Heterogeneous Effects of Family Complexity in Childhood on Mental Health: Testing the “Good Divorce” and the “Good Stepparent” Hypotheses

We address two key research questions. First, is there an association between parental separation and living with a stepparent in childhood and maladjustment in adulthood? Second, we examine the role conflict plays in these associations. We study whether parental union dissolution is only detrimental in cases of heightened post-divorce interparental conflict. We build upon this “good divorce” hypothesis by considering whether the possible association between living in a stepfamily and depressive symptomatology is only present in cases of high stepparent-child conflict (a “good stepparent” hypothesis). Using data from the OKiN survey (Parents and Children in the Netherlands), we analyze the self-reported depressive feelings of Dutch adults aged 25–35. Of our sample, 2233 adults experienced parental separation in childhood (on average, 22 years before data collection); of those, n = 1665 had lived with a stepparent. Our findings clearly indicate that having experienced parental divorce is associated with an increase in depressive symptoms only for those adults who were exposed to heightened post-divorce interparental conflict. Similarly, living with a stepparent is linked to maladjustment only in cases of high stepparent-child conflict. Importantly, we find evidence that a low-conflict stepfather-child tie could even buffer against maladjustment (which is not the case for a low-conflict stepmother-child tie).


Introduction
As the instability of adult intimate unions has grown in recent decades, there has been ongoing scientific and popular interest in the impact of parental partnership dissolution (referring both to marital and non-marital cohabiting unions) on children (Amato 2010;Amato and James 2010;Wang and Amato 2000). Much has been written about the effects of divorce on a range of short-and long-term child outcomes, such as psychosocial adjustment (Sands et al. 2017), educational attainment (Bernardi and Radl 2014;Brand et al. 2019), and (adult) children's own experiences with intimate partnerships (Amato and DeBoer 2001;Ivanova et al. 2011;Wolfinger 2011). Considerable efforts have been made by researchers in this field to assess the magnitude of and the heterogeneity in the possible disadvantages children of divorce might be facing, and to identify the mechanisms that drive the potential association between parental separation and maladjustment (Härkönen 2014;Kim 2011). Importantly, the initial general consensus that family dissolution has negative effects has evolved into a recognition of the difficulties associated with distinguishing the specific effects of parental separation (i.e., the moment when one of the parents leaves the parental household) from the complex interplay of events and processes that lead up to and are precipitated by that family transition (Härkönen 2014). In other words, viewing divorce as an event might be an unjustified oversimplification, given that it is often preceded by the longer process of the disintegration of the relationship, and is followed by the need to negotiate relations between multiple single or stepparent households.
The importance of post-divorce relations for the adjustment levels of the children involved has been highlighted by the notion of the "good divorce" (Ahrons 1994(Ahrons , 2007Ahrons and Tanner 2003). The argument made by proponents of this concept is that as long as "parents − as they did when they were married − continue to be responsible for the emotional, economic, and physical needs of their children" (Ahrons 1994: 3) and maintain a cooperative relationship, any negative consequences of their separation can be minimized. In her work, Ahrons developed a typology of ex-couples based on the nature of their interactions after separation, and reported that about half of the ex-partners could be classified as having high-quality communication. However, the claim that the children of these couples could be successfully sheltered from the negative repercussions of the parental separation has certainly been challenged (e.g., Amato et al. 2011). Furthermore, studies that have paid closer attention to the timing of separation have shown that the vast majority of couples with children report having antagonistic contact, especially in the years immediately following the divorce (Fischer et al. 2005). In other words, although there certainly are some "good divorces," a substantial number of children are still exposed to a hostile interparental relationship in the years following the separation.
Given that the majority of parents repartner after the dissolution of their union (Thomson 2014), children often grow up not only with two ex-partners as parents, but also with one or two stepparents. Therefore, much attention has been paid to the effects of living in a stepfamily on child well-being. The findings of these studies have been rather mixed, with some authors reporting that living in a stepfamily has positive effects on children's adjustment levels, while others have found either negative or no effects (Sweeney 2007; for a review, see Sweeney 2010). One reason why these findings differ (in addition to the wide range of outcomes studied) is that the comparison groups used were not consistent. For example, whereas some researchers compared children in stepfamilies with children in their original, twoparent families, others compared single-parent and stepparent households. It is, however, important to note that there is considerable heterogeneity among stepfamilies, as some stepfamilies are stable and cooperative, while others are characterized by conflict. This variability should matter for child outcomes as well.
