Infant and adult homicide: Incompatibility of predictive models

The utility of demographic and structural variables in examining child fatalities is explored using data from the Uniform Crime Reports, which include all known homicide cases in the state of Texas from 1984 through 1994. The authors' analysis shows that the predictive power of these variables for adult homicide is substantially reduced when predicting child homicide. Social Exchange Theory provides a conceptual framework from which to analyze these types of cases by suggesting conditions under which infants may be lethally injured, usually by a parent or caretaker. This inclination is examined, explained, justified, and treated primarily according to physiological and psychological correlates.