example of a question concerning depressive symptoms: "My mood is melancholy." The Overload subscale is a reversed factor based on the Perceived Stress Questionnaire (Levenstein et al. 1993). A short, threeitem version was developed, with respondents being asked to use a five-point Likert scale ranging from "Not at all" to "Absolutely" to indicate their feelings (e.g., "overburdened") in the past four weeks. In the model, the composite reliability of the scale reached 0.89. The Activity subscale was originally part of the German sensitivity (Befindlichkeit) scales (Abele-Brehm & Brehm 1986). This three-item scale had the same five-point response format as the overload scale, and all three of these items were included in the measurement model with a composite reliability of 0.84. The scale investigates the respondent's level of activity by asking about their feelings (e.g., "full of energy") during the past four weeks.
Parenting. The higher-order Parenting factor consists of two lower-order scales: Negative communication (reversed) and Monitoring. The Negative communication subscale measures the parent's degree of negative behavior toward the child (Schwarz et al. 1997). Respondents used a five-point Likert scale ranging from "Never" to "Very often" to indicate how often they had exhibited particular behaviors (e.g., "yell[ing] at the child because he/she did something wrong"). We reverse-coded the negative items to obtain all positive responses for this factor. The confirmatory factor analysis yielded factor loadings greater than .40 for all items and a composite reliability of 0.74. The Monitoring subscale measures the extent to which respondents were informed about the child's activities and social contacts, based on the Expanded German Version of the Alabama Parenting Questionnaire (Reichle & Franiek 2005). The scale consists of three items, with response options ranging from 1 (Never) to 5 (Very often). Respondents were asked to indicate the frequency of particular behaviors (e.g., "When your child goes out, you ask what he/she did and experienced"). The composite reliability of this factor in the measurement model was 0.76.
Child's behavior. The behavior of the child in question was measured according to the Strengths and Difficulties Questionnaire (Goodman 1997;Woerner et al. 2002). The questionnaire was presented to parents concerning the behavior of their children, as well as to children about their own behavior. The current study is based on the information provided by the child. This questionnaire was originally composed of five subscales indicating the child's overall well-being and behavior. Based on the results of factor and reliability analyses, however, only four could be retained in the measurement model: Prosocial behavior, Emotional symptoms (reversed), Conduct problems (reversed), and Hyperactivity (reversed). All subscales have the same response format, ranging from 1 (Not true) to 3 (Certainly true). The Prosocial behavior subscale consists of four items (e.g., "I try to be nice to other people. I care about their feelings"; composite reliability = 0.68). The Emotional symptoms subscale is assessed according to such questions as "I have many fears, I am easily scared," is reversed and consists of five items (comp. reliability = 0.68). The Conduct problems and Hyperactivity subscales consist of three items each and are reverse-coded into positive items for analysis and interpretation. The following is an example of a question from the Conduct problems subscale: "I take things that are not mine (from home, school or elsewhere)." The following is an example from the Hyperactivity subscale: "I am easily distracted; I find it difficult to concentrate." The composite reliability of the Conduct problems subscale was relatively low (0.41), while the reliability of the Hyperactivity subscale was high (0.77).
Control variables. Previous studies have identified a number of demographic, work, and family-related characteristics that are associated with the consequences of experiencing work-family conflict. The most relevant and significant control variables were selected for this study: age of the youngest child (M = 8.01, SD = 3.43), number of children in the household (one child, n = 101; two or more, n = 868), household income (M = 3870.75, SD = 1551.00), parent's gender (384 fathers, 585 mothers), and whether the parents were divorced (902 not divorced, 67 divorced).

Descriptive results
We started by calculating summed scores for the final factors tested in the structural model. For both WIF and FIW, the summed scores range between 4 (no conflict) to 20 (a lot of conflict). Overall, the subjects in our sample scored higher on WIF (M = 8.90, SD = 3.50) than they did on FIW (M = 6.54, SD = 2.56). The average score for parental wellbeing was relatively high (M = 46.19, SD = 6.81), with a minimum of 13 (very low wellbeing) and a maximum of 64 (very high well-being). The average summed score for parenting behavior was very high (M = 23.46, SD = 2.72), with a minimum of 6 (highly negative parenting) and a maximum of 30 (highly positive parenting). Similarly, the results revealed a high average summed score for child behavior (M = 22.95, SD = 3.99), ranging from 15 (very negative child behavior) to 45 (very positive child behavior).

Measurement model: Higher-order factors
The standardized factor loadings in the measurement model for each higher-order factor are presented in Table 1, along with the standard errors and two-tailed p-values. Within each higher-order factor, all factor loadings are significant and greater than .40, with the exception of the lower-order factor Negative communication (reversed) and Prosocial behavior. As stated previously, the good fit of the measurement model and theoretical importance of these lower-order factors prompted us to retain both in the structural model.

