Introduction
Out-of-home care is considered an important intervention to ensure the security and adaptive development of infants and children when this is no longer possible in their biological family. Children living in foster care (CFC) are at increased risk for developing internalising (symptoms directed inwards, e.g., feelings of sadness) and externalising (symptoms directed outwards, e.g., aggressive behaviour) problems [
1,
2]. Additionally, former maltreatment experiences of children may lead to long-term effects on functional and structural brain development [
3], associated with behaviours that are adaptive in a maltreating family environment [
4,
5], but that may cause problematic interactions in a foster family. Given the high risk of developing behavioural and emotional problems of CFC (e.g., [
1]), the foster family care system may be crucial for the child’s development. Yet, a recent meta-analysis of longitudinal trajectories showed no changes in internalising and externalising behaviour problems and the child’s adaptive development in CFC [
6], highlighting the importance of identifying protective and risk factors within the foster family care system to create optimal support for CFC.
Generally, many factors in (foster) families have been shown to influence children’s development, such as the socioeconomic status of a family [
7], the number of maltreatment experiences in CFC [
1], and the number of foster placement disruptions [
8]. Furthermore, the (foster) parent’s parenting can play an important role in the development of children [
9]. However, most research on the relations between parenting and child development have been conducted in the general population (e.g., [
10,
11]). Therefore, these results may not be directly transferable to CFC. After entering foster care, behaviour problems may persist in spite of caregiving improvements because children have to cope with former maltreatment and/or neglect experiences in addition to adjusting to a new family and living environment [
6]. Furthermore, given the unique history of CFC, different parenting strategies may be adaptive in CFC. Thus, the study at hand aims to examine whether the longitudinal associations between child behaviour problems and parenting differ between foster and biological families, and how parenting behaviour dimensions interact with children’s internalising and externalising problem behaviour in a longitudinal study. The results of this analysis may not only give hints regarding the potential relationships between parenting behaviours and child development, but also lead to tailored interventions for foster families to support the adaptive development of CFC.
Parenting behaviour is often categorised in responsiveness (or warmth), and parental demandingness [
12], although also other categories exist (e.g., parental sensitivity; [
13]). Parental demandingness can be further differentiated into behavioural control (which is often associated to positive child outcomes), harsh control (related to negative outcomes), and psychological pressure (attempts of manipulating the child, that are also linked to negative child outcomes; [
12]).
With respect to
biological families, the body of research examining the relationship between parenting behaviour and children’s behavioural problems is well-established. Meta-analyses of studies in the general population showed small to moderate associations between dysfunctional parenting (i.e. overly strict/unclear and inconsistent parental behaviour or forms of psychological pressure) and an increased risk for depression and anxiety in children [
11,
14,
15], small negative associations between functional parenting (i.e. provision of warmth/appropriate demanding and control) and externalising problems in children [
10,
16], as well as moderate positive associations between dysfunctional parenting dimensions and externalising behaviour [
16]. Also, previous research further showed that these associations seem to be rather reciprocal in longitudinal studies (e.g., [
16]), i.e., studies showed that certain parenting behaviours may on the one hand evoke compatible child developmental pathways. On the other hand, child behaviours also may evoke certain parental reactions and behaviours.
Research in
foster families regarding the parenting and development of CFC is not as elaborated as in biological families. Vanderfaeillie et al. [
17] showed in their longitudinal study that problem behaviours in CFC were significantly associated with the use of negative parenting in foster mothers. Moreover, Chodura et al. [
18] found in their meta-analysis similar associations between parenting and child development in foster families, as had been previously found in biological families; however, most of the studies (34 out of 43) were cross-sectional. Additionally, Orme and Buehler [
7] concluded that approximately 15 percent of foster parents manifest potentially poor or troubled parenting as reported by youth welfare workers (however, of note is that comparable studies in biological families are missing). On the one hand, this is important because of the high vulnerability of CFC [
8]. Because CFC have experienced more adverse life events in early childhood, they may need a more supportive environment for their adaptive development, for instance provided through functional parental behaviours. On the other hand, poor parenting may also occur because of more problematic behaviours shown by CFC, as indicated in longitudinal studies (e.g., [
17]).
