Methods – The study used data on 6,855 matched mother-child pairs from the National Longitudinal Survey of Youth (NLSY). I used multiple imputation of missing data to account for missing data due to sample attrition and respondent non-response. Three analytic methods provided successively rigorous causal estimates of the importance of timing and developmental stage: multiple regression with rich controls, residualized change models, and fixed effects models. Each method examined the association between the average hours per week mothers worked during a developmental period, and children's average BMI, the years they were overweight or obese, and the years they were obese (BMI>=95th percentile). Using both continuous and dichotomous outcome measures was important to assess both the statistical and practical significance of maternal work.
Results – Results from rich controls and residualized change models were generally comparable, finding significant relationships between average maternal work and the child obesity outcomes within and across most developmental periods; few significant associations were found for the effects of maternal work during infancy. Fixed effects models, which provided the strictest test of causality, found that average maternal work hours were only significantly associated with the number of years a child was overweight or obese during middle childhood and late childhood/early adolescence. No similar associations were found for years obese. These findings suggest that an unobserved factor was likely biasing the results of the rich controls and residualized change models.
Conclusions and Implications – This study found that average maternal work at ages 7-10 and 11-14 was significantly associated with an increased contemporaneous incidence of child overweight or obesity. These results inform both future research and anti-obesity interventions. Although much previous research has identified aspects of the environment that are associated with child obesity, few studies have attempted to determine whether developmental stage and timing are important. To the extent that future research can similarly discern which stages of childhood are pertinent to the relationship between other factors and child obesity, interventions can become more effective by tailoring them to children of appropriate age.