Using multi-wave data from a high-risk homeless and matched housed sample, the study tested: (1) A multi-level model of family violence, consisting of a general dyad-level factor and subordinate factors representing more specific aspects of dyad violence, and (2) A developmental-transactional model of distal and proximal processes linking parent-adolescent dyad violence to subsequent romantic partner dyad violence, the Risk Amplification Model (RAM). Measurement models confirmed a multi-level structure of violence across parent-adolescent and partner relationship contexts. Results failed to support the RAM model; however, strong relationships were shown between baseline adolescent anti-sociality and proximal and outcome variables. Overall, findings validated the multi-level model of violence and suggested the importance of controlling for anti-sociality in intergenerational violence transmission research. |