|
Abstract
|
Studies of juvenile homicide have been limited to small samples and have arrived at widely varying clinical formulations. The present study of 72 juvenile murderers uses a three-group typology based on the circumstances of the offense. Group differences in prior adjustment suggest distinct developmental pathways to adolescent violence. Implications for research and for legal decision-making are discussed.
VioLit summary:
OBJECTIVE:
The purpose of this study by Cornell et al. was to investigate a typology based on characteristics of juvenile homicide defendants and their offenses committed. The typology had advocated classification of juveniles according to three distinctions: psychotic--juveniles who had psychotic symptoms, conflict--juveniles who engaged in interpersonal conflict, and crime--juveniles who committed homicide during a crime.
METHODOLOGY:
A quasi-experimental, cross-sectional design was employed using case records of the Michigan Center for Forensic Psychiatry for the years 1977 to mid 1985. The Juvenile homicide sample included 72 adolescents aged 12 to 18. 88 cases were initially identified, but 16 cases were eliminated due to missing information or inability to locate case record. A control group of 35 adolescents, 15-18 years of age, who were charged with larceny and referred for evaluation were selected from case records. There was no discussion of selection techniques for this group. The control group was significantly older than the homicide group and was also significantly less nonwhite. Two-thirds of the evaluations were conducted by doctoral level clinical psychologists, and one-third were conducted by psychiatrists. All were experienced members of the Forensic Center staff and underwent training for reviewing records. Records for each juvenile defendant generally included police and other investigative reports, interviews with family, witnesses, and associates of both victim and defendant, background information regarding the defendant, Forensic Center reports concerning the defendants competency to stand trial, criminal responsibility, juvenile status, mental health, and reports of evaluations by private practitioners working at the Center. Data coding was done in three separate stages. Stage one reduced the voluminous amount of information in case files to a manageable quantity organized in a uniform format and five page abstracts organized by topics were used to identify five variables describing the circumstances of offenses: adolescent-victim relationship, intoxication, weapon used, presence of an accomplice, and post homicide behavior. Coder agreement on the sample of 15 cases averaged 91% for 61 variables employed in this study. One coder was blind to the study hypotheses and coded approximately two-thirds of the cases. Stage two coded information according to 52 background variables clustered into the following categories: family dysfunction, school adjustment, childhood problems, violence history, delinquent behavior, substance abuse, psychiatric problems, and stressful life events. Discrete categories for each variable were assigned whole number weights according to the significance of the problem. Conceptually related variables were clustered and weights were summed into composite scores to compensate for problems resulting from analysis of low frequency items, and to improve scale reliability. This procedure was adopted to provide a concise summary of the juvenile's life history. Each composite variable was evaluated for each item's contribution to the total score, and seven of the fifty two items were deleted because of low correlations. Alphas were calculated for the 45 variables, and ranged from .36 to .79. In stage three of data coding, two authors independently classified the 72 homicide cases into the three subgroups: psychotic, conflict, or crime. Coders were limited to information concerning the offense, and case background. Each coder classified the 72 cases with 89% agreement (Kappa=.80).
FINDINGS/DISCUSSION:
Of the 72 juvenile homicide cases, the coders classified 5 psychotic cases (7%); 30 cases of interpersonal-conflict (42%); and 37 cases of crime (51%). Group differences on the eight composite variables was the primary analyses. The three homicide subgroups differed significantly from the larceny control group, as established by a multivariate analysis of variance (p<.001). One way (ANOVA) on each of the eight composited variables was done. Each ANOVA compared: larceny vs homicide, psychotic vs non-psychotic, and crime vs conflict. Hypotheses for the first series of planned comparisons were two-tailed, and directional hypotheses for the second series were formulated. The three series of planned comparison tests looked at 24 hypotheses, ten of which resulted in statistically significant findings. The adolescent control group scored significantly higher than the homicide group on school adjustment, childhood symptoms, criminal activity (p=<.001), and psychiatric history. Within the homicide group, the psychotic cases were significantly higher than the non-psychotic cases on psychiatric history (p=<.001), but lower on criminal activity. For the non-psychotic homicide cases, the crime subgroup was significantly higher than the interpersonal conflict subgroup on school adjustment (p=<.001), criminal activity, and substance abuse (p=<.001), but lower on stressful life events prior to the offense(p=<.001). When controlling for age and race by analyses of covariance, there was no change in the pattern of significant results. Two discriminant function analyses were applied to demonstrate the lack of overlap among groups, and were conducted by the Michigan Interactive Data Analysis System Discriminant program. The first analysis classified subjects into homicide or larceny control group. Using the maximum chance criterion reported by Huberty, the 78% classification accuracy were significantly better than the 67.3% accuracy expected by chance. The second analysis was limited to classifying subjects in the crime and control groups. The 82% accuracy were significantly better than the 37 of 67 cases (55.2%) which were classified using the maximum chance criterion, (p<.001).
AUTHORS' RECOMMENDATIONS:
According to the authors, the existence of distinct subtypes of homicidal adolescents would have direct implications for research on the etiology of adolescent violence, and would have special relevance for research on the clinical prediction of violent behavior. It was argued that the research on developmental differences among violent juveniles may distinguish developmental pathways to violent behavior. Research related to the prediction of violent behavior, the author claimed, may be impaired by failing to differentiate the motivational circumstances of violent behavior. This proposed typology could have implications for future treatment and disposition of juvenile homicide defendants by changing the waiver criteria among juvenile court systems in the consideration of the juvenile's psychological adjustment and amenability to treatment.
EVALUATION:
This article seeks to explore the complexity of adolescent homicide by not taking commonalities between different types for granted. The finding that very few of these adolescents were classified as psychotic has major importance for how we view juveniles who commit homicide; it is more difficult to segment these offenders into a category of "sick" individuals; therefore, the implication is that treatment is more useful and needed. The sample was large (72), though it is only potentially generalizable to a specific area of the country and to those cases referred to the Michigan Center for Forensic Psychiatry. Additionally, the potential for differential and subjective analysis (despite training) in creating the initial records may threaten internal validity, but the degree to which this is the case is not available for analysis. The contribution of this piece is the beginning of a more deeper and complex understanding of juvenile homicide not as a singular phenomenon but, potentially, as several contained within one. (CSPV Abstract - Copyright © 1992-2007 by the Center for the Study and Prevention of Violence, Institute of Behavioral Science, Regents of the University of Colorado)
KW - Michigan
KW - Homicide Causes
KW - Homicide Offender
KW - Juvenile Violence
KW - Juvenile Offender
KW - Juvenile Development
KW - Youth Development
KW - Developmental Pathway
KW - 1970s
KW - 1980s
KW - Offense Characteristics
KW - Offender Classification
KW - Offender Typology
KW - Offender Characteristics
KW - Violence Causes
KW - Mental Illness
KW - Mentally Ill Juvenile
KW - Mentally Ill Offender
KW - Psychological Factors
KW - Crime Effects
KW - Crime Causes
KW - Juvenile Crime
KW - Conflict
Language: en |