Directly compared, differences have been found in the parenting behaviour of foster and biological parents, but the evidence for the directions of the differences is mixed and limited. Fisher et al. [
19] found that foster parents showed more dysfunctional parenting behaviour compared to a general population sample of parents. In contrast, Lucey et al. [
20] found that foster parents reported less verbal and corporal punishment than biological parents. In this study, differences in parenting behaviour were explained by demographic differences between the two groups: foster mothers were found to be significantly older and married more often than biological mothers [
20]. Furthermore, Linares et al. [
21] found that foster parents reported more clear expectations (regarding rules about family routines) than biological parents
despite demographical differences. Additionally, Ehrenberg et al. ([
22]; using the same sample as in this study) analysed different risk factors for the development of CFC. They found differences in parenting behaviour and (foster) parents’ stress between foster care and biological families that were generally explained by child problems. Furthermore, they assumed interactional processes between parenting variables and child problems, which they did not investigate further because of a different focus of their analysis. Taken together, existing studies often examined longitudinal associations between parenting behaviour and child problems only in families with CFC, however, without comparing the results to biological families. Further, when comparing the two groups, such comparisons were mostly done for only one aspect, either parenting behaviour or child problems.
To summarise, the evidence found in biological families may not be transferable to foster families. There may be differences between both groups due to differences in demographical characteristics between biological and foster families (e.g., the socioeconomic status and age of the parents, [
23]), differences in children’s characteristics and behaviour, and possibly also due to differences in parenting behaviour. There is, however, currently a large research desideratum in identifying longitudinal relationships between parenting behaviour in foster and biological families and children’s internalising and externalising problem behaviour. To our knowledge, no former study used cross-lagged panel models to compare the relationships between parenting behaviours and children’s internalising and externalising problem behaviours in biological
and foster families. Therefore, it remains unclear as to whether parenting more strongly influences child development or vice versa, in particular in foster families.
Considering a well-examined theoretical model, Belsky’s process model of parenting [
24] may help us to understand such interaction effects. According to this model, parenting does have an influence on child development (e.g., child problem behaviour), but is influenced (among other context-related components) by child characteristics (e.g., temperament). Previous studies revealed that callous-unemotional (CU) traits as child characteristics are potentially important moderators of the link between parent–child relationships and developmental outcomes, in particular with respect to externalising behaviours [
25]. Children scoring high on CU traits show a lack of empathy and guilt, emotional ‘coldness’, and disregard for feelings of others together with physiological hypoarousal and reduced responsivity to punishment [
26]. As such, CU traits have been demonstrated as a serious risk factor for externalising behaviour problems and for a particularly persistent course of conduct disorder [
27]. Emerging research suggests that high CU traits can be explained by two different variants [
28]. Primary variants of CU traits are thought to emerge from genetic factors and are associated with insufficient arousal to emotional cues. Secondary variants are supposed to emerge from trauma experiences, leading to emotional numbing, avoidance and inhibition of empathy as reaction on traumatic environments [
28]. While earlier studies propose that the problem behaviours in children with high levels of CU traits become rather irresponsive to parenting—in line with primary CU variants [
29,
30]—more recent evidence suggests the opposite: parental warmth in clinically-referred, conduct-disordered children with elevated CU traits led to a decrease in the risk for antisocial outcomes [
31]. Further, in non-clinical samples, more positive dyadic parent–child relationships predicted a decrease in future behaviour problems only in children with high, but not in children with relatively low CU traits [
32]. This may imply that parental warmth is a dimension of special interest for CFC, assuming they have on average higher levels of CU traits due to experiencing more adverse life events that may lead to the emergence of higher secondary CU traits. Interestingly, a recent adoption study found that low parental warmth displayed by adoptive parents disrupted the heritability of CU traits [
33,
34]; furthermore, a recent twin study suggested that parenting is related to child CU traits and aggression, over and above genetically-mediated effects, with low parental warmth being a unique correlate of CU traits [
35]. However, neither child temperament nor other child characteristics per se shape parenting, but rather the ‘goodness-of-fit ‘ between parent and child might determine the development of parent–child relationships and the developmental outcomes in the child.
