|
Abstract
|
Two questions concerning the effect of physical abuse in early childhood on the child's development of aggressive behavior are the focus of this article. The first is whether abuse per se has deleterious effects. In earlier studies, in which samples were nonrepresentative and family ecological factors (such as poverty, marital violence, and family instability) and child biological variables (such as early health problems and temperament) were ignored, findings have been ambiguous. Results from a prospective study of a representative sample of 309 children indicated that physical abuse is indeed a risk factor for later aggressive behavior even when the other ecological and biological factors are known. The second question concerns the processes by which antisocial development occurs in abused children. Abused children tended to acquire deviant patterns of processing social information, and these may mediate the development of aggressive behavior.
VioLit summary:
OBJECTIVE:
The purpose of this study by Dodge et al. was to examine two questions concerning the relationship between early childhood physical abuse and later development of aggressive behavior. Two questions guided the research: l) whether abuse per se has deleterious effects, and 2) what process is involved when abused children develop antisocial behavior. The authors' perspective was learning theory-based.
METHODOLOGY:
A quasi-experimental, longitudinal, prospective research design was employed using three different kinds of primary data: in-depth interviewing, coded responses to childrens' viewing of videorecorded and cartoon stimuli, and data from teachers', peers', and researcher observations. A representative sample of 309 four-year-old children pre-registering for kindergarten in 1987 were initially selected. Three geographical regions were represented: Nashville, Tennessee (n=103 midsize urban community, with l/4 of the selected sample living in federally subsidized housing); Knoxville, Tennessee (n=100 Appalachian rural and small urban, mixed socioeconomic status), and Bloomington, Indiana (n=106 small city and semi-rural, with much of the sample from working class backgrounds). At pre-registration, parents were solicited, either in person at the school or by mail, at random to become involved in a study of child development. Participation rate was 70% after the contact, and study participants were then visited at home and paid $20. The demographics of the sample are as follows: 53% male, 47% female; 83% white, 16% African-American, 1% other. The parents of 18% of the sample were not married at the time of the child's birth, and 29% of the children lived with single parents. The mean age of parents when the children were born was 27 for mothers and 29 for fathers; average education was 13 years for fathers and 12 for mothers. Physical harm and related variables were measured through in-depth interviews with mothers who were asked questions about the child's early life and misbehavior, disciplining, and possible harm from an adult for both the first year of life and the last 12 months. The interviewer then rated the likelihood of physical harm. Forty-six children (15%) were classified in the likelihood of physical harm as probable or higher. A similar procedure was used to rate exposure to adult physical conflict. Standardized questions to assess family marital status and SES at the child's birth in addition to medical questions about the pregnancy and birth. Each mother also completed a restrospective version of the Infant Characteristics Questionnaire rating child temperament in infancy. Children's social information processing was assessed using videotapes and cartoons, organized into 24 vignettes, and presented to the child who was asked questions about the story involving the child's imagining he/she was the protagonist in each story. Recall of the story was measured and coded according to relevancy (0-2 with 2 fully relevant). Attribution of hostile intent, response accessing tendencies, and response evaluation tendencies were obtained from qualitative data and then coded. Assessment of the child's aggressive behavior was measured six months (plus or minus two months) following the mother and child interviews using teacher ratings, peer nominations, and direct observation. Information from the teacher was obtained from the Child Behavior Checklist, and the aggression scale score was computed. Peer information came from interviews with school peers who nominated up to three children they perceived as starting fights, getting angry, and being mean toward others (definition of aggression). These scores were tallied as the standardized number of nominations received for each item and then averaged across items for one peer aggression score for each child. Trained observers observed each child for 12, 5 minute periods in which observers measured the occurrence of aggressive acts. Univariate statistics, ANOVA, ANCOVA, and multiple regression were used to analyze the data.
FINDINGS/DISCUSSION:
Children who had been physically harmed in early life were significantly more aggressive toward peers than those who were not harmed (p=<.008). The family ecological variables of family SES, single parent, parents ever divorced, and observed marital or mother-boyfriend violence all significantly predicted harm to the child while none of the child/mother biological variables did. When all of these variables plus child's gender were controlled statistically, the multivariate main effect of harm was still statistically significant (p<.024) indicating that harm is predictive of later child aggressive behavior above and beyond the contribution of family ecology or child biological characteristics. Harmed children were found to be significantly less attentive to relevant social cues, more biased toward attributing hostile intent, and less likely to generate competent solutions to interpresonal problems (p<.043 for the model). The seven social information processing variables significantly predicted later aggression as it was measured by teachers (p=<.017), peers (p=<.01), and observation (p=<.001). These variables were encoding of relevant cues, hostile attributional bias, number of responses generated to each social problem, proportion of aggressive responses, proportion of competent responses, positive evaluation of the outcomes of aggressing, and positive evaluation of the outcomes of acting competently. Stepwise regression to examine the mechanism by which harm affects violence shows that the seven processing variables continued to predict later aggression even with harm in the equation. This followed for all three measures: teacher-rated (p=<.043), peer-rated (p=<.035), and observed (p=<.001).
AUTHORS' RECOMMENDATIONS:
The authors recommended further study in this area with more refined measures, particularly the question of why some children follow a path of hostile attributions and aggression and other children alternative paths (such as self-blame and depression). In general, the authors concluded that there are many more dimensions to the effects of abuse on children and that there is a need to keep striving for better measures.
EVALUATION:
The relationship between abuse and later aggressiveness is a continuing theoretical dilemma because of the complexity of determining causation and the continued presence of many confounding influences. The authors' attempt at prospective design is a step in the right direction. Decreasing the potential of temporal confusion through time lagged measures is required to explore an issue such as this. The use of multiple indicators of aggression, particularly from such different sources (peers, teachers, and observation) strengthens faith in the findings. However, and as the authors point out, there are some areas of weakness as well. Measurements of independent variables were questionable. Maternal report of abuse and subsequent interviewer coding of likelihood of abuse is not the best measure. Despite the claims made about inaccuracies in official statistics on abuse, these statistics would be better indicators not only because they do not rely on memory but also because reports of abuse come not just from children and parents but from other sources--neighbors, teachers, and doctors. Additionally, some measure of correspondence between coders would have been useful. Although, as the authors state, there are other mediating factors in the cycle that were not measured, this article and its complexity reveal the nature of the research question and, for what it is able to accomplish, adds to the ever-growing knowledge on the subject. (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 - Tennessee
KW - Indiana
KW - Child Aggression
KW - Aggression Causes
KW - Child Development
KW - Child Abuse Effects
KW - Social Learning
KW - Child Abuse Victim
KW - Child Physical Abuse Victim
KW - Child Physical Abuse Effects
KW - Domestic Violence Effects
KW - Domestic Violence Victim
KW - Child Victim
KW - Early Childhood
KW - Late Childhood
KW - Child Behavior
KW - Child Antisocial Behavior
KW - Behavior Causes
KW - Child Abuse-Aggression Link
Language: en |