Predictors of children in foster care being adopted: A classification tree analysis

Abstract

When returning home is not a viable option, adoption is the primary means by which youth in substitute care achieve permanency. Therefore, understanding the factors that impact adoption is critical to both evaluating and improving the child welfare system. Prior research has mostly studied “main effects” in regard to adoption rates and has not explored the ways in which youth and foster family variables might interact in complex ways to predict adoption. This study uses a classification tree analysis approach known as Optimal Data Analysis (ODA) to predict probability of adoption in 2003 using Adoption and Foster Care Analysis and Reporting System (AFCARS) data. 30,000 adopted youth and 30,000 non-adopted youth were randomly selected for participation in the study. Similar to previous studies, univariate analyses revealed that age, foster parent race/ethnicity, foster parent marital status, and number of previous placements all predicted probability of adoption. However, going further, several combinations of individual variables (multivariate ODA) improved prediction accuracy (e.g., age × structure of foster family × number of previous placements). Results also suggested that the impact of the state on adoption rates varied when essentially controlling for youth and family variables, supporting concerns regarding the crude use of AFCARS data when comparing states.