A Longitudinal, Multilevel Analysis of Child Homicide Rates and Macro-Level State Characteristics: 1979-2007

Background and Purpose: About 2,000 U.S. children die annually from maltreatment, the majority of which are perpetrated by caregivers. Our nation's policies and programs have increasingly paid attention to maltreatment fatalities but we know little about what factors are associated with fatality rates, especially at the state level. Only one study has examined the principles of ecological theory (Bronfenbrenner, 1970) and macro-level factors as they relate to the maltreatment fatalities and confirmed different processes for infants and toddlers (Straus, 1987). The current study examines three decades of fatality rates using multi-level modeling and addresses: (1) What are the rates of maltreatment-related fatalities over time? and (2) What macro-level factors are associated with maltreatment-related fatality rates? This is the first study to examine child fatalities using longitudinal data and to use multi-level modeling.

Methods: This study uses state-level, longitudinal data from the Centers for Disease Control & Prevention, 1979-2007, to examine child homicide rates among children ages 0-9 and how these rates are related to state characteristics. Child homicide victims are predominantly killed by their caregivers; the homicide rate among young children has been determined to be an excellent proxy for maltreatment-related fatalities. The author used multilevel mixed-effects linear modeling to examine the relationship between child homicide and macro-level variables which have been used in previous research: violent crime, divorce, poverty, unemployment, presence of child fatality review team legislation, motor vehicle deaths, and state population. The analyses were conducted separately for children under the age of 1 and children ages 1-9.

Results: Descriptive statistics show that the child homicide rate among infants has increased between 1979-2007 and that there has been little change among victims ages 1-9. The results of the multi-level modeling indicate that the intercepts for child homicide rates vary by state and that state characteristics affect children of in different age groups differently. Higher rates of infant homicide (chi2 Wald=86.56, p<.001) were associated with the presence of child fatality review team legislation (z=2.89, p=.004), higher rates of violence crime (z=6.45, p<.001), lower rates of unemployment (z=-2.07, p=.04), and lower divorce rates (z=-2.81, p=.005). Homicide rates among older children (chi2 Wald=111.78, p<.001) were associated with higher rates of violent crime (z=6.48, p<.001) and motor vehicle deaths (z=3.13, p=.002). No other variables in this latter model were significant.

Conclusion and Recommendations: The findings show that despite years of concentrated efforts, the child homicide rate has not declined between 1979-2007. Further, not all macro-level factors were associated with fatalities. For example, poverty, which is often associated with non-fatal maltreatment, did not explain the rates of child homicide. In fact, there is an inverse relationship between unemployment and homicide rates among infants. Moreover, the presence of child fatality review legislation does not appear to lower the fatality rate. Child homicide rates among victims age 0-9 can best be explained as emerging from a macro context of violence and reckless behavior. The results speak to the importance of violence prevention efforts for families, communities, and states.