Research Questions and Hypotheses
Taking into account the potential demographic differences between samples, the goals of the current study were to investigate (i) whether foster and biological parents differ in their parenting behaviour, (ii) whether longitudinal pathways differ between foster families and biological families and how parenting behaviour and internalising and externalising problems of children are longitudinally related, and (iii) whether the relationship between parenting behaviour and internalising and externalising is moderated by CU traits.
Based on the research findings outlined above as well as Belsky’s process model of parenting [
24], direct longitudinal relationships are expected between parenting behaviours and children’s problem behaviours in both, foster and biological families, although the literature is inconsistent as described above. That is, negative associations are expected between functional parenting and low internalizing and externalizing child behaviour, and vice versa for dysfunctional parenting and child mental health. Across groups, a moderating effect of children’s CU traits is expected for the association between parental warmth and support and children’s externalising problems based on previous findings. Considering the existing literature, we could expect both: a stronger association between parenting and externalising problems in the group of children with high CU traits (e.g., [
32]) or a stronger association in the low CU trait group (e.g., [
30]).
Discussion
The aim of the study at hand was to examine the longitudinal associations between parenting behaviour and children’s emotional and behavioural problems in biological and foster families and the potential influence of children’s CU traits. In summary, no differences were found between the CFC and the CBF groups in parenting behaviours and associations with children’s internalising and externalising symptoms. Independent of the group (CFC versus CBF), the associations showed cross-lagged longitudinal effects between warmth and support and internalising and externalising problems.
The first research question examining if parenting behaviour is different between the two sample groups showed no significant differences in parenting behaviour for both parenting scales. In general, foster parents seem to show the same parenting behaviours aiming to reduce children’s behaviour problems as biological parents. Therefore, the results complement not only previous research in biological families (e.g., [
14‐
16]), but also the rather small body of research on parenting in foster families [
18].
Regarding the second research question examining the longitudinal pathways between parenting behaviour and children’s internalising and externalising behaviour, random intercept cross-lagged panel models were computed. Overall, χ
2-tests indicated that it is reasonable to assume identical longitudinal cross-lagged effects between the two groups. In summary, our results showed a high longitudinal stability for psychological pressure and no cross-lagged effects with children’s internalising/externalising symptoms. In contrast, for parental warmth and support, several cross-lagged effects were found, in particular for child’s externalising problems. Specifically, more externalising problems at T1 predicted less warmth and support at T2 and more warmth and support at T2 predicted less externalising
and internalising problems at T3. These negative longitudinal associations have also been found in previous research examining the emotional and behaviour problems of CFC (e.g., [
13]) and CBF (e.g., [
16]), indicating a rather reliable finding. In conclusion, in particular the provision of warmth and support was shown to have an influence on child behaviour and emotional problems when both parenting dimensions were investigated separately. Associations may have differed had we included both parenting dimensions simultaneously into the model. It is therefore unclear if both contribute individually above and beyond the other or if their shared variance (i.e., construct overlap) is more responsible for these results.
The fact that more autoregressive effects were found for externalising than for internalising behaviours may be explained by the poorer observability of internalising problems in general (e.g., [
59]). The present study can provide early insights into the different association dynamics between internalising and externalising problems and parenting behaviours and thus help to fill the research desideratum discussed above and to better understand the previous mixed evidence for foster families [
19,
20].
The effect sizes were generally rather low between parenting and child emotional and behaviour problems. This also had been found in a meta-analysis and literature review by Chodura et al. [
18], which showed that externalising and internalising problems were associated with parenting behaviours to a low to moderate degree. Additionally, the authors found generally few other variables in families that were associated with internalising and externalising problems in CFC, such as the length of stay in the foster family, possibly indicating that previous experiences prior to entering foster care may have a larger effect. This goes along with Goemans et al. [
6], who also found few variables influencing the trajectories of child development in CFC.
