Family structure effects on U.S. children’s well-being? Re-examining the family instability hypothesis

Previous studies have shown that children who grow up in marriage-based twoparent families fare better in terms of their well-being than children who do not. Other researchers have instead argued that these negative effects are confounded by the children’s parents characteristics affecting selection into specific family structure trajectories, which likewise affect the children’s well-being. However, researchers have been unable to account for the complex and dynamic relation between the socioeconomic conditions of individuals and their trajectories of family formation and dissolution. Here I argue that even when researchers account for selection based on observable background characteristics, the negative effect of changes in family structure experienced during childhood on children’s well-being may be biased. Exposure-confounder feedback bias may be present in this association in the form of time-varying confounders such as socioeconomic conditions which affect family transitions and are affected by them dynamically. Data from the Fragile Families and Child Well-being Study is used to empirically show this is the case. Family instability is here considered as a timevarying exposure. Estimations of effects of family transitions on multiple dimensions of children’s well-being are obtained through estimation of doubly robust marginal structural models and inverse probability of treatment weighting. It is shown that most of the effects of family instability are negative with a few noticeable exceptions. But more importantly, the paper shows that the size of these negative effects can be substantially reduced after partially accounting for an idiosyncratic selection of baseline background confounders and exposure-confounder feedback mechanisms. A discussion of these findings, as they relate to the association between socioeconomic conditions and contemporary family dynamics in the U.S., ensues.