Models for Aggressive Behavior: The Attributes of Violent Characters in Popular Video Games

Communication Studies, Dec 2005 by Lachlan, Kenneth A, Smith, Stacy L, Tamborini, Ron

Liked and disliked characters were also identified using the N.T.V.S. definitions of "good" and "bad" characters. Good characters are those that act benevolently, help others, and/or are motivated to consider the needs of others before themselves. Bad characters, on the other hand, are those that act out of self-interest, accommodate their own needs, and have little regard for others. Neutral characters were those that possessed both good and bad characteristics or those that did not appear long enough in the game for their orientation toward others to be assessed. Because pro-social or liked characters may be more compelling behavioral models (Jose & Brewer, 1984), this variable was collapsed into two categories: liked (good) vs. disliked (bad neutral) in order to draw a distinction between more commonly occurring characters and those that may be especially liked because of these positive behaviors. The decision to collapse bad and neutral characters allowed for comparisons that specifically identified the attributes associated with those characters that may provide the strongest behavioral models.

Several violence-related variables were also assessed. The first is reason. Because aggression is a complex behavior that may be motivated by multiple goals, coders examined the presence or absence of seven specific reasons for each interaction: protection of life, protection of property, anger, retaliation, personal gain, mental instability, and other/unknown. Determinations of violence as "justified" or "unjustified" were established by evaluating the motives for aggression. At the analysis level, any violent interaction that was motivated by protection of life, protection of property, or retaliation was receded as "justified." Violent interactions motivated by other reasons (i.e., personal gain, mental instability, other) were recoded at the analysis level as "not justified," consistent with the criteria for justified motives found in the N.T.V.S.

The consequences of each violent interaction were also evaluated. Depicted harm refers to physical injury or incapacitation of the victim. Depicted harm was coded as none, mild, moderate, or extreme. This was based on the reactions of the victims, and deviated slightly from the N.T.V.S. definition to include the more extreme forms of violence and bloodshed found in video games. If the victim gave no indication of harm, this was coded as none. Mild referred to a small indication of harm (such as a grunt or bruising). Moderate referred to any violent act strong enough to result in bloodshed and or severe visible distress on the part of the victim. Extreme included large scale bloodshed or gore, disfiguring/severing injuries, and death.

Training and Reliability

A total of three undergraduates from Michigan State University coded all of the video game segments in the sample for violence. Prior to coding, the research assistants underwent four weeks of intensive training to master all of the definitions and coding procedures used in this study. During this training, the coders participated in a series of reliability tests (n = 3) designed to assess the consistency of their coding judgments. Using Potter and Levine-Donnerstein's (1999) reliability formula for multiple coders, the coefficients for each of the variables on the last test were as follows: type of act (.94), perpetrator type (.86), perpetrator age (.96), perpetrator sex (.92), perpetrator ethnicity (.96), perpetrator motive (.79), protect life (.79), protect property (.95), retaliation (.84), anger (.95), personal gain (.95), mental instability (.95), unknown (.92), and depicted harm (.80).