HOW GOOD ARE THE DATA?
It is often easier in social enquiries to pose concerns and ask questions than to answer them. To what extent can the data available be taken as telling us definitively and with surety about the relationship between children’s experience of family life and family change and their well-being and achievement in childhood and adulthood and over a wide range of social, educational and economic indicators?
Researchers do, in fact, encounter a number of methodological problems and may advise care and caution in the interpretation of the results. A number of these arise because, despite the interest in and concern about children and the effect on children of family structure and family change, little research has been designed specifically to consider the effects on them in the long term. Researchers have had more frequently to use data collected as byproducts of other work or in research designed for other purposes.
A common critique of clinical studies, for example, has been that the sample from which the data are drawn is not representative of all children facing family disruption and the findings, therefore, cannot be taken as applying to them all, nor are they suitable for drawing conclusions about the children’s development over time. Moreover, without a control group of children who have neither lived in a lone – parent family nor experienced family disruption, there is no benchmark against which to measure whether and by how much the outcomes of the one group differ from that of the other.
National representative longitudinal cohort studies offer solutions to some of these problems. Such surveys do, for example, enable the outcomes for children with different experiences of family structure and change to be compared. Similarly, where data are available for the same children at different points in their childhood as well as in their adolescence and adulthood, it is possible to see whether any observed differences between their outcomes and those of their peers in intact families change over time and to allow for the possibility of these to be short or long term. This is important because problems may abate. Stress and strain may ease, for example, and behaviour and psychological difficulties improve. The ‘effects’ may wash out with time; children may ‘catch up’ lost ground in their educational achievement, for example. On the other hand, problems suppressed in childhood at the time of the disruption and in its aftermath may emerge later, perhaps as difficulties in their own relationships.
Such studies are not, however, without their own limitations, including their limited number. While the three major national longitudinal studies available to researchers were all designed to look at various aspects of children’s development and achievement—whether physical, psychological, social and/or economic—they were not selected or designed specifically to measure the effects of family disruption and lone parenthood.2 What measures of development and achievement are available from these studies may not be suitable or ideal to assess the effects, if any and in whatever direction, on children from their experiences of family disruption and lone parenthood. For example, are the measures of achievement in mathematics and reading at various stages in children’s education career, which the surveys do provide, more valuable and relevant than measures of quick or logical thinking, earlier maturity or a more flexible approach to life? Again, some forms of antisocial behaviour that are measured might be balanced by other positive behaviour that is not.
The data available also need to be interpreted with care. There may be more than one interpretation of behaviour instinctively thought of as a ‘good’ or ‘bad’ outcome. Too much quiet and compliant behaviour may indicate suppressed emotions, which may be damaging to young people’s relationships in the long run rather than reflecting psychological adjustment and well-being. Similarly, the disruptive behaviour said to be characteristic of boys in reaction to the stress and upheaval of family disruption may be more psychologically protective in the long run (Chase-Lansdale and Hetherington 1990).
These findings, moreover, are average outcomes for groups of children. They ought not to be taken, but sometimes are, as being applicable to individual children. As averages they disguise ‘better’ and ‘worse’ outcomes achieved by all children, whatever their family experience.
Nor is the observation of a statistically significant association— between some measured outcome and family disruption—a guarantee that a causal relationship exists. Researchers are clearly well aware of this and try to take account of other causal influences. But it is not yet guaranteed that they have always done so. We cannot be certain, for example, that all possible factors (other than marital disruption and lone parenthood) that may account for the observed relationship have been fully taken into account. There may be intervening variables that account for the observed association, the control variables themselves may be inadequately measured or the direction of the causal relationship may be other than is generally supposed. Financial hardship and poverty, for example, may be a cause or an effect of family change or both. Social class, income, children’s age and gender are all obvious factors that may be influential. While they need to be taken into account, researchers have not or are not always able to do so or to do so satisfactorily. Nor are these the only factors that may be important in determining how children fare, yet few others have been taken directly into account. It has been suggested, for example, that a causal relationship may not hold when other influential factors, such as marital conflict and parents’ ages at the time of the child’s birth, are taken into account and that the findings presented in the next section may be shown not to hold when investigated further (Kiernan 1992a; Ni Bhrolchain 1992; Burghes 1994). Comparison should be made of the effects on children of family disruption and lone parenthood with the effects on children of other social and economic influences, such as parental unemployment, and their own poor health or inadequate education.
The way in which the outcomes are themselves defined and measured may be questioned, as may the statistical techniques applied to them and the interpretation of the results (N1 Bhrolchain 1992,1993). Moreover, while longitudinal data have their advantages, the large time differences between the events and the measured outcomes lead to wariness about concluding that there are causal relationships. By the time the children of divorced parents reach adulthood, for example, the experience of their parents’ divorce may have been many years earlier, and there may have been many intervening experiences that will have influenced their ‘outcomes’.
There is also the difficulty of ensuring that measurements of children’s development are identical at different points in time. Can we be sure that a test of arithmetic at 16 years is equivalent to and as appropriate as that used to measure their attainment at 7 years? And can we know what sorts of measure of behaviour and psychological well-being would be thought equivalent at the two ages? Even if these difficulties are resolved, there remains the issue of what magnitude of difference— between the outcomes of children from different family settings and with different experiences of family life—would be thought to be significant to warrant concern or a social policy response?
That much of the research in this field has had to rely on two of the three longitudinal data sets (see note 2) raises the issue of how, or in what way, results from these studies of children born in the mid – 1940s and late 1950s can be considered applicable to children experiencing family disruption or lone parenthood today. Not only is family life today structured very differently from how it was then, but more families have experienced family disruption, and more children their parents’ parting and living apart. This may have changed both the baseline measurement of outcomes for children in intact families and the outcomes for the children experiencing such disruption. Whether children’s outcomes will be different as a result of their experiences being more normative will in part depend on how the experiences of family disruption and life as part of a lone – parent family themselves affect the development of children. If social stigma is a major player it might be expected that there would have been some amelioration over time in its effects. On the other hand, if individual and family relations at a personal level are more predominant influences, the effects may have changed little as a result of the their greater prominence. In all likelihood, however, both social stigma and personal relations are influential.
Looking to the United States, with its longer history both of divorce and of research into its consequences, researchers were first of the somewhat gloomy opinion that there seemed to have been little ameliorating effect on outcomes from divorce being more widespread. There are, however, some suggestions now that the consequences of divorce are less severe than was previously the case.3 Better welfare provision and the prevailing attitude towards divorce are thought to account for the relatively less damaging effects of parental divorce on children in Scandinavia than in Britain. As Martin Richards (1994) cautions, however, data are not available to test these ideas satisfactorily.