Objective: To challenge the application of an unqualified social learning model to the study of spanking, positing instead a developmental-contextual model in which the effects of spanking depend on the meaning children ascribe to spanking.
Design: Population-based survey data from 1112 children aged 4 to 11 years in the National Survey of Families and Households. Controlled for several family and child factors including children's baseline aggression.
Main outcome measures: Schoolyard fights and antisocial scores on the Behavior Problems Index at the 5-year follow-up.
Results: Structural equation modeling yielded main effects (P < or = .05, change in chi 2) of children's age and race; spanking predicted fewer fights for children aged 4 to 7 years and for children who are black and more fights for children aged 8 to 11 years and for children who are white. Regression analyses within subgroups yielded no evidence that spanking fostered aggression in children younger than 6 years and supported claims of increased aggression for only 1 subgroup: 8- to 11-year-old white boys in single-mother families (P < or = .05, F test).
Conclusions: For most children, claims that spanking teaches aggression seem unfounded. Other preventive effects and harmful effects of spanking may occur depending on the child and the family context. Further efforts to identify moderators of the effects of spanking on children's adjustment are necessary.