Journal Issue: Children and Divorce Volume 4 Number 1 Spring/Summer 1994
How Do Researchers Study Children and Divorce?
To understand how divorce affects children, social scientists predominately rely on two research designs: cross-sectional and longitudinal.6 In a cross-sectional study,7 researchers compare children from divorced and continuously intact two-parent families at a single point in time.8 In a longitudinal study, researchers follow children over an extended period of time following marital dissolution.8
Longitudinal studies usually include a comparison group of children from two-parent families as well. Although both types of research designs have methodological advantages and disadvantages, they provide useful information about adjustment.6,8,9 Cross-sectional studies provide a "snapshot" that shows how children of divorce differ from other children, whereas longitudinal studies allow us to understand how children adjust to divorce over time.
In addition to studies of children, social scientists have studied the long-term consequences of divorce by comparing adults who experienced divorce as children with those who grew up in continuously intact families. Researchers also have carried out a small number of longitudinal studies in which children of divorce are followed into early adulthood.10
Three types of samples appear in the literature.11 Clinical samples consist of children or adults who are in therapy or counseling. Clinical samples are useful in documenting the kinds of problems presented by offspring who adjust poorly to divorce, but these results cannot be generalized to the broad majority of people who never receive professional attention. Researchers obtain convenience samples of children or adults through community organizations (such as single-parent support groups) or other local sources. Convenience samples are relatively easy and inexpensive to obtain, but people in these groups may be atypical in unknown ways. Researchers select random samples of children or adults in a scientific manner such that the sample represents a clearly defined population within known limits.12 These samples may be obtained from schools, court records, or households. Random samples allow us to make valid generalizations about the majority of children who experience divorce.13 Unfortunately, these types of samples are also the most difficult and expensive to obtain.
Researchers match (or statistically equate) children or adults in the two samples (divorced and intact) on key variables known to be associated with both divorce and adjustment.14 For example, parents of low socioeconomic status are more likely than other parents to divorce and to have children who exhibit behavioral and academic problems. Consequently, it is necessary to make sure that the socioeconomic backgrounds of parents in the two groups are comparable.
Researchers then select outcome measures that reflect children's and adults' functioning, or well-being. Common outcome measures for children include academic achievement, conduct, psychological adjustment, self-concept, social adjustment, and the quality of relations with parents. Common outcome measures for adults include psychological adjustment, conduct, use of mental health services, self-concept, social well-being, marital quality, separation or divorce, single parenthood, socioeconomic attainment, and physical health.
Social scientists gather information about children by interviewing one or both parents, questioning the child's teachers, administering tests to the child, or directly observing the child's behavior. Information is usually obtained from adults by interviewing them. Researchers then compare outcomes for those in the divorced and the continuously intact family groups. Statistical criteria are used to judge if differences in outcome measures are large enough to rule out the possibility of their being attributable to chance alone. Observed differences that are too large to be attributable to chance are assumed to be caused by divorce, or at least, by some factor(s) associated with divorce.
Unfortunately, because these studies are correlational, it is difficult to know for certain if divorce is responsible for observed differences between groups. It is always possible that groups might differ in ways that researchers cannot anticipate, measure, and control. For example, an unspecified parental personality characteristic might increase the risk of both divorce and child maladjustment. Firm conclusions about causation require experimentation; because we cannot randomly assign children to divorced and nondivorced families, our beliefs about the causal impact of divorce remain tentative.