Childhood Health and Behavior: A Longitudinal Study

Abstract:  Child development and chronic illness are integrally related. Numerous studies have found associations between chronic illnesses such as asthma and behavioral problems in children. However, most of these studies have not utilized longitudinal data spanning childhood. The National Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth Development (SECCYD), a study of children and their families from birth through adolescence, afforded the opportunity for such a longitudinal analysis. This investigation motivated concurrent statistical research on proper utilization of linear mixed model for longitudinal data. Specifically, the study of childhood asthma and its impact on behavior illustrated that the accuracy of inference depends on the chosen covariance model of the repeated measures. Previous simulation studies have revealed that incorrectly specifying the correlation structure of the repeated outcomes on each individual leads to biased inference, at least in small samples. Here, it is proved that incorrectly assuming a random intercept model (i.e., compound symmetry) leads to optimistically biased inference, in both small and large samples. Simulations illustrate this bias and evaluate a variety of strategies aimed at avoiding it in longitudinal data analysis settings. These findings led to a subsequent study, examining the impact of being born late-preterm (34-36 weeks gestational age) on developmental outcomes throughout childhood.

Refreshments and socializing: 10:30 - 11:00 AM