Structural model
The structural model tests the hypothesized paths between WIF, FIW, and the higherorder factors. The standardized factor loadings, standard errors, and two-tailed p-values of each path are presented in Table 2. Contrary to our expectations (H1), the behavior of the child was not directly affected by a parent's experience of either WIF or FIW. Instead, our results suggest that the influence of work-family conflict is mediated by the other higherorder factors. In line with our second hypothesis (H2), both WIF and FIW were significant predictors of parental well-being. Parents who experienced more conflict due to workinterfering-with-family (WIF: β = -.33, p < .001) and due to family-interfering-with-work (FIW: β = -.38, p < .001) scored significantly lower on the Well-being factor. In turn, higher levels of parental well-being were associated with more positive behavior on the part of the child (β = .15, p < .01). The relationship between work-family conflict and the behavior of the child is thus mediated by parental well-being, thereby confirming our second hypothesis.
With regard to parenting, the analysis indicates that parents experiencing FIW are significantly less likely to exhibit positive parenting behavior (β = -.13, p < .05). With regard to WIF, the results did not provide evidence of a direct effect on parenting, although higher levels of parental well-being (which is negatively affected by WIF) were associated with better parenting styles (β = .30, p < .001) 2 . These results thus suggest that WIF affects parenting indirectly. In turn, parenting has a significant positive effect on a child's behavior (β = .16, p < .01). Taking into account the indirect effect of WIF on parenting, the results provide partial support for our third hypothesis (H3), which predicts that parenting mediates the relationship between work-family conflict and the behavior of the child.
To perform an accurate test of the mediating effects of parenting and parental wellbeing, we fitted the structural model in three steps. First, we considered only the direct effects of WIF and FIW on the child's behavior (along with the control variables), taking into account the direction, strength, and significance of these direct effects, along with the model fit. We then added the direct effect from parental well-being to parenting. in addition to strengthening the directs effects of WIF and FIW -which became more strongly negative and more significant -this reduced the RMSEA from 0.0534 to 0.0488 and increased the CFI from 0.81 to 0.84. The significance of this enhancement was tested and confirmed with a chi² test (Δχ² = 480, Δdf = 1). In the third and final model, we added the mediated effects of FIW and WIF on the child's behavior, through parental well-being and parenting. In light of this mediation, the direct effects of FIW and FIW became insignificant and were fully explained through parenting and parental well-being (see Table 2 below). In this case as well, a chi² test indicated a significant improvement in the model fit (Δχ² = 42.22, Δdf = 6). We can therefore conclude that the mediation effects are indeed crucial to understanding the antecedents of children's behavior. 2 A robustness check was performed to control for reversed causality (i.e., whether it had been parenting style that affected well-being. Comparison of uni-directional models revealed that the effect of parenting style was weaker than that of well-being. In contrast, a reciprocal (non-recursive) model indicated that the effect of well-being strongly overpowered that of parenting. As such, we consider only the effect of well-being in the model. Output is available from the corresponding author on request.
The results displayed in Table 2 are visually depicted in the complete path model presented in Figure 2. According to our analysis, WIF and FIW explain 27% of the variance in parental well-being, with 32% of the variance in parenting being explained by FIW, WIF, and parental well-being. In turn, the two lower-order and two higher-order factors together explain approximately 9% of child behavior.

