After 40+ years of research, one might think that debate about media violence effects would be over. An historical examination of the research reveals that debate concerning whether such exposure is a significant risk factor for aggressive and violent behavior should have been over years ago (Bushman & Anderson, 2001). Four types of media violence studies provide converging evidence of such effects: laboratory experiments, field experiments, cross-sectional correlation studies, and longitudinal studies (Anderson & Bushman, 2002a; Bushman & Huesmann, 2000). But the development of a new genre-electronic video games-reinvigorated the debate.

Two features of video games fuel renewed interest by researchers, public policy makers, and the general public. First, the active role required by video games is a double-edged sword. It helps educational video games be excellent teaching tools for motivational and learning process reasons. But, it also may make violent video games even more hazardous than violent television or cinema. Second, the arrival of a new generation of ultraviolent video games beginning in the early 1990s and continuing unabated to the present resulted in large numbers of children and youths actively participating in entertainment violence that went way beyond anything available to them on television or in movies. Recent video games reward players for killing innocent bystanders, police, and prostitutes, using a wide range of weapons including guns, knives, flame throwers, swords, baseball bats, cars, hands, and feet. Some include cut scenes (i.e., brief movie clips supposedly designed to move the story forward) of strippers. In some, the player assumes the role of hero, whereas in others the player is a criminal.

The new debate frequently generates more heat than light. Many criticisms are simply recycled myths from earlier media violence debates, myths that have been repeatedly debunked on theoretical and empirical grounds. Valid weaknesses have also been identified (and often corrected) by media violence researchers themselves. Although the violent video game literature is still relatively new and small, we have learned a lot about their effects and have successfully answered several key questions. So, what is myth and what do we know?

Myths and Facts

Myth 1. Violent video game research has yielded very mixed results. Facts: Some studies have yielded nonsignificant video game effects, just as some smoking studies failed to find a significant link to lung cancer. But when one combines all relevant empirical studies using meta-analytic techniques, five separate effects emerge with considerable consistency. Violent video games are significantly associated with: increased aggressive behavior, thoughts, and affect; increased physiological arousal; and decreased prosocial (helping) behavior. Average effect sizes for experimental studies (which help establish causality) and correlational studies (which allow examination of serious violent behavior) appear comparable (Anderson & Bushman, 2001).

Myth 2. The studies that find significant effects are the weakest methodologically. Facts: Methodologically stronger studies have yielded the largest effects (Anderson, in press). Thus, earlier effect size estimates -based on all video game studies- probably underestimate the actual effect sizes.

Myth 3. Laboratory experiments are irrelevant (trivial measures, demand characteristics, lack external validity). Facts: Arguments against laboratory experiments in behavioral sciences have been successfully debunked many times by numerous researchers over the years. Specific examinations of such issues in the aggression domain have consistently found evidence of high external validity. For example, variables known to influence real world aggression and violence have the same effects on laboratory measures of aggression (Anderson & Bushman, 1997).

Myth 4. Field experiments are irrelevant (aggression measures based either on direct imitation of video game behaviors (e.g., karate kicks) or are normal play behaviors. Facts: Some field experiments have used behaviors such as biting, pinching, hitting, pushing, and pulling hair, behaviors that were not modeled in the game. The fact that these aggressive behaviors occur in natural environments does not make them "normal" play behavior, but it does increase the face validity (and some would argue the external validity) of the measures.

Myth 5. Correlational studies are irrelevant. Facts: The overly simplistic mantra, "Correlation is not causation," is useful when teaching introductory students the risks in too-readily drawing causal conclusions from a simple empirical correlation between two measured variables. However, correlational studies are routinely used in modern science to test theories that are inherently causal. Whole scientific fields are based on correlational data (e.g., astronomy). Well conducted correlational studies provide opportunities for theory falsification. They allow examination of serious acts of aggression that would be unethical to study in experimental contexts. They allow for statistical controls of plausible alternative explanations.

Myth 6. There are no studies linking violent video game play to serious aggression. Facts: High levels of violent video game exposure have been linked to delinquency, fighting at school and during free play periods, and violent criminal behavior (e.g., self-reported assault, robbery).

More here:
Violent Video Games: Myths, Facts, and Unanswered Questions