Predictors of Attrition in the Fragile Families and Child Wellbeing Study

Abstract

The Fragile Families and Child Wellbeing Study is a longitudinal study which oversamples nonmarital births (“Fragile Families”). Attrition in longitudinal studies is a problem which can affect both internal and external validity of the results. In order to identify which factors predict a participant’s attrition from the study, logistic regression methods were performed. Variables which serve as predictors for a mother’s odds attrition from the study are education, immigration status, and age at time of child’s birth; higher levels of education are a predictor of participation while being an immigrants and increased age are predictors of attrition. The predictors for a father’s attrition are education, immigration status, and relationship with biological mother at time of child’s birth; education and relationship are predictors of participation while being an immigrant is a predictor of attrition. These results are useful to provide insight into how researchers can anticipate attrition in longitudinal studies and how attrition can be accounted for and corrected.