Background
In the past 20 years, family dissolution has become more common in most Western countries and it is estimated that about half of first marriages will be dissolved [
1,
2]. A little more than half of all divorces involve children [
2]. In 2015, 27% of all children in Denmark under the age of 18 years living at home shared an address with only one parent [
3]. In the last decades, several studies have found that children with divorced or separated parents had less favourable outcomes, including academic achievement, psychosocial well-being, self-concept, as well as a higher risk of dropping out of school than children living in intact families [
4‐
7]. These less favourable outcomes in children, which are seen both immediately after the divorce and in a longer perspective, are similar to the outcomes found in interparental conflict [
5,
8]. Indeed, conflict levels between parents before, during, and after the parental divorce may explain more about children’s adaptation to parental separation than the actual event of divorce. Interparental conflict may engender attention problems, self-blaming attributions, elevated conflict with peers as well as general emotional and class-room difficulties leading to reduced academic performance in school children [
8].
Parents are important resources for the child, providing emotional support, practical assistance and guidance and can serve as role models to teach their children social skills [
9]. Thus, the family constitutes a key social setting and, even if parental conflict may not be present, the absence of one parent may be problematic for the child’s socialization [
6,
10]. From this perspective, it has been hypothesized that children have a higher level of social well-being if divorce occurs when they are older rather than younger because a considerable part of the socialization process takes place early in the child’s life. Parental dissolution seems to have relatively few consequences for children at college and university age level, presumably because of their maturity and independence from the family [
6,
9,
10].
Previous cross-sectional and prospective studies have examined the association between family dissolution and social well-being in children aged 11–18 years. The associations were estimated on sample sizes varying from 978 to 13,953 children and based on various measures reflecting social well-being, including popularity, cooperativeness, peer relations, loneliness, being bullied, perceived social disintegration, and lack of joy in school [
6,
7,
11‐
18]. The majority of these studies found that children from dissolved families have a poorer outcome than children from intact families; a few studies found no association. However, most of the studies were based on self-reported data on family dissolution [
6,
7,
11‐
14,
16,
17] introducing potential bias or they would only include data on families legal dissolution by divorce or separation, leaving out couples who live together but are not married [
6,
7,
11,
12,
15,
16,
18]. Furthermore, some studies based social well-being of children on teachers’ or parents’ reporting [
6,
7,
14,
18], thus using other informants than the children themselves. Many of the social outcomes in previous studies refers to the school setting, but few of the studies have focused strictly on this particular setting. The school setting is a central part of children’s daily life and may be seen as the single most important social setting outside the home where children spend many hours during a day [
19]. In a prospective epidemiological study, poor social well-being has been associated with lower academic performance in school and higher risk of severe mental health problems among a representative sample of 2790 adolescents [
20]. The school setting may play a substantial role in early identification of children at risk of poor well-being and in need of support in case of parental separation. Thus, the main objective of this study was to investigate the association between family dissolution and children’s social well-being at school and secondly, to investigate how the association may vary according to the child’s age at the time of the family dissolution. Building on the knowledge in previous studies [
4‐
7], we hypothesized that children from dissolved families had a higher risk of low social well-being at school compared with children from intact families, and that the risk increased the younger the child was at the time of the family dissolution.
Discussion
This historic cohort study found that children from dissolved families had increased odds for low social well-being at school compared with children from intact families. Stratification by age revealed that children between 9 and 12 years had higher odds for low social well-being at school than children between 13 and 16 years. Furthermore, the results showed that the younger the child was at the time of the family dissolution, the higher the odds for low social well-being at school.
These findings can be seen as a support of the hypothesis based on the Parental Loss Perspective, which emphasize that the family constitutes a key social setting and the absence of one parent may be problematic for the child’s socialization [
6,
9,
10]. Following this, children are expected to have a higher level of social well-being if family dissolution occurs when they are older rather than younger, because a considerable part of the socialization process has already taken place. The importance of the child’s age at family dissolution might also be explained by the younger children from dissolved families may have experienced more changes in family structure, e.g. having stepparents. However, when stratifying for number of changes, the youngest children at the time of the family dissolution had consistently higher odds of low well-being compared to older children.
Parental conflict has been well-documented as the factor explaining most of the negative effects of family dissolution [
6,
8,
28]. Unfortunately, we did not have available data on this. If our study aimed at examining family dissolution per se, parental conflict should be adjusted for, and the association in this study would most likely be weaker. Instead, we used family dissolution as an indicator of the process of family dissolution [
2]. By adopting a process-oriented perspective on family dissolution, parental conflict should not be adjusted for as it is a substantial part of the process [
29].
Our results are in line with previous studies examining family dissolution and various aspects of children’s social well-being [
6,
7,
11,
13‐
15,
17]. However, three studies found no association between family dissolution and children’s social well-being [
12,
16,
18] including a Danish and a Norwegian study [
12,
16]. The Danish study of 978 adolescents investigated parental divorce alongside with adolescents experiencing change of residence [
12]. The study found no significant association between the movers and divorce group and a normative reference group in relation to perception of peer-related loneliness. The Norwegian study of 4127 students aged 11–15 found no significant association between children living with divorced single mothers and social disintegration when compared with children living in intact families [
16]. The study did, however, find an association between girls living with divorced single mothers and being bullied. The definition of exposure groups in both studies differed from our study. Furthermore, the information of family dissolution only included family dissolution by legal divorce and was based on self-reports and thus introducing potential bias. A possible explanation for the null-findings in the aforementioned studies has its origin in The Stress Relief Hypothesis introduced by Wheaton (1990) and contends that a stressful life event may actually have beneficial effects on children when divorce is an escape from a harmful, high-conflict environment [
30]. Furthermore, according to the findings of Wallerstein and Kelly, the school was a sanctuary for some children of divorced parents [
31]. Our findings did not support these possible explanations.
