Marital Transitions and Children's Adjustment - August 2000

In most western countries divorce rates have risen steadily since World War II. Although there is some evidence of a leveling-off of the prevalence of divorce in the past decade, most experts suggest that approximately 40% of all marriages end in divorce (Bumpass et al., 1990; Hernandez, 1993; Martin & Bumpass, 1989). The rate of separation has been most extensively studied in the United States, although available reports suggest comparable, if slightly lower, rates of separation in other western countries. Marcil-Gratton (1998) provides a helpful summary of this information in the particular context of Canada as well as specific findings from the National Longitudinal Survey on Children and Youth (NLSCY). Among the most obvious changes influencing marriages and families in Canada are the high rate of separation, the number of births to non-married couples and single-women, and the increasing number of families headed by cohabiting (rather than married) couples.

Because of the large and growing number of children who will live in a single-parent family, a considerable effort has been directed into understanding how this family form may shape children's development (Hetherington, 1999; see Lipman, Boyle, Dooley, & Offord, 1998 for a report from the NLSCY). It is important to note, however, that because most divorced individuals will remarry, and many of these remarriages will involve children, a large and growing number of children will also spend part of their childhood years in a stepfamily (see Cherlin & Furstenberg, 1994; Furstenberg & Spanier, 1987). Equally important, because the rate of divorce in second marriages is higher than first marriages (Clarke & Wilson, 1994), the picture that emerges from available sociological and demographic data is one of a series of marital transitions experienced by adults and children. Given the marked frequency of marital and family transitions, it is natural that policy makers, health professionals, and the general public raise questions about the implications of the changing family patterns for children's and adult's well-being.

Not surprisingly then, the consequences of family change through marital separation¹ and re-partnering on children's psychological adjustment have been widely discussed and debated in the social sciences literature and the popular press. Recent reviews of the research literature (e.g., Hetherington, Bridges, & Insabella, 1998) highlight the very robust finding of higher mean levels of a wide variety of adjustment difficulties in children in stepfamilies and single-parent families, compared with children in "intact" or non-divorced families. Furthermore, these adjustment differences exist not only in the short-term, but may also persist into adulthood to influence subsequent generations of children (Kiernan & Hobcraft, 1997; Kiernan & Mueller, 1998; O'Connor et al., 1999b; Rodgers & Prior, 1998; see also, Aquilino, 1996; Webster et al., 1995). However, despite the wide-spread research attention directed toward this issue to date, basic questions remain about the causes and consequences of these changes on children and adults. Most importantly, we still know relatively little about why some children appear to be comparably resilient to stress that accompanies family upheaval, whereas others suffer serious difficulties.

The observation that there are "winners, losers, and survivors" (Hetherington, 1989) following family transitions led investigators to adopt a risk and resilience perspective (Hetherington, 1999). This perspective is concerned not so much with assessing mean differences between diverse family types, but instead on elucidating the risk and protective mechanisms explaining individual differences (i.e., variation) among children in their adjustment associated with family transitions.

1.1 The aims of the current research

The current research had several aims. The first, most general aim was to assess the causes and consequences of family changes for children in Canada. Much of what we know about the frequency and sequelae of family type change on child outcomes is based on research findings from the U.S. and, to a lesser extent, the United Kingdom (Amato, 1996; Amato & Rogers, 1997; Brown & Booth, 1996; O'Connor et al., 1999b; Office of National Statistics, 1997; Rodgers & Prior, 1998; Schoen & Weinick, 1993; Thompson, 1994). We do not know whether the findings obtained in U.S. and U.K. samples (and the implications of those findings for policy and practice) can be generalized to Canadian families. The large and representative nature of the NLSCY sampling strategy is particularly unique in this research field, and will distinguish this study among the better extant projects assessing the connection between family and marital transitions and children's well-being.

A second, more specific aim was to identify the risk and protective factors that explain why some children have difficulty adjusting to family transitions but others appear unscathed. As noted above, although we know that there are significant and meaningful mean level differences in children's behavioural difficulties according to family type membership, researchers have been less successful in explaining why there is such great variation in children's adjustment. Moreover, there is even debate concerning whether the risk for adjustment difficulties arises from the family type per se (e.g., see Cherlin et al., 1991; Forehand et al., 1997). It may be, for example, that the causal risk factors predated the divorce (e.g., in the form of marital conflict, Amato & Rogers, 1997; Davies & Cummings, 1994) or pre-dated the current family formation (e.g., in the form of the number of previous relationship transitions and parental psychopathology, Capaldi & Patterson, 1991; Dunn et al., 1998; O'Connor et al., 1998). There is even suggestive evidence that the connection between parental divorce and child adjustment may be partly genetically mediated (O'Connor et al., 2000). The longitudinal design of this study is therefore essential because it allows us to study changes in child adjustment following a (further) parental separation and family re-organization. We are therefore able to discern to what extent changes in family re-organization predict changes in children's well-being.