In this chapter, we contribute to the ongoing debate about the concept of the "good divorce," and examine the levels of depressive feelings reported by young adults who experienced parental separation in childhood. We begin by addressing the question of whether growing up with separated biological parents can be linked to higher levels of maladjustment in young adulthood. We then examine whether any association found in the first part of our study might be driven by the quality of the post-divorce tie between the ex-partners. In other words, we explicitly look at whether separations that are not followed by elevated levels of conflict are detrimental to the long-term adjustment of children. We go a step further than earlier investigations of the "good divorce" hypothesis by also considering the quality of a young adult's stepparent-child relationships in childhood, and proposing a "good stepparent" hypothesis. It should be pointed out that when we refer to "effects" in the analyses and in the title, the term is meant in a statistical sense, as it is clear that these effects are only approximations of the causal effects that a divorce may have.
In our work, we utilize data from the recently collected OKiN survey (Ouders en Kinderen in Nederland; Parents and Children in the Netherlands; Kalmijn et al. 2018), which was specifically designed to address questions about the long-term repercussions of family complexity. We focus on young adults (aged 25-35) born in the 1980s who were not living with their two biological parents by age 15. Like many other Western countries, the Netherlands has experienced a marked increase in divorce since the 1960s. Therefore, the cohorts of young adults included in our analyses are among the first group of children who were affected by the Dutch "divorce revolution" while growing up -although they are certainly not the first cohorts to experience the impact of that substantial shift in matrimonial patterns.

Parental Separation and Post-divorce Interparental Conflict
Considerable efforts have been made to identify the mechanisms that might explain the association between divorce and the adjustment levels of individuals. In particular, many authors have closely examined the question of whether there is negative selection into divorce (for example, based on socioeconomic status; Gennetian 2005; Ginther and Pollak 2004;Grätz 2017). In other words, scholars have been attempting to determine the extent to which the link between divorce and (child) maladjustment is causal or spurious. A number of researchers have indeed found that the magnitude of the effect of parental divorce on (adult) children's well-being decreases after a range of characteristics that might function as confounders are taken into account. Moreover, some authors have used person fixed-effect models to show that there is no association between changes in family structure and changes in behavioral problems for children (Aughinbaugh et al. 2005). However, the majority of studies that have looked at this question have found that at least some of the negative effects of marital separation on individual adjustment persist after accounting for selection (Amato 2010;Amato and Sobolewski 2001;Kim 2011). Therefore, in line with previous work, we expect to find that when we compare adults whose parents separated while they were children to adults with continuously married parents, the well-being of the adults in the former group will be lower.
In addition to differing on methodology, scholars have debated the theoretical mechanisms that underlie this association, which might include the loss of resources (e.g., financial resources, but also parental time and ability to provide support) following parental separation, and the stress precipitated by the family transition. Whereas the resource model has often been used in investigations of how child educational attainment is affected by parental divorce (Bernardi and Boertien 2016;Bernardi and Radl 2014;Jonsson and Gähler 1997;Thomson et al. 1994), the stress model has frequently been applied in studies of child and parental psychological well-being following separation (Lansford 2009;Pearlin et al. 2005). Importantly, evidence of the detrimental effects of the stress precipitated by family transitions has led some researchers to argue that a child's level of adjustment may be highly dependent on the level of interparental conflict (and the resulting stress and anxiety for children) that surrounded the marital separation.
Exposure to interparental discord has been shown to have clear, long-term negative repercussions for individual well-being (Amato and Sobolewski 2001;Musick and Meier 2010). A number of studies have shown that conflict within a marriage can be detrimental for children. For example, children may perceive that they play a role in creating the tensions between their parents (Pryor and Rodgers 2001), or they may suffer because the ability of their parents to engage in warm and effective parenting is compromised (Amato 2000). This line of reasoning in turn implies that ending an acrimonious relationship might be beneficial for everyone involved, as it limits the risk of conflictual interactions occurring. Indeed, researchers have shown that when a partnership is of poor quality, the negative effects of divorce on adjustment are diminished for both the adults (e.g., Monden and Kalmijn 2006) and the children involved (Strohschein 2005). It is, however, important to note that dissolving a discordant relationship does not necessarily mean that the ties between the ex-partners are severed, especially when there are children involved. Whereas couples without children can have a "clean break" from a problematic relationship, parents have to maintain some form of communication while parenting their shared offspring (Fischer et al. 2005).