Strengths and limitations
The major strengths of our study were that our analyses were based on a population sample extracted from national registries with full set of variables, enhancing statistical strength and eliminating recall bias and exposure misclassification. Furthermore, questionnaire data on social well-being were reported by the children themselves and not by teachers or parents as in previous studies [
6,
7,
14,
18]. Teachers and parents might not be suitable to report children’s perspective since they only have a partial picture of the child. The teachers are unable to evaluate the child’s well-being outside the class setting and the parents are unable to evaluate the child’s well-being in a school setting. Furthermore, teachers may only meet the children during lessons [
32]. Support for using children as informants was found in a study concluding that teachers judge a child of divorce less on the basis of his/her observed behaviour and more on the basis of preconceived stereotype expectations to a child of divorced parents [
33]. Also, a meta-analysis found that effect sizes based on the reports of parents and teachers tended to be weaker than effect sizes based on reports from children. This suggests that parents and teachers either underestimate the children’s problems [
6] or that children exaggerate their problems. Only children attending 4th–9th grade in ordinary public schools were included, as the questionnaire is considered more valid from the 4th grade [
34]. In addition, excluding children who lost a parent due to death and children not living with any parent e.g. children placed in care, maintained the focus of the study on the influence of family dissolution.
The study has some limitations. The first is related to family dissolution being constructed using an annual registration of family structure estimated at the 31st of December the previous year. The only knowledge of family structure in the year of the child’s birth was the one applicable on 31st December where the child had to live with both parents to be included in the study. This entailed that a child who lived with both parents from birth but experienced family dissolution before the 31st of December in the first year of life would not be included. Unfortunately, information of how many children this applies to was not available, but approximately 4 % of parents in Denmark separate when the children are 1–2 years old [
35]. Furthermore, it was only possible to track one change in family structure per year in the registries, reducing the validity of this particular variable. The reduced accuracy of family dissolution also affected the accuracy of the child’s exact age at the time of the family dissolution, leaving room for small variations. Using family dissolution as an indicator of parental break-up as opposed to divorce may, however, be viewed as a strength, because people may live together without being married. However, circumstances such as living apart due to work conditions while being in a continuous relationship should also be considered.
A second limitation relates to social well-being at school measured on a scale that has not yet been validated. Thus, ability of the scale to detect true positive and true negative cases with the particular cut-off value in the study is unknown. As a result, there is a risk of outcome misclassification. However, since the possible misclassification most likely did not depend on exposure it would be non-differential implying bias towards the null hypotheses. A study assessing the psychometric properties of the questionnaire proposed a different four-factor structure [
27]. However, our sensitivity analysis did not change the results, supporting the internal validity of the scale used to measure social well-being in this study. The scale aimed at measuring social well-being in a school context, thus the construct validity of the scale depended on whether or not the children’s answers related to their social well-being at school and not their social well-being in general. We were unable to test this. The framing of the questions by including “school” could suggest that the validity was not compromised. Only the question about loneliness did not guide the child to focus on the school setting, thus leaving room for answers about general loneliness. The validity might be compromised because of the questionnaire being filled in while being among classmates. This phenomenon has been seen in interviews with children who would avoid answering questions, that they did not want their classmates to know the answers of [
36]. Assuming this was due to low social well-being at school, this could indicate two problems. If one way of avoiding answering was to underreport their actual problem, it would most likely be independent of their exposure status causing non-differential outcome misclassification. If another way of avoiding answering was to use the option “I don’t want to answer”, this would increase the likelihood of being excluded. Assuming excluded children had low social well-being at school, this would entail selection bias if exclusion of children also depended on exposure. A significant difference in the distribution of exposure group was found among included and excluded children, where family dissolution was more common among excluded children (not shown). Thus, the study was most likely subject to selection bias causing underestimation of the association. Selection bias could also be apparent in children not filling in the questionnaire. Unfortunately, this could not be investigated as no data were available.
Even though the use of registry data was a major strength of this study, it should be noted that the data were collected for administrative purposes or solely in order to produce public statistics and not for research purposes [
37]. Some registry data on parents’ education and ethnicity were missing, but there was no reason to believe that the missing data would result in selection bias as it did not depend on exposure status.
As we used data from the first National Well-being Questionnaire it was not possible to adjust for prior social well-being at school, i.e. if the children from dissolved families had low social well-being before the family dissolution. However, if data on prior social well-being at school was available, the question of whether it should be adjusted for emerges. A study found evidence of children being affected by the disruption process at least 2–4 years prior to the actual family dissolution, possibly as a result of parental conflicts [
6,
28,
29].
Conclusion
In conclusion, this study found that children experiencing family dissolution had a significantly higher risk of low social well-being at school compared with children from intact families. Furthermore, the younger the child was at the time of the family dissolution, the higher the risk of low social well-being at school.
The school may be an important setting where children at risk of poor well-being as a result of parental separation can be identified and receive help and support. Here, the health visitors in collaboration with a pedagogical and psychological consulting team could be central; example by offering group sessions to children who experience family dissolution. Future studies should address the importance of the child’s age at the time of the family dissolution as well as include possible predictors of the increased risk among the youngest age group to improve identification and support of these children.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.