Maternal Employment Experiences and Children's Behavior: A Reanalysis and Comment

Abstract

Recently, MacEwen and Barling (1991) published an article on the effects of maternal employment experiences on children's behavior. The authors suggest that most empirical research on the impact of maternal employment has focused on whether children of employed mothers suffer ill effects and suggest that, popular notions notwithstanding, there appear to be no consistent negative effects of maternal employment experiences on families. However, when a mother's experiences of her role is negative, whether her role is that of employed mother or homemaker, then there are detrimental effects to herself and her children. MacEwen and Barling propose two mechanisms to explain why and mothers' employment affects children's behavior. They reason that mothers' negative employment experiences create personal strains. The personal strains, turn, affect parenting behaviors. Finally, parenting behaviors affect children's behaviors. MacEwen and Barling operationalize maternal employment experiences terms of interrole conflict and satisfaction with employment, and they specify three forms of children's behavior as outcome variables, namely, anxious/withdrawn, attention/immaturity, and conduct disorder behaviors. The writers propose that maternal employment experience affects children's behavior by way of personal strain and parenting behavior. The personal strain variables are cognitive difficulties and negative mood. The parenting behavior variables are rejecting and punishing behaviors. The authors specify an empirical model which most variables at each stage affect the next stage of variables. Notably, however, the empirical specification does not provide for direct effects of maternal employment experiences on children's behavior, a specification that, unfortunately, maximizes the magnitude of the indirect effect coefficients and, further, leaves the basic issue unresolved, that is, whether the proposed process explains how maternal employment experience affects children's behavior. The authors use a series of hierarchical regression equations which they first enter two control variables, age and education, and then enter subsequent predictor variables. They estimate a model which, typically, successive variables are assumed to be directly affected by variables immediately preceding them the model. Variables with hypothesized direct effects on the dependent variable are entered last, which produces conservative estimates for the hypothesized direct effects, as is appropriate. Next, the investigators trim the model by deleting nonsignificant paths, then reestimate the trimmed model which they present graphically with accompanying beta weights their Figure 2 (reproduced below). The trimmed model reveals two significant effects the authors had not hypothesized: Negative mood, an indicator of personal strain, produces a direct effect on two measures of children's behavior, attention/immaturity and conduct disorder. Also, the trimmed model fails to confirm three hypothesized effects: Parental rejecting behavior affects neither children's conduct disorder nor attention/immaturity, and parental punishing behavior has no effect on children's anxious/withdrawn behavior. The authors discuss beta coefficients for the trimmed model and assert that in general, the trimmed model confirms the hypotheses outlined the introduction (p. 640). Their analyses and observations proceed from left to right with a coefficient-by-coefficient discussion of how each preceding variable affects the next step the model--that is, how maternal employment experience affects personal strain, how personal strain affects parenting behavior, and, finally, how parenting behavior affects children's behavior. The authors again conclude that their analyses confirm their theoretical model, and they acknowledge and speculate about unanticipated relationships their analyses expose and about relationships the estimates fail to confirm. …