Parental Divorce and Externalizing Problem Behavior in Adulthood. A Study on Lasting Individual, Family and Peer Risk Factors for Externalizing Problem Behavior when Experiencing a Parental Divorce

ABSTRACT

This study explored the relationship between experiencing a parental divorce in childhood and externalizing problem behavior (EPB) in adulthood. We examined individual, family, and peer influences for those growing up in a broken home. Our findings indicate that growing up in a broken home has enduring consequences for EPB. However, factors influencing EPB in adulthood did not differ according to family structure. Regarding changes in EPB in adulthood, having more deviant friends was more hazardous for those growing up in a broken home, while parental rejection and low self-control were more hazardous for those from an intact family.

Introduction

A parental divorce can undermine children’s wellbeing and development, with consequences lasting far into adulthood (Amato 2000; Hetherington and Stanley-Hagan 1999; Spruijt and Duindam 2005; Theobald, Farrington, and Piquero 2013; Van der Valk et al. 2005; Wallerstein and Lewis 2004). One negative outcome of a parental divorce is a higher likelihood that offspring will “act out”, that is, show externalizing problem behavior (EPB) (Blum et al. 2000; Neumann et al. 2010). EPB is outward-oriented behavior that reflects maladjustment and an inability to engage constructively in society (Buehler et al. 1997), and which often produces distress in others (Gerard and Buehler 1999).

Nowadays, a rising number of children grow up in unstable family homes (OECD 2011). This is also the case in the Netherlands, were some 20 percent of children under age 16 have divorced parents (Statistics Netherlands 2017). Considering that children who experienced a parental divorce are more at risk of developing EPB, it is relevant to seek insight into factors that might reduce problematic behaviors, particularly in these more vulnerable divorced family homes.

A strong correlation has been found between parental divorce and youth EPB, but research is inconclusive on the long-term effects of divorce (Amato 2010). Some studies suggest that parental divorce continues to affect offspring development into adulthood (Cherlin, Lindsay Chase-Lansdale, and McRae 1998; Theobald, Farrington, and Piquero 2013; Wallerstein and Lewis 2004). Others show the effects of divorce as mainly short term (e.g., Amato 2000; Amato and Anthony 2014). The current study seeks to expand knowledge in this field by exploring the long-term relationship between experiencing a parental divorce in childhood and EPB in later life in the Netherlands.

Ample theory and research, however, has also focused on antecedents of EPB other than a parental divorce (e.g. Farrington 2005; Rhee and Waldman 2002; Ribeaud and Eisner 2010). The problem behavior theory (J. Jessor and S. L. Jessor 1977) and the multiple risk model (Van der Ploeg and Scholte 2003) suggest that beyond one’s own behavior and personality traits (such as self-control), family characteristics (like parental rejection and family socioeconomic status) and peer influence are important. Exploring the influence of these three dominant dimensions – individual, family, and peers – on EPB simultaneously, in addition to the experience of a parental divorce, expands on previous research by bringing out the particular consequences of growing up in a broken home (Theobald, Farrington, and Piquero 2013). Moreover, we elaborate on this by examining how these influences differ between those who grew up with divorced parents and those in an intact family. As such we aim to find out what make some children growing up with divorced parents more susceptible to EPB. At last, we intent to contribute to the field, since relatively little research has looked at who is most likely to be affected by a parental divorce, in terms of a higher chance of EPB in later life, into adulthood. If this is the case, insights on interventions can be sought to better protect those who experienced a parental divorce in their youth.

Our research questions read as follows: To what extent does experiencing a parental divorce in youth lead to EPB in adulthood? To what extent are individual, family, and peer-related antecedents of EPB in adulthood moderated by having experienced a parental divorce in youth?

Hence, this research seeks new and challenging insights regarding EPB in later life. We are particularly interested in the relation between experiencing a parental divorce at a young age and EPB later on, in adulthood. We applied two methodological approaches. First, we retrospectively investigated the effect of a parental divorce on EPB by applying the ordinary least squares (OLS) method to Dutch data collected in 2014. Our data is from the CrimeNL database, which is unique in its longitudinal approach and focus on criminality in the Netherlands. Our second approach, use of lagged dependent variable (LDV) models, allowed us to look at the development of EPB over a one-year period in adulthood, thus exploring whether experiencing a parental divorce in childhood remained equally influential in later life, for those in the Netherlands.

Theoretical framework

Parental divorce and externalizing problem behavior

Children who experience a fundamental transition in their living arrangements have been found to experience lower levels of wellbeing thereafter (Brown 2006). One such transition is a parental divorce. Children with divorced parents also tend to experience lower levels of wellbeing overall than children in an intact family (Amato and Keith 1991). In line with these findings, studies have shown that children growing up in a broken home (in this study a home with divorced parents) are more likely to develop EPB (Amato and Anthony 2014; Blum et al. 2000; Furstenberg and Teitler 1994; Neumann et al. 2010; Thijs et al. 2015; Thomson, Hanson, and McLanahan 1994; Van de Rakt 2011). Hence, a parental divorce seems clearly disadvantageous for children’s healthy development.

