Structural and Cultural Determinants of Child Homicide: A Cross-National Analysis

Article excerpt

Models of structural characteristics that may influence national infant and child homicide victim rates are derived from prior research. Expected effects of structural characteristics from a "social control" perspective are compared with expected effects from a "guardians, suitable target" perspective. Gartner's (1991) claim that structural factors influence victim rates only in nations with low social insurance expenditures is also evaluated. Statistical analysis of three infant and child age groups with homicide rates from 1965-1969, 1970-1974 ... 1985-1988 fails to support the claim that high and low social insurance expenditure strata differ. Further, no independent effects of Gartner's (1991) three measures of family structure are found. Indicators of family stress/resources, female status, the culture of violence, and a proxy for unmeasured variables and measurement error all contribute to produce high levels of explained variance in each age group.

The systematic comparative study of national child homicide victim rates was absent from sociological literature before the appearance of Fiala and LaFree's (1988) pioneering work, and subsequent articles by Gartner (1990,1991). These works empirically assess potential structural and cultural determinants of cross-national differences in infant and older child homicide victim rates. This article organizes, expands, and then tests Gartner's basic (1990) and revised (1991) conceptual scheme.

Gartner focuses on two theoretical perspectives. The first considers structural factors that affect adult-child relationships within the nuclear family or kinship group. The character of these attachments will affect the child's risk of becoming a victim of adult violence from family members because they increase or decrease formal and informal social controls that affect the risk of homicide. A nation's "culture of violence" is also seen as influencing the impact of social controls that affect potential murderers (Williams & Flewelling, 1988).

The second perspective considers structural conditions that may mediate the role of "guardians" and other "situational factors" as a means of social control. To this conceptual scheme we add a measure that should correct for unmeasured causal variables, a lack of comparability, and unreliable measurement of independent and dependent variables.

Theories of homicide are largely based on adult victims, whether discussing covariates (Land, McCall, & Cohen, 1990) within the United States or cross-national studies (see Gartner, 1990, and Cutright & Briggs, 1993, for literature review). The determinants of child and adult homicide rates, however, may well be different Perhaps the major contribution of Fiala and LaFree (1988) is their emphasis on relationships among family members. Rather than relying on previous theoretical arguments about the causes of adult homicide rates, Fiala and LaFree viewed child homicide as an extreme form of child abuse and directed their literature review to that topic. This theoretical redirection is of special importance to this analysis.

Several characteristics that differentiate child from adult homicide are apparent in U.S. data. For example, just over 3% of 1989 homicide victims ages 15 or older (17,954) had an unreported proximate cause of death-e.g., firearms, cutting/stabbing (U.S. Department of Justice, 1989, p. 11). In contrast, 26% of the infant victims (254), 11% of the victims ages 1- 4 (340), and slightly over 6% of the 5-14 year olds had an unreported method. Medical examiners are more diligent in reporting the method of killing among older than among younger victims.

A second difference between child and adult homicide victims is their gender. According to the U. S. Department of Justice (1989, p. 10), about 52% of victims under age 10 are male- the percentage we would expect due to the higher male than female sex ratio at birth, and among young children. …