The magnitude of post-divorce conflict has been highlighted as an important moderator in the association between the dissolution of the partnership of the parents and the maladjustment of their children. It has been a key reason why researchers have pointed out that shared post-divorce residence for children might be harmful, as it can expose them continuously to heightened interparental discord (Harris-Short 2010; Poortman and van Gaalen 2017). Post-divorce conflict can affect children in a myriad of ways. It can, for example, intensify feelings of conflicting loyalties for children, as it could force them to choose between the contending parents. This pressure may, in turn, undermine the children's relationship with one or both of their biological parents (usually the father; Hornstra et al. 2020). It has, however, been argued that if parents are able to maintain a low-conflict, collaborative relationship following the dissolution of their partnership, the stress associated with the divorce could be temporary. Thus, any negative effects of the dissolution on the child's well-being would be short-lived (Ahrons 1994;Kelly and Emery 2003). This line of reasoning implies that the child should experience no long-term negative effects if the parents have a "good divorce." The "good divorce" became the ideal of the cultural elite during the 1970s and 1980s, when social science research was warning of the potentially negative effects of divorce on children (Dronkers 1997).
The existing evidence on this potential buffering effect has been mixed (for an overview, see Kelly and Emery 2003). Some authors have reported that children with collaborative divorced parents display fewer behavioral problems in adolescence (Amato et al. 2011;Beckmeyer et al. 2014), but others have found little evidence to support the good divorce hypothesis (Amato et al. 2011). In their analysis of ten child outcomes, Amato et al. (2011) identified only one significant difference that seems to indicate that cooperative parenting after a divorce confers an advantage. However, since their analysis was based on a small sample of approximately 300 young adults, the chances of finding insignificant results was high. In our contribution, we use a much larger sample; and, importantly, we also account for the parents' own behavioral problems while the child was young. Although we do expect to find a negative association between parental separation and adjustment in adulthood (divorce hypothesis), we also expect that once the level of conflict is considered, we will observe no differences in the long-term well-being of adult children raised in households with separated parents and those raised in households with continuously married or cohabiting parents (good divorce hypothesis).
Most of the previous studies that tested this hypothesis did so primarily for children who were still living at home. In our work, we generalize this idea to long-term effects (i.e., to the consequences for the children of divorced parents when they are adults and living independently). Whether any effects that were operating in an individual's childhood and youth last into his/her adulthood remains an open question. When children become young adults, their well-being is usually less dependent on the behavior of their parents, and they tend to be less exposed to their parents' conflicts. We therefore would not expect large effects but there are also reasons to believe that effects will still be present. Children need practical and emotional support from parents when making the transition to adulthood. Moreover, early mental health problems may lead to an accumulation of adverse experiences, and could therefore persist into adulthood.

Stepparents: The "Good Stepparent" Hypothesis
The majority of separated parents go on to re-partner (Thomson 2014). Thus, studying the long-term effects of parental union dissolution on child well-being in isolation from the possible addition of a stepparent to the household in which the child lives might lead to erroneous conclusions. It is, for example, possible to misrepresent the consequences of divorce by conflating the impact of two distinct transitions: namely, the separation of the parents and the re-partnering of the resident parent. Therefore, in this chapter, we not only address the "good divorce" hypothesis as presented above, but also consider how the addition of a stepparent to a singleparent household might affect the well-being of the child in young adulthood.
The assumption that stress is a key mechanism underlying the association between household transitions and adjustment levels has also been made in studies that examined the extent to which living with a stepparent affects a child's wellbeing (Fomby and Cherlin 2007). Similar to the argument made about parental union dissolution, this hypothesis suggests that adding a new parental figure to a single-parent household involves a number of potentially stressful adjustments for both the parent and the child. In line with this argument (and, albeit to a less extent, with the resource hypothesis), a number of studies have shown that the transition from a single-to a stepparent household is not associated with more positive outcomes for the children involved (Coleman et al. 2000;Hanson et al. 1996;Thomson et al. 1994). Therefore, we expect to find that young adults who experienced the addition of a stepparent to the household were worse off than their counterparts who experienced only the transition from two-parent to a single-parent household (stepfamily hypothesis).