1.2 Conceptual and methodological considerations in research on family influences

Identifying the risk and protective processes that explain children's adjustment in diverse family types is complicated for several different reasons. We will highlight the particular issues that are especially relevant for this project.

The first consideration is the definition of family "type". Defining family type is a surprisingly complex task. Numerous definitions of family type have been proposed. Most often the definitions are developed for very specific purposes or, more often, because of the particular and often idiosyncratic sample of families included in a study. For example, for census purposes, a stepfamily is often defined simply in terms of a family in which there is at least one dependent child from a previous relationship from one or both partners. This definition is not satisfactory for our purposes, however, because psychological research indicates that risk factors are far less common in "simple" stepfamilies than in "complex" stepfamilies (Hetherington et al., 1998). Accordingly, in this report, we distinguish between different stepfamily forms (see definitions in section 2.2).

The second conceptual and methodological consideration in the study of family transitions and child adjustment is the covariation among risk processes. That is, although there is now support for several key risk and protective variables (Amato & Keith, 1991; Hetherington et al., 1998), it is clear that there is a substantial overlap among the risks involved. Thus, family type and separation likely act as a proxy for multiple kinds of risks, from disrupted parenting to community norms of family life. Furthermore, even within the family there are multiple and overlapping risks, such as parent-child conflict, sibling conflict, marital conflict and socio-economic adversity (Hetherington & Clingempeel, 1992). Progress in understanding the processes distinguishing resilient from stress-affected children will occur to the extent that we are able to go beyond identification of a risk indicator to determination of specific risk mechanisms indexed by the indicator (Rutter, 1994). Only by this means can rational decisions be taken on "target" areas for interventions (e.g., how to support families experiencing a divorce or remarriage). Multivariate analyses may help determine which risk factor is "driving" the risk process, but this approach does have limitations. A better strategy, and one available to us in this study, is to focus on within-individual change over time.

A third conceptual and methodological consideration in the study of risk mechanisms concerns the distinction between between-family and within-family differences in child outcomes. This distinction has been highlighted in recent research in developmental psychology and behavioural genetics. On the one hand, we can explain variation in children's adjustment according to the characteristics of the family they live in. That is, we can examine why children in different families have disparate patterns of adjustment, or between-family differences in children's outcomes. In other words, by virtue of being in a particular family we might assume that children would "share" certain experiences - including family type and parental separation - as well as the effects of those experiences. Accordingly, children in the same family might be more similar to one another than children in different families (of course, sibling similarity may be explained by other factors as well, notably genetics).

On the other hand, we also know that children in the same family differ from one another on the range of important outcome variables (Plomin & Daniels, 1987). Because risk factors that operate at the family level (or between-family) may not explain why children in the same family differ from one another, it is important to consider the complementary view that risk factors may also operate at the individual child level. We refer to individual child-level effects as within-family variability. Research findings highlighting within-family variation or sibling differences in children's adjustment are important for several reasons. First, they force a re-consideration of the assumption that many psychosocial risks operate in a family-wide basis and therefore affect siblings similarly. Second, these findings open up new avenues to the study of resilience because there is no certainty that risk processes explaining between-family variation in children's adjustment are the same as those that explain within-family variation.

Almost without exception, prior research on children's adjustment to divorce and re-partnering was based on designs that assess one child per family. This is also true of the research on risk and resilience in children more generally, involving risks such as poverty (e.g., McLoyd, 1990), marital conflict (Jenkins & Smith, 1990), poor mental health of parents (Rutter & Quinton, 1984) and cumulative risk (Jenkins & Keating, 1998). Our knowledge of the factors that underlie between-family and within-family differences in child outcome has been obfuscated by the one-child-per-family research design. Unfortunately, when only one child per family is assessed, it is impossible to explain the extent to which variation in child adjustment can be explained by risks operating at a family-wide level (as separation and family type is presumed to operate), the individual child level, or an interaction between the two.

A novel feature of the current study is to assess whether children in the same family are differently affected by membership in a "non-traditional" family. Moreover, the longitudinal angle provides a unique opportunity to identify within-family risk and protective processes, and to distinguish those risk factors that operate at the family level from those risks that operate at the individual child level. Specifically, we examine to what extent siblings are differentially affected by their parents' separation, and what factors explain these within-family variations. Findings from this novel analytic approach would provide additional insight into the origins of children's resilience to adversity and, in addition, offer new directions for clinical and policy work with families in transition.

The hypotheses to be tested as well as the exploratory analyses are described in detail in each of the three results sections. The following section sets out the research strategy used to answer the questions raised in the introduction.

• ¹We refer to family (type) change or family transition as any change brought on by a separation or re-partnering of the residential parents. In addition, throughout this report we refer to marital separation. However, a sizeable minority of partnerships are cohabiting (non-marital) relationships, and this is particularly so in stepfamilies. Furthermore, we adopt the convention of using parental separation as the index variable rather than the legal divorce.