Furthermore, CFC showed significantly more emotional and behaviour problems than CBF in the study at hand, and additionally the longitudinal stabilities of internalising and externalising problems were significantly higher in CFC than in CBF (see also [
56,
60], using the same sample as the current article). This may explain some of the low cross-lagged effects in foster families. However, no differences were found between both groups regarding these longitudinal associations. That is, foster parents may show the same parenting behaviours as biological parents, indicating appropriate parenting responses, even though CFC show more and longer lasting problem behaviours. Foster parents still are seen as the most important target group for interventions to support the development of CFC, possibly due to the fact that there may be the most potential for change. This is also in line with cross-sectional and intervention studies that indicate that parenting behaviour and the parent–child relationship in foster families may positively influence the child’s stress response systems, such as the hypothalamic-pituitary-adrenocortical (HPA) axis and may thereby mitigate the possible consequences of early life stress ([
61,
62], using the same sample as the current article). Alterations in the HPA axis may constitute a risk factor the development of later mental and somatic health problems [
63]. Additionally, there may be adjustment processes that can be seen only in examinations longer than one year, i.e., rather long-term effects. Consequently, (foster) parents and youth welfare workers should not neglect the potential effects on children’s problem behaviour, even though they may be small.
The third research question investigated whether CU traits moderated the associations between parenting behaviour and children’s emotional and behaviour problems as a potential factor which could explain additional variance of these associations. In line with earlier studies, our findings suggest that CU traits indeed moderated the relationship between parental warmth and support and children’s externalising behaviour [
29,
30]. This effect, again, supports Belsky’s model, indicating that child characteristics may be important moderators of the link between parent–child relationships and developmental outcomes and explains further the longitudinal effects observed in the current study. This is, CU traits as child factors may set directions for parenting behaviour: children with high CU traits may show difficult behaviour resulting in lower associations between parental warmth and support and children’s externalising problems. However, in contrast to some previous studies in non-foster care families [
31,
32], this effect was driven by a higher association between positive parenting behaviour and the child’s externalising behaviour present only in those children scoring low in CU traits (no interaction effect was found with group, indicating no differences in this moderating effect for foster and biological parents). This finding is in line with a recent study showing that positive parenting was most relevant when distinguishing between youth with conduct disorder and high versus low levels of CU traits using a machine learning approach [
64] and with intervention studies showing lower effect sizes in subjects with higher CU traits. While our assessment methods differed from those of Kochanska et al. [
32], shared positive affect is most closely related to the ZBQ subscale
warmth and support. In line thereof, we found no moderation effects for the ZBQ scale
psychological pressure, and no moderation effects for internalising problems. This further supports our conclusion that in particular the provision of warmth and support has an important influence on the children’s behaviour problems. Note, however, that in the present study children’s levels of CU traits and parent’s levels of psychological pressure were generally low (CU traits:
SD = 0.24 for CBF and
SD = 0.40 for CFC), which might have influenced our findings. In summary, future research is needed to further examine the influence of early adverse life events, elevated CU traits, and parental warmth and support, on children’s development.
Limitations and Recommendations for Future Research
The present study used questionnaires filled in by mostly one (foster) parent. Possible effects may therefore be traceable to an effect of the same source. Yet, this is unlikely given the overall small associations between parenting and child problems. Further, Chodura et al. [
18] generally found low to medium effect sizes in their meta-analysis, considering all types of studies. Nevertheless, to further support the observed effects, future research should employ a multi-informant, multi-method approach in measuring parenting and children’s problem behaviour (e.g., including behavioural observations, reports from teachers, both parent reports). These operationalisations could give deeper insights into parents’ and children’s behaviour but may also be associated with reduced participation rates of populations [
65].
Regarding the operationalization of parenting, the internal consistencies of the parenting scale psychological pressure also should be taken into account. The low internal consistencies may have influenced the cross-sectional and longitudinal associations with child problems. One explanation could be the age of children in the sample groups. Internal consistencies of the ZBQ were higher in the validation studies [
37] with
α between 0.73 and 0.75 and should therefore be interpreted with caution.