We believe, however, that the effects on a child of the simple transition from one type of household to another are less important than the quality of the tie between the child and the new co-resident parental figure. In other words, as in the case of divorce, we need to recognize the likely heterogeneity in the ties that are created as a result of the transition. While a large number of studies have highlighted the challenges stepparents face in establishing a positive relationship with their stepchildren (Stewart 2007), there is also evidence of considerable diversity in the quality of these dyadic ties (King 2007;White and Gilbreth 2001). It has, moreover, been shown that high-quality stepparent-child relationships are associated with higher adjustment levels well beyond childhood (Jensen and Harris 2017). Therefore, in this contribution we propose a "good stepparent" hypothesis: i.e., we expect to find that adult children who grew up with a stepparent with whom they had a highquality relationship fare better than children who were raised by a single, separated parent. Again, these expectations apply to the long-term effects of stepfamily experiences.

Data
We use data from the recently collected Dutch survey Parents and Children in the Netherlands (Ouders en Kinderen in Nederland, OKiN; Kalmijn et al. 2018). The sampling frame was based on the Dutch population register (Bakker et al. 2014;Prins 2017). A systematic oversample was created based on with whom the targeted respondents (adults born between 1971 and 1991) were living when they were 15 years old. Three sampling strata were defined: (a) households with non-separated parents (both parents, as listed on the birth certificate, were present in the household; 25%); (b) households with one separated parent and no new partner (i.e., only one biological parent was present in the household; 33%); and (c) households with a separated parent and a new partner (i.e., one biological parent and the partner of the parent were registered as living in the household; 42%). Although the officially registered household constellation did not always match the actual situation of the household, using the strata was a highly effective way to oversample children whose parents were not continuously together throughout their childhood. The actual household situation of the respondent during youth was assessed in the questionnaire.
The fieldwork was carried out by Statistics Netherlands in 2017. The targeted respondents (also referred to as anchors) received an introduction letter inviting them to participate using an internet link. The invitation included an unconditional incentive (€5 gift certificate), and non-respondents received several reminders. If they did not respond after a month following the last letter, they were asked to participate in a face-to-face computer-assisted interview. The final response rate was 62% (N = 6485 adult children), which is above average for the Netherlands (De Leeuw and De Heer 2001). Given that our focus is on young adults, we decided to limit the sample to 25−35-year-old respondents. This led to a subsample of 4056 individuals. Of those, we dropped the anchors who reported that they either did not know or could not provide any information about their father or mother, as they were not asked questions about the level of interparental conflict after separation (n = 166 observations lost). We kept the respondents who experienced the death of a parent while young and those who were born to a single parent in order to give the reader the opportunity to compare the association between separation and maladjustment with the effects of living in other household constellations. The final analytical sample consisted of 3890 individuals (57.4% of whom experienced parental separation in childhood). On average, the participants with separated parents experienced the dissolution of their parents' partnership at 7.61 years of age (with a standard deviation of 4.08 years), which is, on average, 22 years prior to the interview (range 9-34; median: 22 years). Descriptive information about all of the variables described below is provided in Tables 13.1a and 13.1b (Table 13.1a shows the  unweighted descriptives and Table 13.1b shows the weighted descriptives).

Variables
Depression In this contribution, we operationalized maladjustment as the selfreported level of depressive feelings using the eight-item version of the Centre of Epidemiological Studies-Depression Scale ( Van de Velde et al. 2009). The respondents were asked to rate how well eight statements described how they felt during the past week on a scale from 1 = rarely or never to 4 = most of the time or always (e.g., "I felt depressed"; "My sleep was restless"; "I was unable to get going"). The scale was calculated by taking the mean of the eight items, with higher values denoting more depressive feelings. The reliability of the scale was α = .86 and the mean in our analytical sample was M = 1.62 (SD = 0.56).

