|
Abstract
|
This article provides a meta-analytic review of the experimental effects of media violence on viewers' aggression in unstructured social interaction. In the reviewed experiments, children or adolescents were exposed to violent or control presentations and their postexposure behavior was coded for aggression during spontaneous social interaction. Exposure to media violence significantly enhanced viewers' aggressive behavior when the findings were aggregated across studies, but the effect was not uniform across investigations. Only suggestive evidence was obtained concerning moderators of the effect: Marginally stronger relations were obtained in those studies using a cross-section of the normal population of children (vs. emotionally disturbed children) and in those studies conducted in laboratory settings (vs. other contexts).
VioLit summary:
OBJECTIVE:
The goal of this article by Wood et al. was to examine the effects of media violence on aggression levels of children and adolescents.
METHODOLOGY:
A quasi-experimental method was employed for this study. Meta-analysis was conducted on experiments published that assessed unconstrained personal aggression. 23 research reports were identified from a search of citations in earlier reviews and the databases of PsychInfo, Eric, and Dissertation Abstracts International. Studies were selected that used random assignment, had dependent variables that consisted of aggressive acts directed toward persons or inanimate objects, and aggression was assessed through a direct observation of behavior by raters. Variables that were coded were 1) whether the study was published or not, 2) sex composition of the subject sample, 3) whether the subjects were initially angered or frustrated, 4) control condition, 5) type of subject (normal, juvenile delinquent, emotionally disturbed), 6) definition of aggression (interpreted or interpreted plus objective), 7) setting for exposure, 8) setting for observation of behavior, 9) age of subjects, 10) nature of interaction partners during assessment of aggression, 11) date of publication, 12) adequacy of the experimental design and data analysis to handle dependencies in responses of independent subjects, and 13) length of media presentation (in seconds). The direction of the effect was coded to indicate whether the experimental group was more aggressive; the control group was more aggressive, or they were equivalent. The effect size was calculated when adequate data were available. The magnitude of the effect was calculated in relation to their associated means and standard deviations. Vote counting, D-stat, OLS regression, and homogeneity statistic (Q) were used to analyze the data.
FINDINGS/DISCUSSION:
For 16 of the study outcomes, greater aggression was found in the experimental groups. In 7, greater aggression was found in the control groups. The result was said to approach significance (p<.10). Of 19 studies in which only neutral media presentations were used as control conditions, 13 found greater aggression in the violent-film condition, and 6 found greater aggression in the control condition. This was nonsignificant. The 12 studies for which exact effect sizes could be calculated showed a flat distribution of effect sizes and no skewness in either direction. In mean-weighted analyses of effect-size calculations, a significant increase was found in aggression after exposure to violent media. The findings were also found to be significantly heterogeneous which means that the weighted effect sizes were sufficiently different from each other and, as such, could safely be said to come from different populations. The outcomes for laboratory settings were found to be somewhat larger than for field settings (p<.10). These estimates were found to be heterogeneous with <.10 level of significance. The studies using emotionally disturbed subjects revealed essentially no effect. Slightly larger effects were found with normal subjects and juvenile-delinquent adolescents. The difference across means was marginally significant (p<.10). Estimates within these groupings were found to be homogeneous. The effect was significantly higher with better designed and analyzed studies. Other variables were not significant, and OLS regression revealed that neither the duration of the media presentations nor date of publication moderated the impact of violent media on viewers' aggressiveness. It was argued in the end that this study, in concert with other studies, suggests that aggression is multiply determined and that media is not the only factor.
EVALUATION:
This study represents an important methodological achievement in social science--analysis of studies of the same phenomenon. Through meta-analysis, we can come to summary conclusions about a variety of studies. This study takes great pains to account for varieties of methodology and to test the studies' findings on common ground with appropriate controls. 23 studies is not a large number, but the authors were limited to what was available. In subdivisions of studies, such as type of subjects, the numbers of studies become small. Additionally, conclusions based on findings significant at <.10 should be taken cautiously. Overall, this study is an important beginning to making collective sense out of studies of aggression effects. Given the nature of the study and its dependence on studies that have been done, this type of study must be replicated on a regular basis to increase the base of studies analyzed to be even more grounded in the conclusion that violent media can increase aggression levels. (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 - Meta-Analysis
KW - Media Violence Effects
KW - Aggression Causes
KW - Juvenile Aggression
KW - Child Aggression
KW - Social Interaction
KW - Exposure to Violence
Language: en |