Discussion
Balancing work and family life poses a challenge for many parents, as the responsibilities associated with each of these roles demand a substantial amount of time and effort. Previous studies have demonstrated that, in addition to affecting the well-being and functioning of the parent, the experience of work-family conflict can have negative spillover effects for all family members. The current study investigates the effect of parental experiences of work-family conflict on the behavior of a child. Previous research and theoretical insights have suggested that this relationship does not run directly, but that the shared and influential environment surrounding the family members should be taken into account. The present study is the first to introduce parental well-being and parenting as mediators in this pathway. First, our results do not confirm our expectation that higher levels of work-family conflict (both WIF and FIW) experienced by a parent have a negative effect on a child's behavior (H1). This apparent absence of such a direct relationship suggests that mediating variables play a crucial role. Moreover, the insignificance of this direct effect in the full model indicates that any differences are fully mediated by parental well-being and parenting. These mediators thus explain the entire relationship between WIF/FIW and the behavior of the child.
Second, we predicted that parental well-being would mediate the relationship between work-family conflict (both WIF and FIW) and the child's behavior (H2). This expectation was based on the well-documented long-term, direct negative impact of work-family conflict on the mental health and well-being of the parent, which in turn has a negative effect on the functioning of their children (Aycan & Eskin 2005;Cho & Tay 2016;Cooklin et al. 2016;Dinh et al. 2017). This expectation was further supported by the theoretical mechanism underlying these relationships with regard to the immediate environment that is created and shared by the parents. Consistent with our expectations, the results revealed that higher levels of both WIF and FIW have a negative impact on parental well-being and that higher levels of parental well-being are associated with better behavioral statements. It would be plausible to expect that conflict between the roles of work and family might abate the possibility and capability of parents to cope with such incompatible demands. In turn, this is likely to strain parental well-being and, ultimately, the functioning of children through the effects of their shared physical and relational home environment. Results from a recent study by Dinh et al. (2017) provide evidence to support this effect, emphasizing the longitudinal and intergenerational scope of this issue. According to that study, parental well-being and parent-child interaction mediate this pathway. In addition, the initial or chronic experience of work-family conflict is associated with an increase in mental health problems for children, and reductions in work-family conflict lead to reductions in mental health problems in children.
Third, our results do not provide full confirmation for the hypothesis that good parenting mediates the relationship between work-family conflict (both WIF and FIW) and the behavior of the child (H3). With this hypothesis, we argued that both the interference of work with family and the interference of family with work would have a negative effect on parenting performance by engendering a sense of pressure on available time and energy, thereby elevating stress levels (Cooklin et al. 2015;Strazdins et al. 2006). We further predicted the existence of a positive relationship between parenting performance -which contributes to the composition of the family environment and interactions -and the behavior of the child (Dinh et al. 2017;Keyser, Ahn, & Unick 2017;Ong et al. 2018). The results provide confirmation for all of the predicted paths, with the exception of the relationship between WIF and parenting. As such, the spillover effects of work responsibilities to the family are apparently unrelated to parenting styles. One possible reason for this finding could be related to the differentiated associations of WIF and FIW. For example, WIF is more strongly related to work than it is to family characteristics, while FIW is more strongly related to family outcomes (e.g., parenting) than it is to work characteristics (Byron 2005; Eby et al. 2005;Ford, Heinen, & Langkamer 2007;Amstad et al. 2011). The outcomes/consequences relating to work (e.g., work performance) or family characteristics (e.g., couple relationship) were most strongly affected by the source of conflict. Applied to the current study, this means that impaired parenting performance is a direct and dominant reaction to the domain in question (i.e., the family), and it is therefore more strongly related to FIW than it is to WIF. Furthermore, the positive association between parental well-being and parenting indicates the existence of an indirect path from WIF (which has a negative impact on parental well-being) to parenting style. Although this could be an alternative explanation for the absence of a direct effect, further investigation is needed.
Although this study provides new insights, it is subject to several important limitations. First, we were not able to provide long-term insight into this phenomenon. This is because several variables of interest were included only in later waves, while others were not included in every wave. In addition, although we do take into account the possibility of reciprocal associations between our variables, this issue still remains unclear. Recent studies provide evidence primarily for reciprocal associations, although there is also support for the unidirectional scenario (e.g., between work-family conflict and the mental health of the child). This indicates the existence of a complex interplay between all variables, as well as the need for appropriate interventions in policy and practice (Vahedi et al. 2018;Westrupp et al. 2016). Another limitation of the current study has to do with our use of self-reports from parents about their well-being and parenting, along with self-reports from children about their own behavior. This may have introduced a certain level of socialdesirability bias. Finally, in this study on conflicts between work and family, we explicitly distinguish between the two directions of work-family conflict. This assumes that the competing demands of work and family inevitably lead to problems. As various scholars have noted, however, work and family can also enhance each other (Greenhaus & Powell 2006). Although this was not the focus of the current study, it could offer an alternative point of view for future research concerning the specific positive or negative implications of the enriching potential of work and family roles.
In conclusion, this study makes a meaningful contribution to the existing literature by offering new insights into the link between parental experiences of work-family conflict and the behavior of their children. It is one of the first studies to investigate this pathway. Strengths of the study include its acknowledgement of the mediating role played by parental well-being and parenting, and the distinction between the specific consequences of WIF and FIW. Our findings provide evidence that experiencing conflicting work and family roles has a negative impact on parental well-being and parenting performance. In addition, parents with a healthy level of general well-being and adequate parenting have a posi-tive influence on the behavior of their children. Our results nevertheless provide no evidence that the specific spillover effects of work responsibilities to the family (WIF) are related to parenting. As the two most important spheres of life -work and family -are likely to come into conflict with each other, the impact of such conflict on the well-being and functioning of mothers and fathers should not be neglected. The impact of this powerful interface in shaping the social environment has critical intergenerational implications for the long-term functioning and development of children. Additional research is needed in order to disentangle these equations and provide insight into the vulnerability of children in imbalanced parental work-family environments. Such knowledge could be used to inform adequate policy and practice arrangements that could reduce potential conflicts between the roles of work and family. Elements of such policies and arrangements could include flexible working hours, job security, family-friendly working environments, and access to family-leave schemes (e.g., maternity/paternity leave and parental leave).

Information in German
Deutscher Titel