Family and Household Composition in Childhood
Information about the family and household transitions in childhood were obtained through retrospective questions. The respondents were asked whether their parents were together at the time of the respondent's birth (single parent, n = 264 of analytical sample), whether either of their parents had died (deceased parent, n = 286 of analytical sample), and whether their parents had separated (separated parents, n = 2233 of analytical sample). The last two questions referred to the period before the anchor turned 18 (or moved out of the household, if preceding 18). If their parents were not together throughout their childhood, the respondents were asked whether their father or mother had any new partners at that time. Identical questions were then asked about the partners with whom the father and the mother had the longest relationships. The respondents were asked whether they lived with any of these new partners, and what the relationship between the anchor and the stepparent was like. Of the respondents aged 25-35 in our sample, 1690 had lived with a stepfather and 687 had lived with a stepmother for at least part of their childhood. Note that the data were collected retrospectively. This means that the new unions of the anchor's parents may have been terminated. In retrospective surveys, there is some danger of the underreporting of dissolved partnerships. In addition, a respondent's evaluation of an anchorstepparent relationship may depend on whether the union was still ongoing (see below).

Post-separation Interparental Conflict
If the participants reported that their parents separated during their childhood, they were asked to assess the level of conflict between their parents using the following question, "Were there serious fights between your parents in the first years following the separation?" on a four-point scale (1 = never to 4 = often). Those respondents who reported that their parents never or sometimes fought were rated as "low conflict," while those who reported that their parents fought regularly or frequently were coded as "high conflict." A total of 327 respondents in our analytical sample said that they did not know how much conflict there was between the parents. We decided to code these participants as having experienced a low level of conflict, based on the assumption that regular or frequent fighting would have been noted by the respondent. Of the anchors whose parents had separated, 1519 (68.0%) reported observing low levels of conflict, while 714 (32.0%) reported observing high levels of conflict.

Conflict with Stepparent
The respondents who reported that either of their parents had a new partner were asked to assess the level of conflict they had with that person by answering the following question about the period before they started living independently: "Were there tensions and/or conflicts between you and that new partner of your [mother/father] during that time?" (1 = never to 4 = often). As in the case of the interparental conflict items, we coded the "never" and "sometimes" categories as "low conflict" and the other two as "high conflict." Of the anchors who had lived with a stepparent, 68.0% (n = 1149) of those who lived with a stepfather and 65.2% (n = 448) for those who lived with a stepmother reported experiencing a low level of conflict.

Control Variables
In our analyses, we accounted for a number of potential confounders in the key association of interest. First, we controlled for the father's and the mother's problem behaviors during the anchor's childhood. Three indicators were considered: namely, frequent alcohol use (1 = yes, 0 = no), mental health problems (1 = yes, 0 = no), and addiction (1 = yes 0 = no). The three health behaviors were combined into a single scale that represented the count of problem behaviors of the parent during the adult child's youth. We also accounted for the socioeconomic status of both parents, which was based on a scale composed of each parent's standardized highest educational attainment and occupational status during the anchor's childhood (as reported by the participant, and indicated by the International Socio-Economic Index of Occupational Status, ISEI). In cases in which the information needed to estimate the SES status of the parent was missing, we used the Dutch register data to impute these values based on each parent's origin (native Dutch, Western foreigner, non-Western foreigner), current income, age, and estimated home value. We had to implement this procedure for 177 fathers and 116 mothers (out of a full analytical sample of 3890). We also controlled for whether the respondent reported having lived in an institution (and not with family members) at any point during his or her youth (of our full analytical sample, only 53 anchors fit in this category). We also included controls for having a father of non-Dutch origin, the current age of the respondent, and gender.

Analytical Approach
We estimate a number of linear regression models, with the self-reported level of depressive feelings as the dependent variable. The first model only examines the association between experiencing parental separation and maladjustment. The second model addresses the "good divorce" hypothesis by focusing on the level of post-separation conflict between the parents. The reference category consists of children whose parents remained together during their childhood. The third model adds an effect for having lived in a stepfamily. The coding is cumulative; hence, this effect compares children from stepfamilies with children from single-parent divorced families. The fourth model adds variables for high-and low-conflict stepfamilies, separately for stepfathers and stepmothers. Again, the coding is cumulative. We present predictive margins to facilitate the interpretation of these more complex models. Additionally, given that the level of depressive feelings was rather low in our analytical sample (M = 1.62, SD = 0.56 for a variable which ranged from one to four), we also estimated logistic regression models, with the probability of reporting depressive symptoms in the top 15% of the distribution. In our analyses, all of the continuous variables (including the dependent variable in the linear regression models) were standardized to M = 0, SD = 1, which means that the findings can be interpreted as standardized effect sizes. As we had a number of missing values for the parents' problem behaviors (n = 273 for mothers and n = 558 for fathers), we implemented sequential imputation using chained equations that included all of the variables from the main analyses and a number of anchor characteristics (educational level, unemployment status, current partnership status, parenthood status, and whether the respondent had experienced a partnership dissolution).