Possible self-selection effects which may lead to more homogeneous samples should not be overlooked [
66]. The present study tried to reduce this effect by contacting case workers of youth welfare institutions and asking them to provide the study information to as many foster families as possible. Additionally, the CBF group was recruited in similar regions to reduce differences in demographic characteristics, which may have increased homogeneity. The limited generalizability also should be noted in terms of the results regarding CU traits. CFC are a group of children with certain adverse experiences and the results, therefore, are not simply applicable to other population groups.
According to Éthier et al. [
1], victims of chronic maltreatment show more emotional problems, aggressive behaviour and social withdrawal problems than victims of transitory maltreatment. The chronicity and severity of maltreatment and neglect experiences of CFC groups therefore should be considered in future research. It might be helpful to distinguish between maltreated children with and without clinical symptoms of psychiatric disorders, which would have exceeded the statistical power in the present sample. Éthier et al. [
1] also pointed out that children’s developmental improvement in the transitory maltreatment group emerged not directly after the end of maltreatment, but rather gradually. This may show that chronically maltreated children need more time to show improvements in development after the transition into a non-maltreating environment [
1], which is important information for youth welfare workers as well as for foster parents and may give better insights into differential developmental pathways in CFC.
Summary
Former maltreatment experiences of children in foster care (CFC) may lead to emotional and behaviour problems that need tailored parenting behaviours of foster parents. Former research does, however, leave a gap regarding the longitudinal associations of parenting behaviour and child’s internalising and externalising problems in foster families, and whether these differ from the associations in biological families. families are poorly understood, and it is unclear whether these differ from biological families. Thus, the goals of the current study were to examine i) differences in parenting between foster and biological parents, ii) the longitudinal associations with children’s internalising and externalising problems and iii) the potential moderation of these by the children’s callous-unemotional traits (CU traits). Data from 86 foster families and 148 biological families (mean age = 4.16 years) were analysed in a longitudinal study with three measurement times, approximately six months apart. Parenting was assessed via self-report on the warmth and support and psychological pressure scales of the Zurich Brief Questionnaire for the Assessment of Parental Behaviours. Internalising and externalising problems of the child were assessed by caregiver-report on the Child Behaviour Checklist. CU Traits were assessed at first assessment by caregiver reports of the Inventory of Callous-Unemotional Traits (ICU). First, to examine group differences in parenting, linear mixed models were computed. Second, to examine the longitudinal associations between parenting and child internalising and externalising problems, random intercept cross-lagged panel models were calculated and it was tested whether the lagged effects differ between groups. At last, multiple regression analyses were computed including group and interaction effects of parenting and CU traits. Parenting behaviour did not significantly differ between the foster and biological family group in both parenting scales. Additionally, the random intercept cross-lagged panel models showed that the lagged effects did not significantly differ between groups. Significant longitudinal cross-lagged effects were found for parental warmth and support and children’s externalising problems across groups. Overall, longitudinal effect sizes were small to medium-sized. Further, CU Traits moderated the relationship between warmth and support and externalising problems of children: in the presence of low CU traits, parental warmth and support and children’s externalising behaviour was negatively associated. No significant interaction effects were found for the other parenting and child problems relationships. The study at hand gives some support for the potential influence of parenting behaviour on internalising and externalising problems in CFC and CBF. In particular, parental warmth and support was found to influence children’s behaviour and emotional problems. These findings suggest that parenting behaviours and child psychopathology do influence each other over time reciprocally and to a similar extent in foster and biological families. However, regarding the additional influence of certain child characteristics, such as CU traits, it could be shown that there is a moderation effect on the association between parenting and child development. This might go along with a greater longitudinal stability of some psychopathological symptoms and thereof a reduced responsivity to parental warmth. In summary, the results give hints to tailored parenting behaviours to encourage adaptive developmental pathways in children, that may not always be naturally shown by (foster) parents, but can be promoted in parenting trainings.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.