According to the divorce stress adjustment perspective (Amato 2000), two situations can occur after children or adolescents experience a disruption of the family home. These two situations are described by the crisis model and the chronic strain model. According to the crisis model, a parental divorce temporarily disturbs life, but does not have an enduring effect. Most children adjust well over time, with negative consequences being concentrated in the first two years after the divorce. Thereafter, everyday life routines recommence (Amato 2000). Luckily this is the case for plenty of children who experience such a disruption in their family life (Amato and Anthony 2014; Morrison and Cherlin 1995). Nonetheless, many children experience negative consequences of a parental divorce for a more extended period (Cherlin, Lindsay Chase-Lansdale, and McRae 1998), which is in line with the chronic strain model. Studies have found that the effects of growing up in a broken home, such as a higher chance of forming risky habits, can extend far into adulthood (Hetherington and Stanley-Hagan 1999; Spruijt and Duindam 2005; Van der Valk et al. 2005; Wallerstein and Lewis 2004). Based on the chronic strain model, we assume:

H1: There is a positive relation between experiencing a parental divorce in youth and EPB in adulthood

The moderating effect of a parental divorce

Since both the crisis model and the chronic strain model can occur after the experience of a parental divorce, this indicates that not all who experience a parental divorce in their youth will exhibit EPB in later life. Whether children who experienced a parental divorce do exhibit EPB later has been found to depend on personal and social factors (Amato 2010; Theobald, Farrington, and Piquero 2013). Prior research suggests three dominant determinants of EPB. Next to individual characteristics, the social environment, specifically parents and peers, is highly influential when it comes to such problem behavior. Our research therefore particularly addresses the moderating effect of experiencing a parental divorce in youth on the relationship between these three dimensions and EPB.

Family dimension

Children in families with low socioeconomic status (SES) are found to be particularly at risk of developing EPB (Bjerk 2006; Duncan, Brooks-Gunn and Klebanov 1994; Yoshikawa 1994). A family SES generally is represented by the following highly correlated indicators: parental income, parental occupation and parental education (Duncan, Featherman, and Duncan 1972; Hoff, Laursen, and Tardif 2002; Sirin 2005). For different reasons, parental education is considered to be a stable aspect of a family SES, and a proper indication of the social, economic, and cognitive resources of families in modern societies (Krieger, Williams, and Moss 1997; Mandemarkers and Kalmijn 2014; Mueller and Parcel 1981). Moreover, parental education is a common indicator of parental SES in prior research on deviant behavior (see e.g., Osgood et al. 1996; Wright et al. 1999).

We are interested in whether the influence of parental SES on EPB differs between those who experienced a parental divorce and those who did not. Research found that high SES parents are better able to cope with disruption of the family and to protect their children from potential negative consequences (Walper, Thönnissen, and Alt 2015) For instance, higher educated mothers are found more competent to provide a safe and stable environment for their offspring after a divorce than those with less years of education. Also, higher educated mothers are better negotiators and therewith retrieve more child support and alimony (Mandemakers and Kalmijn 2014; Seltzer 1991). This is important, since in a broken home economic strain accounts for part of the disadvantages the children experience (Thomson, Hanson, and McLanahan 1994). Compared to lower educated mothers, higher educated mothers have better opportunities on the labour market, and they are more aware of the possible risks of a divorce, also for the upbringing of their children, due to their higher level of literacy (Kalmijn 1999; Seltzer 1991, 1998; Van de Werfhorst and Kraaykamp 2001). As such, the risk of growing up in a divorced home might be mitigated if sufficient parental socioeconomic resources are available (Garriga and Härkönen 2009; Mandemakers and Kalmijn 2014; Theobald, Farrington, and Piquero 2013).

H2: Those who grew up in a family with a low SES show more EPB in adulthood, and this relation is even stronger for those who experienced a parental divorce in youth.

Besides family SES, parenting style is considered important for a child’s development (Aquilino and Supple 2001; Hoeve et al. 2009; Wright and Cullen 2001). Parenting style as a factor is generally recognized in criminological and pedagogical studies on EPB in adolescence (e.g., Hoeve et al. 2007). Parental rejection and neglect are thought to be especially harmful to a child’s wellbeing (Hoeve et al. 2009). Parental rejection, in the form of extremely negative attitudes towards a child, can stimulate EPB, which could then endure for the longer term (Fauber et al. 1990; Roelofs et al. 2006). Fauber and colleagues (1990) found that parental rejection was associated with other adjustment problems among children growing up with divorced parents. The loss these children experienced due to the divorce led to greater anxiety regarding parental attachment, especially of the custodial parent. When the custodial parent practiced a rejecting style of parenting, this was likely to increase the stress on the child, in turn increasing EPB.

