Objectives. This study compares two explanations for the effects of an employment-based anti-poverty intervention, the New Hope project, on parenting and child behavior. (1) Did the New Hope intervention directly affect child behavior and, if it did, is this the result of New Hope effects on earlier parenting practices? (2) Did the New Hope intervention directly affect parenting practices and, if it did, is this the result of New Hope effects on earlier child behavior?
Methods. Assessments were collected two and five years after parents were randomly assigned to participate in New Hope or in a control group. The two models were tested using Baron and Kenny's (1986) recommendations for regression to test mediation.
Results. The results support the second research question, but only for boys; neither model was supported for girls. New Hope led to increased positive behavior and reduced behavior problems for boys after two years; children's behavior accounted for experimental effects on parents' reports of their parenting behaviors after five years.
Conclusions. Employment-based interventions designed for low-income parents can affect children directly; expanding such programs may help improve children's life chances.