Purpose: The purpose of this study was to discover whether there were subgroups of urban maltreated children who were more or less at risk of adverse developmental outcomes, how these groups of maltreated children compared with non maltreated children from the same neighborhoods, and how these groupings related to stress reactivity as measured by cortisol levels.

Methods: This paper reports findings taken from a longitudinal study of abuse and neglect of urban young adolescents. The racially diverse (38% African American, 37% Latino, 12% white, 12% biracial) sample consisted of 303 maltreated and 150 comparison children who were 9-12 (X= 10.3) at their inclusion in the study. Children participate in a protocol that included a number of measures including the Self Image Questionnaire for Young Adolescents (SIQYA) (Petersen, et al., 19840, the Self Perception Profile for Adolescents (SPPA) (Harter, 1988), and a protocol to measure cortisol responses to stress. A modified Trier Social Stress Test (Kirschbaum, Pirke, & Hellhammer, 1993) was used with saliva samples taken before and after a stressful event to model a cortisol response curve for each child. Cluster analysis was used on two factors found in the SIQYA (positive and negative affect) to discover distinct groups of maltreated children. These groups were then used in an analysis of variance with the competence scales of the SPPA to determine if these groups were related to self image.

Results: Six distinct groupings of maltreated children were found on the SIQYA. The means of the groups were significantly different at the .01 level on each of the 6 competence scales of the SPAA. Post hoc comparisons indicated that one group was always larger than the others and one was consistently smaller than the others in all but 3 of the 36 comparisons. Interestingly when the comparison group was entered in the comparisons with the other groups, it ranked second, below the group with the largest means indicating that this group of maltreated children had the most positive self images. The eta correlations for the 6 competence scales ranged from .32 to .20. When the patterns of cortisol reactivity for the groups were examined, differences significant at the .01 level were found on patterns of reactivity with the maltreated group highest in self image different from all the other groups. These children with the highest competency scores showed a flattened cortisol reactivity pattern: they did not show a strong reaction to a stressful situation.

Implications for Research and Practice: The results of this study reinforce the knowledge that the reactions to child abuse and are individual and occur within a complex of factors inherent in the child and the environment that have differential effects on the way children react to the trauma of child abuse. Some children may develop coping patterns which buffer the impact of child abuse and other stressful events. Understanding what these patterns are and how they develop will lead to interventions which can better ameliorate the negative effects of child abuse.