H3: Those who experienced parental rejection in their youth show more EPB in adulthood, and this relation is even stronger for those who experienced a parental divorce in youth.

Individual dimension

Prior studies have found that certain individual or personal characteristics are correlated with levels of EPB (Pratt and Cullen 2000). One of these characteristics is lack of self-control (Windle 2000). Self-control is a behavior-steering trait that is shaped early in life (Buker 2011; Teasdale and Silver 2009). The self-control theory of Gottfredson and Hirschi (1994) posits that those with low self-control seek immediate gratification and have less ability to resist the temptation of the immediate rewards universally derivable from crime (Akers 1991). Accordingly, studies have found that those with a more impulsive and risk-seeking personality, and thus low self-control, are more likely to exhibit EPB (Akers 1991; Barnes et al. 2006; Crosswhite and Kerpelman 2009; Gottfredson and Hirschi 1990).

Experiencing a parental divorce can be perceived as a threatening situation and challenge, because according to the parental absence model, divorced parents generally have less time and resources to care for their children compared to intact families (McLanahan and Sandefur 1994). Coping is a mechanism used to overcome or minimize the impact of a perceived threat or challenge (Agnew 1993; Folkman 1984) and is related to self-control (e.g., Folkman 1984). Children with low self-control are more attracted to risky behavior and likely to need more guidance than children with greater self-control. Therefore, those with low self-control may be more disadvantaged by the consequences of a divorce than those with greater self-control (Eisenberg et al. 2003; Finkenauer, Engels, and Baumeister 2005).

H4: Those who have a low level of self-control show more EPB in adulthood, and this relation is stronger for those who experienced a parental divorce in youth.

Peer dimension

According to age-graded social control theory, behavior is partly determined by influences of the social context outside of the family (Laub and Sampson 2001). This theory is part of the dynamic development perspective, which states that friends and partners are highly influential over the life course. The role of friends can also be of a higher importance when growing up in a broken home. Following the parental absence model (McLanahan and Sandefur 1994), there will be less support and control from divorced parents. This lack of support might be filled by peers. However, when this is filled by deviant peers, an increase in EPB is to be expected, as the number of deviant friends in one’s social network is a strong and consistent predictor of violence and aggression (Ferguson, Miguel, and Hartley 2009; Haynie and Osgood 2005; Rokven 2016; Sarnecki 2001; Steinberg 1987). Association with delinquent peers often precedes initiation of delinquent acts (Elliott and Menard 1996; Sarnecki 1990). Delinquent peers can function as delinquent role models, transmitting deviant values and leading to their internalization (Agnew 1992; Brendgen, Vitaro, and Bukowski 2000).

When it comes to the influences of deviant peers, social control processes in the family are important as well (Haynie and Osgood 2005). Exposure to deviant peers is a risk factor for smoking, alcohol abuse, and use of illicit drugs, all of which are aspects of EPB, and this relationship has been found to vary by family structure (Eitle 2005). Some previous research suggests that the extent of whether having deviant friends is harmful to those growing up with divorced parents depends on the type of EPB under study (Eitle 2005). However, since different types of problem behaviors tend to correlate (Jessor 1992), an accumulation of negative effects may be expected in those who experienced a parental divorce. Thus, those with more deviant friends may be more at risk of developing EPB and having more deviant friends may be more harmful to those who experienced a parental divorce than those who have not experienced this.

H5: Those who have more deviant friends show more EPB in adulthood, and this relation is stronger for those who experienced a parental divorce in youth.

Data and measurements

This study used data from the CrimeNL Dutch panel survey.¹ CrimeNL is a longitudinal study of individuals’ experiences with crime. It targets Dutch-speakers, ages 15 to 45, within the ten largest cities of the Netherlands.² The main topics are crime experiences, victimization, reports of offensive behavior, lifestyles and insecurity (Tolsma et al. 2014, 2015). The third CrimeNL wave (2014) also collected extensive data on the family situation and parental socialization in the respondents’ youth, which makes it highly suitable for this research. Respondents participating in wave 3 (2014) were selected as the starting point of our panel analysis (T–1). To explore whether EPB may develop or alter even in adulthood, wave 4 in 2015 was used to determine changes in levels of EPB over a one-year period (T). Some 356 respondents participated in both waves. Unless otherwise noted, respondents with missing values on one of the relevant variables were removed by listwise deletion. This left 270 respondents for analysis.

