Children and parental divorce: The meaning of small effects

Rodgers (1996) presents a critique of Australian studies of children of divorce, including our own research. He argues that the small and often nonsignificant effects usually reported should not be disregarded, because even small disadvantages are serious when they apply to a large group. He argues further that the effect sizes found in the Australian studies are similar to those of US studies, which commonly do report significant findings; the difference is that the US studies usually have larger samples and thus greater power. As a corollary, the Australian studies would have had similarly significant results had they used equally large samples. We present an analysis of the effect sizes we obtained for five independent variables (divorce, gender, father - child relations, mother - child relations, and father's occupation) on three outcome variables (anxiety, depression, self-concept). Compared to the effects of the other four predictors, the effects of divorce were very weak, and in the case of anxiety, to the advantage of the divorce group. We conclude that what is needed is not generalised comparison between divorce and nondivorce groups, but more fine-grained analysis of the risk, protection, and booster factors involved.