Results
We first focus on some descriptive results (presented in Table 13.1a unweighted and  Table 13.1b weighted), separately for the young adults who grew up living continuously with their two biological parents and those who did not. First, we can see that, as expected, the anchors who experienced the separation of their parents reported higher levels of depression than their counterparts with parents who were continuously together though the difference is certainly not large. The highest average levels of depression were reported by the anchors who were born into a single-parent household. If we look at the weighted percentage of anchors who reported very high levels of depression (in the top 15% of the distribution), we can see that 19% of the young adults with separated parents, but just 11% of the young adults with parents who were continuously together, reported having high levels of depression. In other words, the results of these basic bivariate analyses suggest that there are some persistent differences in the mental well-being of young adults based on whether their parents had separated.
Another interesting finding displayed in Tables 13.1a and 13.1b is that although the majority of young adults with separated parents reported observing low levels of conflict after parental separation, one in three respondents indicated experiencing a conflictual interparental relationship. An additional check showed that those who observed low levels of post-divorce conflict did not differ substantially in terms of the age at which they experienced a separation from the high-conflict group (M = 7.56, SD = 4.12 for the low-conflict group and M = 7.72, SD = 3.98 for the high-conflict group). As we implied in the introduction, the vast majority of the OKiN participants were living with a stepparent following the breakup of their parents' union and unsurprisingly, more participants were living with a stepfather than with a stepmother. The results shown in Tables 13.1a and 13.1b indicate is that there was indeed considerable heterogeneity in the quality of the child-stepparent tie: about one-third of the participants reported having regular and frequent confrontations with a stepparent, regardless of whether the new partner was male or female.
We now turn to Table 13.2, which displays the results from the linear regression analysis with self-reported depressive feelings as the dependent variable. It is important to point out that although we refer here to "effects," we cannot, of course, ascertain a causal relationship per se given the nature of our data. In the first model, we see a positive and significant effect of experiencing parental separation on depressive symptoms in young adulthood, although the magnitude of the effect is not large (.11 of a standard deviation; compared to, for example, the effect of being born to a single parent, which is .22). It is, however, important to keep in mind that this association was found for a sample of young adults who experienced that transition an average of 22 years ago. We would be surprised if we found a very large effect of a single important life event in childhood so many years later. Therefore, this result supports previous findings about the long-lasting repercussions of parental separation on the children involved. Another interesting result is the finding that the participants who had experienced the death of a parent did not report higher levels of depression in adulthood than those whose parents were continuously together.
Yet the findings of the subsequent models are also essential to our work. Model 2 shows that the association between divorce and maladjustment is only present if it is followed by high levels of interparental conflict. The magnitude of the effect is larger than it is in the first model (.260 vs .109), and it is comparable to the detrimental effect of being born to a single parent. No statistically significant association is found between having experienced a low-conflict parental separation and maladjustment, which suggests that children who experienced low-conflict separations are doing as well as children in families with stable parental unions. Thus, the "good divorce" hypothesis posited by Ahrons (1994) is supported.
Model 3 and Model 4 focus on the "good stepparent" hypothesis. Looking at Model 3, we observe an association between adding a stepparent to the singleparent post-divorce household and maladjustment. In other words, the results of this model suggest that having experienced an additional transition to a stepparent household (and potential stress associated with it) is not necessarily detrimental in the long run. However, as Model 4 shows, the failure to find a significant association is due to the clearly diverging effects of having a low-conflict or a high-conflict relationship with the co-resident stepparent. For example, having lived with a stepfather with whom the anchor had a low-conflict relationship is shown to be associated with having lower levels of depressive feelings than if the anchor had not lived with a stepparent in childhood. However, having lived with a stepfather with whom the tie was strained is found to be associated with a significant and sizeable increase in depressive symptomatology (about a fifth of a standard deviation). For the young adults who reported living with a stepmother, the story was slightly different. The negative impact of having had a conflictual relationship with a stepmother is still clearly visible, but the protective effect of having had a low-conflict tie is not. In other words, the results of this model indicate that whereas having a positive tie with a stepfather can buffer against depressive symptoms in young adulthood, having a negative tie with a stepfather is associated with maladjustment. In the case of a coresident stepmother, at best, the experience does not result in harm in the long run.