Methods

First, we conducted a linear regression analysis to investigate the extent that EPB in adulthood (T–1) can be explained by the experience of a parental divorce in childhood and by other family, individual, and peer characteristics. Secondly, we used panel analysis to explore whether increased levels of EPB over a one-year period in adulthood were associated with the expected factors. We applied lagged dependent variable (LDV) modelling, since it allows inclusion of time-invariant variables (experiencing a parental divorce, parenting style at age 12, and family SES) (Johnson 2005). In the LDV model, the dependent variable of wave 3 (b₂Y_i1) was regressed on the dependent variable of wave 4 (Y_i2) (Johnson 2005). The formula is as follows:

Y_i2 = b₀ + b₁X₁ + b₂Y_i1 + b₃S_i + E_i(1)

The effects of an individual predictor measured at T–1 can be interpreted as the extent that a respondent’s level of EPB changed over time, in this case over a one-year period in adulthood. LDV modelling is the preferred method when the dependent variable has a causal influence on the subsequent wave (Johnson 2005). In our case, it is theoretically plausible that EPB at wave 3 causally influenced EPB at wave 4, given that research has found those who exhibit EPB earlier in life are likely to carry on this behavior later in life as well (e.g. Laub and Sampson 2003).

Measurement of dependent variable: externalizing problem behavior

Our study was interested in externalizing problem behavior (EPB), defined as a range of signals of maladjustment (Achenbach 1982; Spruijt and Duindam 2005). In our dataset, information on EPB was based on the self-reported delinquency scale (Donker et al. 2004). EPB was measured with the following items: fare dodging, illegal downloading, use of soft drugs, use of hard drugs, alcohol abuse, and absence from work or school without a reason. These items provided a reliable measure of the construct EPB (wave 3: α = 0.683; 7.3% missing; range 0.00–1.00, wave 4: α = 0.649; 0.06% missing; range 0.00–1.00), as a Cronbach’s alpha of 0.6 is generally considered acceptable (Nunally 1978).

Measurement of independent variables: time invariant variables

At the individual level we took self-control as a stable personality trait (Blokland and Nieuwbeerta 2005). Self-control was measured using the 13-item Brief Self-Control Scale (Tangney, Baumeister, and Boone 2004). We conducted a principle factor analysis to verify that these items indeed measured self-control. Two items had to be omitted due to too low communalities or factor loadings. These were, “I often do things without thinking through alternatives” and “I say inappropriate things”. Self-control, constructed using the remaining 11 items, reliably measured the theoretical construct (α = 0.811; range 0.91 – 4.00). Higher scores indicate greater self-control.

Whether the respondent experienced a parental divorce was measured as residing in an intact family or having divorced or separated parents. In light of our focus on long-term parental influences on offspring, information on how old the respondent was when the parents separated was considered. This resulted in three categories: (0) “intact family”, (1) “divorced parents < 12 years old (parents divorced when respondent was 0 to 12 years old)”, (2) “divorced parents 12 years and older (parents divorced when respondent was 12 years old and older)” (0.8% missing).

Family socioeconomic status (SES) was measured as the highest completed level of education between the parents. If the respondent could not indicate the level of education of one parent, the level of education of the other parent was used (4.2% missing; range 0.00–3.00).

Parental rejection was represented by four items: “My parents punished me hard, even when it was considered futile”, “I was beaten by my parents more that I deserved”, “My parents always told others I was lazy”, and “I was treated as the black sheep of the family”, based on Arrindell’s s-EMBU scale (Arrindell et al. 2001, 1999). A principal factor analysis indicated that the item “My parents always told others I was lazy” did not measure the same construct as the other three items. It was therefore omitted. Taking into account the internal consistency of the items (α = 0.682; range 0.00–3.33), we can conclude that the remaining items provide a reliable measure of parental rejection (Nunnally 1978). A scale was consequently computed, with higher scores on the scale indicating greater parental rejection.

Measurement of the independent variables: time-varying variables

Next to individual and family characteristics, we explored the role of deviant friends. Respondents were asked whether any of the following statements applied to themselves and their friends and acquaintances: “absent from work or school without good reason”, “drinks too much alcohol (for instance, more than 10 glasses a night)”, “uses drugs (marihuana, pot, hash, coke, XTC, etc.)”, “hangs out on the street at night”, and “does stuff that is against the law”. The answer categories were (0) “not true for anyone”, (1) “true for some”, (2) “true for most”, and (3) “true for all”. Scores on the latter two categories were quite low, so we merged these. Furthermore, principle factor analysis for both waves indicated that the item “hangs out on the street at night” needed to be deleted because of a too low communality. Reliability with the four items was calculated as 0.704 for T–1 and 0.653 for T. Higher scores on the constructed scale indicate more deviant friends (4.8% missing T–1; range 0.00–2.00, 6.5% missing T; range 0.00–2.25). This was included in the panel analysis as a time-varying factor.

