New Studies, II

Writers intent on denying or questioning the connection between family disintegration and bad outcomes for children constantly remind us that “correlation does not prove causation.” The mere fact that children in one-parent homes are much more likely to be poor, for example, does not prove that single parenthood itself increases the likelihood that a child will be poor. The actual cause of child poverty could be . . . something else, which researchers have not “controlled” for.

In her Chicago-Kent Law Review article (which I discuss elsewhere), Martha Albertson Fineman trots out this accusation with considerable fanfare. The fact that children from two-parent homes do better in school, she says, does not show that single parenthood increases the likelihood that a child will do poorly in school. The real source of the problem could be, probably is . . . something else. Shame on all those scholars who forget to take these other variables into account. Yet when it comes to naming names, Fineman suddenly loses her voice, failing to cite a single specific example of the problem she is describing. This reticence is not surprising, since the charge itself is largely nonsense. Twenty years ago, maybe she could have located enough meaningful examples to sustain the point. But today, researchers typically bend over backwards on this issue of “controls,” especially when the topic is the effects of family structure on children. 

For example, a recent study based on data from the National Household Survey on Drug Abuse finds that rates of drug use and abuse are lowest among adolescents living with their two biological parents. And yes, the researchers “controlled” for everything they could think of that might reduce - do what scholars call “make go away” - the apparent causal influence of family structure. Yet the correlations they found between family structure and drug use do not appear to be artifacts of other conditions: “Economic resources and residential mobility fail to explain these relationships, thus casting doubt on their ability to explain the association between family structure and an important adolescent behavior.”

In general, skin color, place of residence, family income, gender, parents’ educational attainment - all of these variables matter. But so does family structure.   

Sources: Fineman, ibid. John P. Hoffmann and Robert A. Johnson, “A National Portrait of Family Structure and Adolescent Drug Use,” Journal of Marriage and the Family 60, no. 3 (August 1998). 

First published Winter 2000.