In addition to performing these analyses, we examined the question of whether having experienced family complexity in childhood is strongly predictive of high levels of maladjustment in adulthood (i.e., reporting depressive feelings in the top 15% of the distribution). The logistic regression models, displayed in Table 13.3, show the association between the experience of having lived in different household constellations during childhood and the probability of reporting very high levels of depressive feelings in adulthood. The results of the first model indicate the young adults whose parents split had 39% higher odds (b = 0.335) of reporting extreme levels of depression compared to their counterparts with continuously married parents. In other words, whereas the baseline probability of reporting such high levels of depression for the participants with non-separated parents was about 10%, those with separated parents had a probability of about 14%. Again, the reader should keep in mind that we are referring the impact of family events that happened, on average, two decades ago. In Model 2, we again see that the negative repercussions of divorce are only visible if the parents had frequent conflicts following the divorce. Having experienced that situation increased the odds of extreme depression by 80% (b = 0.59) compared to having had continuously married parents.
Model 3 and Model 4 examined the association between having lived with a stepparent and extreme levels of depression. It is clear that even though the direction of the coefficients is consistent with the analyses displayed in Table 13.2, the associations are not statistically significant. In other words, neither living with a stepparent nor the quality of the respondent-stepparent tie is found to have a significant effect on the risk of experiencing extreme depression in early adulthood. Still, in order to give the reader an impression of the associations, we estimated the predicted probabilities of reporting high levels of depression depending whether the respondents experienced their parents' divorce as good or bad, and whether they had a good or a bad relationship with the co-resident stepfather (the more likely co-resident stepparent figure). These associations are plotted in Fig. 13.1. We can see that for an individual who experienced a low-conflict parental separation and had a low-conflict tie with the stepfather, the predicted probability of having severe depression was 0.11, which is very comparable to the probability for an individual whose parents were together throughout his or her childhood. However, for a young adult who experienced a high-conflict parental separation and then had a high-conflict tie with his or her stepfather, the predicted probability of having severe depression was twice as high, at 0.22. In summary, although the logistic regression does not provide statistically significant findings in support of the "good stepparent" hypothesis, we consistently find that conflict between the biological parents and conflict within the stepparent-child tie are much more detrimental to an individual's long-term wellbeing than simply having experienced specific household transitions.

Discussion
In this chapter, we used a unique survey among young Dutch adults to study the long-term association between parental union instability and maladjustment. We first examined the question of whether parental separation had a negative effect on individual well-being into adulthood, and then paid specific attention to the role conflict played in this association. Our work thus explored a hypothesis presented by Ahrons (1994), which states that as long as parents are able to maintain a  Table 13.3 for full model)) cooperative relationship after separation, children will be shielded from the longterm effects of divorce (the "good divorce" hypothesis). We built on this proposition by also considering whether the effects of the subsequent -and potentially stressful -addition of a stepparent to the household were also contingent on the level of conflict between the child and the stepparent. Our findings support both the "good divorce" hypothesis and what we have called the "good stepparent" hypothesis. First, in line with previous work, we found evidence that parental union dissolution has long-term effects on the children involved (Amato 2010;Amato and Cheadle 2005;Härkönen 2014). The young Dutch adults in our sample who saw their parents separate (which took place, on average, two decades earlier) reported having somewhat higher levels of depression than their counterparts whose parents were continuously together. However, our findings also indicated that it was not the separation itself that left that a lasting mark, but rather the level of conflict following the transition. The results of our analysis can be interpreted as providing clear support for the "good divorce" (Ahrons 1994) hypothesis: when the parents were able to minimize their overt disagreements after separating, their adult children's levels of adjustment did not differ from those of adults whose parents did not separate; whereas when the parents had heightened post-separation conflict, the mental wellbeing of their adult children clearly suffered. To the best of our abilities, we controlled for possible confounding factors, such as the parents' own problematic behaviors and socioeconomic positions while the young adults were children, and still found a negative association between having experienced a high-conflict separation and an adult child's well-being.