Measurement of control variables

We included several control variables shown to be relevant in studies of EPB. First, we calculated the age of respondents, using the birthdate and year of the interview. Age ranges were 17–48 in wave 3 and 18– 49 in wave 4. To ensure that respondents’ socialization is completed, we omitted respondents younger than 21 (n = 15). Formally, in the Netherlands adulthood starts at the age of 18. However, parents have to financially support their child(ren) until the age of 21. Therefore, in the Netherlands youngsters up to 21 are most likely still (highly) dependent on their parents. As we are interested in the relation between parental socialization and EPB in adulthood, we want to omit a possible direct parental influence. Note however that our results and conclusions do not change when we run our models on a sample including respondents (children) 18 years and older. Age was included in the panel analysis as a time-varying factor. The highest completed education of the respondent was measured in four categories: (0) “no education or primary education only”, (1) “lower secondary education”, (2) “higher secondary education”, and (3) “tertiary education”. Because the dataset included respondents still in education, those who were currently studying were taken up in the higher categories corresponding to their preparatory schooling levels (see also Rokven et al. 2015) (0% missing T and T–1; range 0.00 – 3.00).³ The time-invariant control variable gender was measured at wave 3 as (0) “male” or (1) “female”, with no missing values.

Table 1 presents the descriptive statistics. These statistics indicate that on average the level of EPB has risen slightly from one year to another respectively from 17% to 17.8%. The average age of the respondent was 35 years old, and on average they were highly educated (mean 2.763 wave 3; mean 2.767 wave 4). Further, there is an equal representation of male and females in the dataset, and more respondents did grow up in an intact family (80,4%) than in a divorced home (19,6%). Respondents on average hold a neutral level of self-control (mean 2.386), as it ranges from 0.91 to 4.00. Further the table indicates that on average the level of family SES is high (mean 2.267), and that on average there is not a great deal of parental rejection (mean 0.493). The overrepresentation of higher educated and older people in our dataset, needs to be mentioned, as it might slightly bias the results.

Table 1. Descriptive statistics.

	Wave 3 (T–1)				Wave 4 (T)
	Minimum	Maximum	Mean/Proportion	Std. Deviation	Minimum	Maximum	Mean/Proportion	Std. Deviation
Time-Varying Variables
EPB	0.00	1.00	0.170	0.223	0.00	1.00	0.178	0.216
Deviant friends	0.00	2.00	0.559	0.438	0.00	2.25	0.571	0.445
Age respondent	21.00	48.00	35.219	7.159	21.00	49.00	36.248	7.183
Education level respondent	0.00	3.00	2.763	0.527	0.00	3.00	2.767	0.540
Time-Invariant Variables
Gender	0.00	1.00
Male			49.1
Female			50.9
Family structure	0.00	2.00
Intact family			80.4
Parental divorce (< 12 years old)			11.2
Parental divorce (12 and older)			8.4
Self-control	0.91	4.00	2.386	0.598
Family SES	0.00	3.00	2.267	0.873
Parental rejection	0.00	3.33	0.493	0.687
Number of respondents	270

Source: CrimeNL wave 3 and 4 (2014, 2015)

Analyses and results

OLS regression and the relation between parental divorce and EPB in adulthood

Table 2 presents estimates of the relation between family, individual, and peer characteristics and EPB in adulthood. Model 1 shows the relation between a parental divorce experienced in childhood and EPB in later life. The results indicate to a certain extend the chronic strain model. Those who experienced a divorce when they were younger than 12 years old, tend to exhibit higher levels of EPB in adulthood compared to those who grew up in an intact family. Among those who experienced a parental divorce at age 12 or older, levels of EPB did not differ significantly from those who grew up in an intact family. These findings suggest that younger children are more affected by a parental divorce. Indeed, previous studies show that young children may blame themselves for the family’s dissolution or come to fear total abandonment. Furthermore, young children have less social support to fall back on than older children (Hetherington, Stanley-Hagan, and Anderson 1989).

Table 2. Regression analysis on EPB T–1, number of respondents = 270.

	Model 1		Model 2		Model 3
	b	SE	b	SE	b	SE
Family structure (ref = intact)
Parental divorce (< 12 years old)	0.178***	0.044	0.079*	0.037	0.001	0.219
Parental divorce (12 and older)	−0.014	0.048	0.000	0.038	0.000	0.038
Self-control			−0.049*	0.019	−0.041*	0.020
Family SES			−0.014	0.014	−0.017	0.014
Parental rejection			−0.015	0.016	−0.023	0.017
Deviant friends			0.213***	0.029	0.197***	0.031
Control Variables
Age respondent			−0.006***	0.002	−0.006***	0.002
Education level respondent			−0.017	0.022	−0.022	0.023
Gender (Female = 1)			−0.062**	0.022	−0.057*	0.022
Interactions
Self-control * parental divorce (< 12 years old)					−0.049	0.061
Parental SES * parental divorce (< 12 years old)					0.040	0.043
Parental rejection * parental divorce (< 12 years old)					0.051	0.045
Deviant friends * parental divorce (< 12 years old)					0.083	0.077
Intercept	0.153***	0.015	0.505***	0.113	0.506***	0.117
Adjusted R squared	0.053		0.403		0.409

Source: CrimeNL wave 3 and 4 (2014, 2015); Significance levels: Ϯ=sig<0.10 *=sig<0.05 **=sig<0.010 ***=sig<0.000

Model 2 adds the other hypothesized antecedents of EPB. For the family dimension, no significant effects were found. Hence, we found that taking into account all other factors, parental rejection and family SES do not significantly relate to EPB in adulthood. However, the results do reveal an association between an individual’s self-control and number of deviant friends and EPB in later life. Hence, those with less self-control and those with more deviant friends tend to show more EPB in adulthood. Nevertheless, when taking into account the hypothesized relevant individual, family, and peer factors, experiencing a parental divorce during childhood is still significantly - though marginally - associated with levels of EPB in adulthood.

