Child Custody Arrangements: Their Characteristics and Outcomes



This review was impeded by gaps and weaknesses in the social science literature on custody arrangements in general and on shared custody in particular.[4]

3.1 Research Topics

Relatively few studies explore the factors that affect the choice of custody arrangement made by separating and divorcing parents, or the effects of those decisions on children and parents. An important omission is the absence of data on the number and characteristics of parents who (a) initially selected their final custody arrangements; (b) settled on a specific custody arrangement after mediation or negotiation; and (c) were ordered by the court to establish a specific arrangement. The changes over time in custody arrangements—the move from shared custody to other arrangements, for example—are little studied, with the consequence that the reasons for these shifts in arrangements are not well understood. Furthermore, we know little about how shared custody is operationalized in daily living. "How parents actually work out, structure, and manage co-parenting subsequent to divorce is largely unknown" (Arendell, 1995a). Data on the costs associated with different custody arrangements are also very limited.

Little data are available on the effects of custody arrangements on children of different ages. Nord and Zill (1996) comment that the following questions are almost never researched:

Although researchers have made efforts to answer the question concerning "what is the best custody arrangement for the children when divorcing parents are in high conflict," only one or two good-quality studies have directly tackled this issue. More research should be conducted in this area.

3.2 Study Designs

3.2.1 Sampling

Researchers often use samples that are not randomly drawn from the total population of separating and divorcing couples. Furthermore, the emphasis has been on middle-class families. There is scant information on families outside of the middle class and those from minority groups. Samples of convenience, such as volunteers and couples who are in mediation or referred to counselling (clinical samples), are frequently found in the literature. Exceptions to this general rule are the studies based on data from the National Longitudinal Survey of Children and Youth (NLSCY) conducted by Statistics Canada and Human Resources Development Canada, and several national surveys in the United States. Therefore, possible bias in the selection of study participants resulting from the use of non-random samples is a major drawback in the research literature on child custody arrangements after separation and divorce.

3.2.2 Court Versus Survey Data

Some studies use court data exclusively. However, in real life, custody arrangements often bear little resemblance to what is found in court files (Maccoby and Mnookin, 1992). The factors that are related to the type of custody arrangement in court files may not be the same as the factors that affect the custody arrangement in the long run. More survey data, particularly longitudinal data on divorcing couples, are required to address a number of research interests, such as the extent and nature of changes over time in inter-parental conflict, children's adjustment to divorce, and the reasons why changes in custody arrangements are made.

3.2.3 Cross-sectional Versus Longitudinal Designs

A cross-sectional design compares different groups of people at the same point in time. A longitudinal design involves data collection from the same people in two or more time periods or "waves." Because of the lower cost and shorter time frame, most research uses a cross-sectional rather than longitudinal design. Cross-sectional designs have a number of disadvantages, a primary one being that controls cannot be introduced for pre-separation family dynamics and parental and child adjustment cannot be taken into account.[5] Cross-sectional data are therefore unable to capture the complete picture of the effects of these factors on custodial arrangements or parenting roles. Studies that fail to take into account the pre-existing difficulties of families have a tendency to overstate the effects of separation.

3.2.4 Data Sources

Some of the most sophisticated and rigorous in-depth studies of custody arrangements were done in California; for example, the Stanford Child Custody Project was conducted in two counties near San Francisco. Other studies were also confined to local communities in the United States, rather than being national in scope. The only research on expenditures by non-resident parents and the costs of visitation or contact by non-resident parents was done in Australia. The applicability of the findings to Canada is not known.

Both surveys and smaller scale research, with some exceptions, use parents (mostly mothers) to report on their post-separation experiences and on the well being of the child. Fathers are generally excluded from the study design. Validity problems arise with regard to parental assessments of their children's adjustment. Parents may be reluctant to reveal problems encountered by their children and their subjective assessment of the children's post-divorce adjustment may be particularly vulnerable to this tendency. Alternatively, parents may not be aware of behavioural problems such as delinquent activity. Relatively few researchers conducting surveys and smaller scale studies actually interview or otherwise assess the children (e.g. by means of teacher reports).

3.2.5 Other Problems

Because of difficulties in locating parents and children in non-standard custody arrangements, such as sole paternal custody, split custody and shared custody, much of this research on these arrangements involves small numbers and unrepresentative samples from specific communities. Because of the formidable expense and other problems inherent in obtaining a large enough sample for analysis, these arrangements have not been well studied. This is a reason why many analyses "piggy-back" questions on custody arrangements onto large scale surveys, such as has been done in Canada with regard to custody and access issues added to the National Longitudinal Survey of Children and Youth.

Some research, even that published in the late 1980s and early 1990s, uses survey data from two decades ago. Many aspects of custody arrangements have changed in the past 20 years.

3.3 Analysis

Multivariate analysis using regression or logistic regression is rarely undertaken in research on living arrangements. Since a number of independent variables (e.g., family income; conflict between parents) are related to the dependent variable (e.g., child outcomes), it becomes important to determine which, if any, of the independent variables is significantly correlated with the dependent variable, taking into account the various correlations that may exist between the independent variables. Multivariate analysis helps to determine the relative influence of all of the various forces that can affect the dependent variable. If possible, all "confounding" factors should be included in the analysis. For example, to determine the effects of custody arrangements on the well-being of children, the analysis should control for socio-economic status: social class variables such as parental income affect well-being and are also strongly associated with custody arrangements.

Because many surveys are conducted for reasons other than an examination of child custody arrangements, the variables required for a complete examination of child custody are often not available. Analyses are therefore impeded because the data source lacks all the information desirable for analysis.

Finally, surveys in the United States frequently include respondents who were never married and do not control for marital status (divorced or separated parents versus those who never married). This may affect the relationships among the variables. Large differences in many aspects of custody, access and child support arrangements were found for various types of unions in the National Longitudinal Study of Children and Youth in this country (Marcil-Gratton and Le Bourdais, 1999) and we suspect that these differences would be even larger in the United States.

3.4 Summary and Conclusions

Weaknesses in the social science literature on child custody arrangements include major gaps in the topics examined, flaws in study designs, especially sampling, and inadequate analysis.

The quality of available child custody research varies greatly. Some of the research is poor in quality. In some cases, researcher bias appears to negatively affect the soundness of conclusions drawn. This applies primarily to papers that were not subject to the rigour of peer review, but even some journal articles that were presumably vetted by colleagues show bias, usually in favour of shared custody. Lower quality research in peer-reviewed journals tends to be found in less prestigious journals and in papers from the 1980s and earlier. Particularly problematic are some review articles that contain misleading conclusions and inaccurate citations. For example, research is cited to support a generalization but no such support is found upon review of the original study.

Despite the drawbacks to the research literature on child custody arrangements, several excellent, well-designed studies have been undertaken.