Another interesting outcome of our work is that we did not necessarily find an additional negative effect of parental repartnering after separation, which could have been expected based on the assumption that multiple household transitions lead to additional stress, and, thus, to maladjustment for the children involved (Amato 2010). We also did not find an automatic benefit of adding another adult to the household in which the child was living after the parental divorce. Previous studies that examined the adjustment levels of children reported there is no parental repartnering benefit (Hanson et al. 1996). In other words, in contradiction to the "resource hypothesis," we found that the addition of another parent figure and their resources to the household did not necessarily offset the disadvantage of having separated parents relative to having a stable two-parent family. Similarly, we did not find that simply having lived in a stepfamily after the parental separation was associated with high levels of depression. However, we uncovered important differences in the well-being of adult children after taking the quality of the stepparent-child tie in childhood into account.
In line with earlier work that pointed to the benefits of a high-quality stepparentchild tie (Jensen and Harris 2017), our findings supported what we coined "the good stepparent" hypothesis. We found that having a low-conflict relationship with a resident stepfather could act as a buffer against depressive symptomatology in adulthood. In contrast, when the relationship between the child and the stepparent was characterized by conflict, we observed cumulative negative effects. Of the young adults who experienced both a high-conflict divorce and a conflict-ridden tie with a resident stepparent in childhood, 22% reported very high levels of depression, compared to 11% of the young adults whose parents stayed together throughout their childhood. We consider this to be a noteworthy finding, especially given the extreme levels of depression reported in this specific case and the amount of time that had passed since the family transitions.
However, we also observed an interesting gender difference: whereas a highquality stepfather-child tie was a protective factor and a low-quality stepfather-child tie was linked to higher levels of depression, at best, the stepmother-child tie was not associated with maladjustment. In other words, having a high-quality relationship with a stepmother does not appear to confer the same benefits as a having highquality relationship with a stepfather. The precise reason for this gender difference is unclear. It is worth noting that although shared custody arrangements are on the rise in the Netherlands (Poortman and van Gaalen 2017), the majority of the adults we studied lived primarily with their mother (and thus with a stepfather, and not with a stepmother) after separation. Yet having had more exposure to one stepparent figure than another cannot explain why we found comparable disadvantages of having a high-conflict stepfather-or stepmother-child tie, but that only stepfathers could act as a buffer. This remains an open question in our work.
We conclude by noting certain limitations of our study, and opportunities for further research. One limitation is that we were unable to control for child effects. It is, for example, possible that the deterioration in a child's psychological well-being resulted in the child having a poor relationship with the stepparent. Hence, some of the evidence we found in support of the "good stepparent" hypothesis may be due to reverse causation (Kalmijn et al. 2019). A second limitation is that we did not have measures of potential conflicts between the parent and the stepparent. While there is likely some correlation between the conflicts the child had with the stepparent and the conflicts the parent had with the stepparent, this correlation is far from perfect. To check this bias, we estimated an extra model in which we controlled for stepfamily instability (whether the parent and stepparent separated later). Our results were robust even after considering this additional control variable. Related to this second limitation, we have to acknowledge that all of our measures of conflict were based on retrospective information from a single source -the adult child. We cannot eliminate the possibility that a participant's current state of mind was affecting his or her perception of the past. However, our findings are very much in line with those of other studies that were based on prospective data, and that also examined the impact of conflict on individual well-being (e.g., Musick and Meier 2010). Despite these limitations, our study has several advantages, including our use of a systematic oversample of children from stepfamilies and elaborate measures of confounding parent and stepparent traits, and our focus on the generation who grew up during the divorce revolution. Our finding that family turmoil affected the depressive symptoms of adults so many years after their parents divorced and re-partnered is striking, and can be seen as bad news, given that the young adults we studied were in a life course phase that can be particularly challenging, as it tends to be characterized by the clustering of multiple transitions (becoming a parent, entering employment, etc.). But the good news is that we found no divorce effects at all when the level of conflict between the separated parents of the anchor was low, and when the stepparent-child ties were harmonious, as was often the case for these young adults. White, L., & Gilbreth, J. G. (2001). When children have two fathers: Effects of relationships with stepfathers and noncustodial fathers on adolescent outcomes. Journal of Marriage and Family,63(1), 155-167. https://doi.org/10.1111/j. 1741-3737.2001.00155.x. Wolfinger, N. H. (2011. More evidence for trends in the intergenerational transmission of divorce: A completed cohort approach using data from the general social survey. Demography, 48 (2), 581-592. https://doi.org/10.1007/s13524-011-0025-4.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.