Model 2 also shows that age and gender are significantly related to EPB. In line with previous research (e.g. Herrenkohl et al. 2000; Junger-Tas, Ribeaud, and Cruyff 2004), we find less EPB among females and older respondents.

To explore why those who experienced a parental divorce in youth might be more susceptible to EPB in adulthood, model 3 tests differences between growing up in an intact family and those growing up with divorced parents in the relationship between EPB and the hypothesized individual, family, and peer characteristics.⁴ This model shows, however, that possible factors influencing EPB in adulthood - as in family SES, parental rejection, self-control and deviant friends - do not differ depending on family structure. Hence, our results suggest that an individual’s self-control and number of deviant friends are relevant factors determining EPB in adulthood, apparently regardless of one’s family structure during childhood. For children who grew up with divorced parents, a lower level of self-control or a larger number of deviant friends does not seem to be more disadvantageous in the long term than for children who grew up in an intact family.

LDV modelling and the relation between family structure and changes in EPB in adulthood

The findings of the linear regression (Table 2) suggest that a parental divorce has long-term effects on levels of EPB. However, the data also allow us to explore whether the positive relation between a parental divorce in childhood and levels of EPB remains when examining changes in EPB over a one-year period in adulthood, by applying LDV models.

Model 1 in Table 3 reveals that growing up with divorced parents is associated with more EPB over a one-year period, compared to growing up in an intact family. It appears that growing up with divorced parents has a lasting influence on one’s behavior, and can result in increases in EPB during a one-year period in adulthood. This points to a possible accumulation of disadvantages associated with the experience of a parental divorce in youth.

Model 2 adds the three dimensions (family, individual, and peers). The results suggest that solely the peer dimension significantly relates to changes in EPB over a one-year period in adulthood. Apparently, an increase in the number of deviant friends leads to an increase in EPB, even later in life. Moreover, in additional analyses (not shown), when an increased number of deviant friends was added to the model, the influence of family structure was significantly reduced. Hence, the accumulation of disadvantages, such as growing up with divorced parents, appears to be partly explained by increased deviance in one’s friendship circles.

Table 3. LDV model on EPB, number of respondents: 270, number of times = 2.

	Model 1		Model 2		Model 3
	b	SE	b	SE	b	SE
Family structure (ref = intact)
Parental divorce (< 12 years old)	0.090**	0.027	0.064*	0.027	−0.140	0.152
Parental divorce (12 and older)	0.005	0.034	0.002	0.033	0.005	0.032
Self-control			−0.001	0.014	−0.011	0.015
Family SES			0.000	0.010	−0.001	0.010
Rejection			−0.005	0.012	0.010	0.012
Deviant friends			0.089***	0.024	0.072**	0.025
Control Variables
Age respondent			−0.003**	0.001	−0.004**	0.001
Education level respondent			−0.028 Ϯ	0.015	−0.038*	0.015
Gender (Female = 1)			−0.032 Ϯ	0.016	−0.039*	0.016
Interactions
Self-control * parental divorce (< 12 years old)					0.087*	0.044
Parental SES * parental divorce (< 12 years old)					−0.016	0.032
Rejection * parental divorce (< 12 years old)					−0.092***	0.034
Deviant friends * parental divorce (< 12 years old)					0.118 Ϯ	0.061
Lagged EPB (T–1)	0.727***	0.037	0.573***	0.045	0.581***	0.044
Intercept	0.045***	0.011	0.244**	0.086	0.313***	0.087
Adjusted R squared	0.619		0.646		0.652

Source: CrimeNL wave 3 and 4 (2014, 2015); Significance levels: Ϯ = sig<0.10, * = sig<0.05, ** = sig<0.010, *** = sig<0.000

The control variables indicate that EPB in adulthood is likely to decrease for females and those of older ages. Education level was barely significant, implying that increasing educational attainment slightly reduces the likelihood of increased EPB. Additionally, those who already showed EPB (EPB T–1) were more likely to show increased levels of EPB over the studied one-year period in adulthood.

