LONGITUDINAL STUDIES

LONGITUDINAL STUDIES

Although literally hundreds of studies have shown some psychological effects of media violence, most of those have been short term and in the laboratory, often using the methodology of showing participants a film and subsequently measuring their behavior or attitudes in some way. Although these findings are important, they do not directly address the long-term cumulative effects of watching hundreds of hours of violent television as recreation throughout one’s childhood. There have been a few studies that have addressed this issue, most notably those by Rowell Huesmann and Leonard Eron and their colleagues.

Longitudinal studies over ten years in the United States and Finland provided the first evidence of a causal relationship between real-world viewing of TV violence through childhood and violent behavior as a child and a young adult (Eron & Huesmann, 1984; Eron, Huesmann, Lefkowitz, & Walder, 1972; Huesmann & Eron, 1986; Lefkowitz, et al., 1977;

Violence: Watching All That Mayhem Really Matters 282

PitkanenPulkkinen, 1981). Through careful design and control of other variables, Eron et al. (1972) and Lefkowitz et al. (1977) concluded that they could rule out other plausible third variables, such as dispositional violence, as being the cause of both violent TV viewing and violent behavior. Huesmann and Eron’s research has shown that the amount of television watched at age 8 is one of the best predictors of criminal behavior at age 30, even after controlling for individual aggressiveness, IQ, and socioeconomic status. Those who were heavy TV viewers in preadolescence were also more likely to more severely punish their own children many years later than were those who watched less TV during the critical years of ages 8 to 12. Watching more TV in adolescence did not seem to matter.

In a 3-year longitudinal study, Huesmann, Lagerspetz, and Eron (1984) further explored the role of several intervening variables on the relationship between viewing media violence and violent behavior in U.S. and Finnish children in elementary school. The study collected a mass of data between 1977 and 1980 from the children, their parents, the children’s peers, and the school. Data gathered included measures of TV viewing, attitudes, behaviors, ratings of self and others, and family demographics. A few of the highlights are presented here.

As had been found in many other studies, there was a positive correlation between violent TV viewing and peer-rated aggression, which was stronger for boys than for girls and for U.S. people than for Finns. The overall level of violent behavior was also higher in the U.S. children. One of the most striking results for both samples was the strong correlation of violent behavior and self-rated identification with the violent TV model, especially for boys, The best predictor of later violent behavior was the interactive product of violent viewing and identification with the violent character. There was no evidence that violent TV affects only those children naturally more predisposed to violent behavior or that children who fantasize more were affected any differently. Neither were there particular effects of level of parental violent behavior or parents’ TV viewing on children’s violence.

The most recent longitudinal research presents the strongest evidence yet that exposure to media violence from ages 6 to 10 predicts violent behavior in young men and women in early adulthood (Huesmann, et al., 2003). Following up on children first tested in 1977–1978 (Huesmann & Eron, 1986), all of the same persons who could be found (about 60% of the original sample) were tested as young adults from 1992 to 1995. Data were obtained on their adult TV violence viewing and their adult aggressive behavior, the latter obtained from self-reports, spouse/significant other reports, and archival crime and traffic-violation data. Results showed that childhood TV violence viewing was a significant predictor of adult aggressive behavior. These effects were strongest if there was strong viewer identification with the aggressive model and if the perceived realism of the

283 A Cognitive Psychology of Mass Communication

violence was high. Unlike the earlier longitudinal studies, the 2003 study found these predictive effects for both men and women. The effects persisted even when the factors of socioeconomic status and intellectual ability were controlled. These results were replicated by another 17-year longitudinal study of 700 children (Johnson, Cohen, Smailes, Kasen, & Brook, 2002).

Although the evidence from longitudinal and other studies has consistently supported the existence of significant negative effects of viewing media violence, studies that look at multiple contributors to societal violence typically estimate that media account for 10% to 15% of the variance in the dependent measure (Perse, 2001; Sparks & Sparks, 2002). Some critics tend to dismiss the media component in the face of the 85% to 90% of the variance attributable to other factors. In fact, however, this 10% to 15% range is neither surprising nor is it insignificant. In any complex social behavior, there are naturally going to be multiple causes. In the case of interpersonal violence, there are numerous known strong causes such as parental teaching, poverty, drugs, gangs, absent parents, easy access to weapons, and decline in family values, all of which clearly contribute greatly to this problem. Thus, it would be surprising, indeed completely incredible, to expect media to account for half of the societal violence! Its importance is further demonstrated when we realize that media may be one of the easier causes of societal violence to change. There are also some good reasons to think the contribution of media to societal violence may be underestimated by these studies (Perse, 2001).

The clearest conclusion about the mass of research on the effects of media violence requires taking a convergent validation approach. Any study taken by itself is subject to some criticism in terms of methodology, interpretation, or ecological validity. When looking at the big picture, however, the weight of the evidence clearly falls on the side of violent media having several negative behavioral and attitudinal effects, especially modeling and desensitization. These effects typically are not uniform and frequently are moderated by other variables, but they do seem to be causal in nature. The “no effects” or “no clear picture yet” conclusion is simply not tenable scientifically, although it is surprisingly the view that is increasingly presented in popular news reporting.

RESEARCH VERSUS THE PUBLIC PERCEPTION Bushman and Anderson (2001) conducted a meta-analysis of over 200

studies of the effects of media violence, collectively involving over 43,000 participants. They concluded that the evidence for negative effects of media violence is strong and its strength is increasing, especially in the post-1990 research. Overall, effects are larger for experimental (rs=.25) than

Violence: Watching All That Mayhem Really Matters 284

correlational studies (rs=.10–.15), as one would naturally expect, but the trend is the same. Interestingly enough, however, during this same period when the research evidence on the negative effects of media violence was becoming stronger, the popular news reporting of the same research erroneously suggested that these effects were becoming weaker.

What are some possible reasons for this serious deviation from reality? Bushman and Anderson suggest three. One, the media industry has a considerable vested interest in denying a strong link, since they make enormous profits from selling violence. Television networks, movie studios, cable channels, and sometimes print media outlets as well, are part of the same huge conglomerates with considerable pressure to maintain or increase their audience size for their advertisers. A second reason suggested by Bushman and Anderson is what they call a “misapplied fairness doctrine.” One of media’s strongest abiding principles is fairness—presenting all sides of an issue. This causes media to look very hard for an opposing view, e.g., pro- vs. con-opinions on the negative effects of media violence. The resulting presentation of both sides of this issue leaves the reader/viewer with the mistaken impression that research opinion is more or less evenly divided on the issue. This scrupulous attempt to be fair ends up obscuring the fact that, although there are two sides of opinion on the issue (violent media do or do not cause violent behavior), the overwhelming majority of scientific opinion comes down on the negative effects side. Finally, Bushman and Anderson argue that the research community has often failed to advocate its position with strength and clarity. Researchers by training are very cautious and conservative about overgeneralizing the effects of their research, and most have no training and little or no experience in talking to the press.