ADOLESCENT AGGRESSION AND TELEVISION

Most criminal acts in the United States, at least most violent criminal acts, are committed by young males in late ado1escence.l There could be many reasons why this is so-biological, psychological, sociological, political, and economic. Empirical studies have uncovered relations between aggression and a number of variables from these classes. Some of these relations are causative, some concomitant or perhaps merely coincidental. One particularly striking finding has been that heightened aggression among adolescents is related to the violence of the television programs they watch.* However, the extent and the content of a child's television viewing are inevitably correlated with other potential causes of aggression. For example, most violent crimes are committed in lower-class, ghetto areas by individuals of limited IQ, who have dropped out of school, are unemployed, come from disorganized families, and in general have limited resources for coping with problems.' These are also the persons who spend much time watching television and who prefer violent Thus, on the basis of correlational data alone, it is difficult to attribute cause or effect to either of the variables in question. Causal relations can be best demonstrated by experimental manipulation or inferred from large-scale observational studies done over time using repeated observations on the same subjects. The studies by Berkowitz and his students with college-age subjects exemplify the former approach, and the studies we have done, following youngsters for periods up to 10 years until late adolescence, illustrate the latter approach. Berkowitz, for example, has demonstrated that exposure of university students to violent films increases the likelihood and magnitude of subsequent aggressive behavior, especially if the viewer is angered or frustrated prior to viewing the film.6 In our 10-year longitudinal study, we found that the best single predictor of how aggressive a young man would be at age 19 was the violence of television programs he had watched at age 8.' By use of a cross-lagged panel design and partial correlation to control for possible third variables, it was demonstrated that the most plausible hypothesis to explain this relation was that continuous viewing of television violence at the earlier age caused the aggressive behavior that was measured at the later age. FIGURE 1 describes the cross-lag correlational analyses of our 10-year study. These data have already been published, so we will not dwell on them at any length.' You see here the significant difference between the two cross correlations over a 10-year lag. The violence of television programs watched by boys at age 8 is more highly related to their aggression 10 years later than is the aggression of boys at age 8 to the violence of television they watched