Model 3 in Table 3 presents the interaction effects of the three distinguished dimensions among respondents who experienced a parental divorce before age 12.⁵ The impact of each of the three dimensions on changes in EPB over a one-year period in adulthood was different for those who grew up in an intact family versus those with divorced parents. Regarding the individual dimension, our results suggest that the relationship between self-control and changes in EPB differs between those growing up with divorced parents and those in an intact family. The interaction estimate in model 3 (b = 0.087) is positive, meaning that the relation between self-control and changes in EPB becomes less negative and eventually even positive for those who experienced a parental divorce (b = 0.076)⁶. Those who experienced a more vulnerable position (i.e., a parental divorce) from a young age (younger than age 12), benefited less from having a high level of self-control in moderating changes in EPB over a one-year period in adulthood, compared to their peers who grew up in an intact family.

For the family dimension, differences were found for parental rejection. Those who grew up in an intact family and experienced parental rejection were in fact likely to have an increase in levels of EPB over the studied one-year period (b = 0.010). Those who experienced parental rejection in a disrupted family were likely to exhibit reduced levels of EPB (b = –0.082). Thus, parental rejection during childhood seems to relate differently to changes in EPB in adulthood, depending on whether one grew up with divorced parents or in an intact family.

The influence of the last dimension, peers, also appears to differ depending on family structure. Regardless of whether a parental divorce was experienced, having more deviant friends over time is related to increased levels of EPB over the studied one-year period. However, for those who experienced a parental divorce this relationship is stronger (b = 0.190) than for those who grew up in an intact family (b = 0.072). Hence, an increase in deviant friends seems hazardous for the development of EPB, and even more so for those who experienced a parental divorce.

Conclusion and discussion

More and more children grow up with divorced parents. This is reason for concern, as studies have repeatedly linked experiencing a parental divorce in childhood and externalizing problem behavior (EPB). EPB is not only distressing to the individual and family, it also poses a risk to society, such as in crime and in high healthcare costs (e.g. Laub and Sampson 2003). Our study sought to explain in more detail the link between family structure and EPB. Though the relation between parental divorce and EPB has been well-established, research on whether this is an enduring correlation remains scant. We asked whether the experience of a parental divorce in childhood influences levels of EPB in adulthood in the Netherlands. We then focused on whether having experienced a parental divorce continues to be influential in later life, by studying changes in levels of EPB over a one-year period in adulthood. We argued that experiencing a parental divorce in childhood is likely to affect EPB in multiple ways. Specifically, we argued that three recognized, but different explanatory dimensions (individual, family, and peers) are mitigated or intensified among those who experienced a parental divorce. To empirically answer our research questions, we applied linear regression and LDV models using data from the Dutch CrimeNL panel study (Tolsma et al. 2014, 2015).

Our results showed that those who grew up with divorced parents tend to exhibit more EPB in adulthood, which hints towards the chronic strain model (e.g., Amato 2000; Spruijt and Duindam 2005; Van der Valk et al. 2005; Wallerstein and Lewis 2004). There were even indications that having experienced a divorce in childhood may relate to an increase in levels of EPB in adulthood. We thus, to a certain extent, found support for a long-term influence of growing up in a family disrupted by divorce among children who experienced that before age 12. Considering that life course persistent EPB often start at a young age, it is important to start and prevent it as early as possible (Moffitt 1993; Tuvblad et al. 2011). Policies should focus more on those growing up in divorced families and specifically to those who are susceptible to EPB, in order to get an early detection and prevention of the development of long lasting EPB. When considering other known factors, the relation between parental divorce and EPB is only marginally significant. An individual’s level of self-control and, especially, number of deviant friends seem to substantially mediate the relation between experiencing a parental divorce in childhood and EPB in adulthood. We found parental SES and a rejecting parenting style to be unrelated to EPB levels in adulthood.

Regarding changes over a one-year period in adulthood, an increase in the number of deviant friends implies an increase in EPB levels over a one-year period in adulthood. An explanation for this may be found in the balance perspective addressed by McGloin (2009). It states that individuals tend to shape their own EPB levels to the mean level of problem behavior among their peers. As such, increased deviance among friends would lead individuals to increase their own level of EPB (McGloin 2009). We also found that the number of deviant friends mediates the relationship between having experienced a parental divorce and changes in EPB levels in adulthood. Thus, to prevent or reduce EPB in adulthood, especially for those growing up with divorced parents, guidance regarding friendship patterns seems to be of importance.

Our second research question asked whether the three dimensions (i.e., individual, family, and peers) have a weaker or stronger impact on EPB when growing up with divorced parents versus in an intact family. Looking at all dimensions, we found no differences in adult EPB levels between those two family structures. Thus, the impact of self-control, parental rejection, family SES and number of deviant friends appears to be no more or less pronounced among children with divorced parents.

