The ties that bind II: Reunification for siblings in out-of-home care using a statistical technique for examining non-independent observations

The present study was undertaken to incorporate sibling information and account for sibling bias when modeling longitudinal child welfare outcomes using a generalized linear model such as logistic regression. The paper is divided into two parts. The first part examines recent studies that have included sibling data in their multivariate models, and references statistical literature that underscores a method-generalized estimating equations (GEE)-that accounts for non-independent, cluster-correlated observations in analyses where logistic regression might otherwise be considered. The second part of the paper applies this technique to examine the likelihood of reunification from foster care for a statewide sample of siblings who entered care for the first time in 2000 (n=15,517). Results indicate that some sibling-specific factors (e.g., being placed together in care) strongly predict reunification, while another (i.e., number of siblings), surprisingly, does not. Implications for child welfare policy, practice, and future research are discussed.