Carolyn E Ievers-Landis1, Christopher J Burant, Rebecca Hazen. 1. Division of Developmental/Behavioral Pediatrics and Psychology, Rainbow Babies & Children's Hospital, University Hospitals Case Medical Center, Cleveland, OH, USA. carolyn.landis@uhhospitals.org
Abstract
OBJECTIVE: The objective of this study was to offer a practical demonstration of the use of bootstrapping in structural equation modeling (SEM) with smaller samples by evaluating family dinnertime rituals at the intersection between the family social environment and pediatric type 1 diabetes mellitus management. METHOD: Participants were 77 children and adolescents aged 6 to 18 years and their mothers from a larger study of child and parental challenges related to managing the treatment regimen and correlates of adherence. SEM was used to test a model of maternal reports of the family social environment as exogenous variables with mealtime rituals and child hemoglobin A1c levels as endogenous variables. The bootstrapping procedure and the Bollen-Stine bootstrapped χ test were then applied to test the stability and appropriateness of this model. RESULTS: Using SEM, the final model had robust goodness-of-fit indicators and revealed that the family social environment characteristics of control and cohesion were related to mealtime rituals, which was associated with hemoglobin A1c levels. The bootstrapping procedure indicated that the parameter estimates were very stable, thus lending greater credence to the model. The Bollen-Stine approach provided evidence that the entire hypothesized model was not significantly different from the model derived from bootstrapping. CONCLUSIONS: This investigation offers a practical demonstration of the bootstrapping technique and the Bollen-Stine approach for testing models on smaller samples in SEM. These techniques may be used with relatively small samples, which are common in developmental/behavioral pediatrics and pediatric psychology research.
OBJECTIVE: The objective of this study was to offer a practical demonstration of the use of bootstrapping in structural equation modeling (SEM) with smaller samples by evaluating family dinnertime rituals at the intersection between the family social environment and pediatric type 1 diabetes mellitus management. METHOD:Participants were 77 children and adolescents aged 6 to 18 years and their mothers from a larger study of child and parental challenges related to managing the treatment regimen and correlates of adherence. SEM was used to test a model of maternal reports of the family social environment as exogenous variables with mealtime rituals and child hemoglobin A1c levels as endogenous variables. The bootstrapping procedure and the Bollen-Stine bootstrapped χ test were then applied to test the stability and appropriateness of this model. RESULTS: Using SEM, the final model had robust goodness-of-fit indicators and revealed that the family social environment characteristics of control and cohesion were related to mealtime rituals, which was associated with hemoglobin A1c levels. The bootstrapping procedure indicated that the parameter estimates were very stable, thus lending greater credence to the model. The Bollen-Stine approach provided evidence that the entire hypothesized model was not significantly different from the model derived from bootstrapping. CONCLUSIONS: This investigation offers a practical demonstration of the bootstrapping technique and the Bollen-Stine approach for testing models on smaller samples in SEM. These techniques may be used with relatively small samples, which are common in developmental/behavioral pediatrics and pediatric psychology research.