However, surprisingly, factors in all three dimensions did have a differential influence on changes in EPB over a one-year period in adulthood. If levels of EPB changed in adulthood, then the relation with self-control, parental rejection and deviant peers was different for those from divorced versus intact families. In the individual dimension, having a high level of self-control provided less protection against an increase in EPB in adulthood among those who had experienced a parental divorce. For the family dimension, parental rejection influenced changes in EPB in adulthood, though parental rejection was more problematic among those who grew up in an intact family than among those who experienced a parental divorce. Children with divorced parents may be exposed to multiple disadvantages, stressful situations and disadvantageous parental influences in the period before and after the divorce, probably more so than children in an intact family (Laub and Sampson 2001; McLanahan and Sandefur 1994). Growing up in such a situation may diminish the impact of parental rejection and self-control on changes in EPB in adulthood. In the peer dimension, an increase in deviant friends led to increased levels of EPB in adulthood, even measured over just a one-year period. Though the consequences of having deviant friends are problematic for all, nevertheless, more acute among those who experienced a parental divorce. This may be related to less attachment to the parental home than to peers (Laub and Sampson 2001; McLanahan and Sandefur 1994). We recommend future research to focus more in detail on these interesting findings. Overall, policymakers should consider that changes in EPB in adulthood are differentially influenced by factors like family, self-control and peers by those having experienced a divorce or not.

It warrants mention that the conclusions drawn here might be slightly biased due to an overrepresentation of higher educated and older people in the dataset. Both groups are known to be less affected by EPB and generally have higher levels of self-control (e.g. Tittle, Ward, and Grasmick 2003). Future research could improve on this deficiency. In the meantime, caution should be exercised in generalizing our results to the whole population of the Netherlands. Parental education as an indication of parental SES may be another limitation in our study. This might also explain why in contrast to Theobald, Farrington, and Piquero (2013) we did not find a moderating effect of those who grew up with divorced parents and family SES. Although research has shown that parental educational level is a proper indication of a family SES (Mandemakers and Kalmijn 2014; Wright et al. 1999), to provide more insight, we recommend including a more comprehensive measurement of family SES in future studies exploring EPB.

Also, for future research, it would be valuable to have data for a longer period, as we now were limited to study changes in a one-year period in adulthood. Longitudinal data for a longer period will make it feasible to detect how family life, and especially a parental divorce in childhood, shapes behavior and behavioral changes in adulthood. Further, in this research we could only focus on broken homes, defined as those marked by a divorce or separation. However, step-parent families may differ from both intact families and divorced single-parent households. In particular, the quality of the relationship between the child and parents or step-parents could be considered. Previously, social control theory addressed the role of quality bonds in discouraging delinquent behavior (Hirschi 1969). As such it might be interesting to further test the relationship between EPB and family structure. A question could center on the role of relationship quality, rather than quantity, in tendencies towards EPB (Mack et al. 2007). To do so, future research would need to disentangle various types of vulnerable home situations of children and adolescents, and seek to determine whether they have different effects on EPB in the long term. Furthermore, it should be noted that our data did not allow us to take into account factors predating a divorce. We are well aware that the effect of a divorce might be diminished when pre-divorce factors are accounted for (Furstenberg and Teitler 1994). However, our findings on the influence of divorce on EPB are in line with others that did take pre-divorce factors into account.

Overall, this research produced relevant insights on detrimental consequences of a parental divorce that carry on into adulthood. This underlines the importance of research on this particular group. Children growing up with divorced parents seems to be more inclined to develop EPB. Therefore, it is of interest to focus on this group and on factors that could make them less susceptible to EPB.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Sacha Sillekens

SACHA SILLEKENS holds a bachelor’s degree in sociology and a research master’s degree in social and cultural science both from the Radboud University, Nijmegen, the Netherlands. She currently resides in Copenhagen, Denmark. Her research interests include youth criminality, criminality, child abuse, victimization, social capital, gender equality, youth behavior, parenting, work-family life, and media education.

Natascha Notten

NATASCHA NOTTEN works at the Radboud University, Department of Sociology. Her research interests include social stratification in media use, educational inequalities, cultural participation and risk behavior. More specifically, she focuses on family issues, parenting practices and the intergenerational transmission of (un)equal opportunities. Natascha is an expert on parents’ socialization efforts regarding media literacy and their lasting consequences.

Notes

¹ CrimeNL is a collaboration between the Department of Sociology of Radboud University and Statistics Netherlands.

² The cities are Amsterdam, Rotterdam, ‘s Gravenhage, Utrecht, Eindhoven, Tilburg, Almere, Groningen, Breda and Nijmegen.

³ Correspondingly, the Dutch higher secondary school tracks (HAVO and VWO) are included under tertiary, and the lower secondary-school track (VMBO) is categorized as higher secondary. It is plausible that these respondents will complete that specific educational level.

⁴ Testing the interactions individually did not alter the results. Including the interaction with the other family structure levels also did not alter the results, and all those interactions were insignificant.

⁵ Including the interaction with the other family structure levels did not alter the results, and all those interactions were insignificant.

⁶ Calculation example of the interaction effect: Self-control: b = −0.011 self-control*parental divorce: b= 0.087 (−0.011+0.087=0.076).