This study investigated the effects of video game play on aggression. Using the General Aggression Model, as applied to video games by Anderson and Bushman,  this study measured physiological arousal, state hostility, and how aggressively participants would respond to three hypothetical scenarios. In addition, this study measured each of these variables multiple times to gauge how aggression would change with increased video game play. Results showed a significant increase from baseline in hostility and aggression (based on two of the three story stems), which is consistent with the General Aggression Model. This study adds to the existing literature on video games and aggression by showing that increased play of a violent first person shooter video game can significantly increase aggression from baseline.
Barlett and company investigated on the effects of different game controllers and length of game play aggression. What I like about this study is the emphasis on using a single video and experimentally controlling video game factors that may affect results. Some examples of video game factors are the amount of violence, character development and story development. The results have found that play is associated with state aggression, hostility, physiological arousal. In combination with playing a light gun, the effect is stronger than playing with a traditional controller. However, play time and frustration was not found to be significant.
In this study, the types of game controllers (light gun vs. controller) will tell us whether there are differences in aggression. One reason is the weapons effect, where the mere presence of a weapon increases aggressive behaviours.
The other factor being investigated is the amount of time a player spends on a violent video game in a session, not in the long term. IMO, Barlett et al. did not wrote much about it. What is being explained is simply a snowball effect: exposure to aggression or committing an aggressive act leads to increased aggressive cognition, emotions and physiological arousal. This could lead to increases in aggressive behaviours. So the logic is more exposure to violence equals more aggression. Well it seems there are two theories being tested: (1) there will be an increase, but it will not increase continuously during play. (2) There will be an increase and it would continue to increase as long as play continues.
Frustration is included within the study. It’s defined by the feelings of not being able to achieve goals.
Barlett et al. provided some info about Craig Anderson’s (2004) meta-analytic study and his criteria of “best-fitting” video game studies. What’s “best-fitting” to include in his meta-analysis is that it used a pre-post design (good), clearly defined non-violent and violent video games (no brainer), and “produced evidence that the video games differed from one another to not contaminate the experimental conditions.” This last criterion is puzzling me and perhaps I’m reading it wrong, so it’s best fitting if the study used different video games? If we were to compare two non-violent games that are different and achieve different results… so does it mean that other factors within the experiment can influence results? I just don’t see how it works out.
- The longer you play, the more hostile and aroused you become.
- Playing with a more interactive controller is associated to higher increase in aggression than a standard controller.
Participants: 99 undergraduates (85 males, mean age= 19.42, SD=1.67). average game play hours per week: 11.52 (SD=10.01)
Materials: Time Crisis 3 is selected because it’s an FPS, it supports either standard controller or a light gun. This allows all participants to experience the same gaming environment, so minimizing any external confounds in the experiment. Finally it represents the latest generation of video games, so it applies to today.
Physiological arousal: a heart rate monitor that measures heart pulses from the index finger.
State Aggression: Nine story stems. Participants take a character’s point of view and how they would react, from a list of 5 answers (least violent to most violent), to stories of bad behaviours. For example, “He kicked your dog.” Answer one: call his home and say that his daughter kicked your dog. Answer two: kick his dog. Etc… Well the stories and answers are more refined than my example. But you get the idea. 3 categories of point of view of story stems: sports player, parent and judge.
Hostility: One item that measures state hostility on a 5-point likert scale. Just one!?
Frustration: One item on a 5-point likert scale. WTF?! Where the hell is your inter-item reliability?
Suspiciousness questionnaire: designed to verify whether participants figure out what the experiments is investigating. If you know what the experiment is about, you would be subject to expectation biases.
Participants are taken their heart pulses, complete pre-experiment questionnaire and three story stems. After, they are given a tutorial on the game and the controller type (either standard or light gun).
To manipulate frustration, they either play on level 1 or 2 to vary difficulty as they say. Play time last 15 minutes.
They are again taken their heart rates, complete the same questionnaires and three different story stems. Again, they are given a tutorial on the other type of controller, different from the first play. They continue their play with the other controller type for 15 minutes.
Finally, their heart rates are taken, with the questionnaires in addition to the demographics and suspiciousness questionnaires, and another set of three story stems.
Frustration: a repeated measures ANOVA was conducted and it was found that there were no significant differences in frustration across time. So it’s excluded in subsequent analyses.
A 3 (game time) X 2 (level) X 2 (controller type) repeated measures ANOVA was conducted for data analysis.
Results for physiological arousal: there is a main effect for game time, but there is no significant simple effect. So there’s an increase in physiological arousal, but it does not differ across time. An interaction effect is found between game time and controller type, the light gun condition produced higher physiological arousal than standard controller.
State aggression: Same thing, main effect for game time, no simple effect and an interaction effect with controller type. So an increase in aggression scores, regardless of time, from using light guns produced higher scores. Note: one category (judge) of the story stems was omitted from analyses due to response distortion.
Hostility: Same thing, main effect for game time, no simple effect and an interaction effect with controller type. So increase in hostility scores is attributed to game time, but not across time and light guns produced higher scores.
My first thought is that you can’t readily generalize this study to the Wii-zapper, especially if you make an aesthetic comparison. The Wii does not have the typical visual characteristics of a gun, while the Time Crisis gun do. You normally don’t see a white gun or it looks more like a nerf gun to me. So this call for another study on what characterizes a weapon effect. A silly wii-mote vs. a gun replica.
Taking a closer look at the results, I can’t say anything about physiological arousal. So I don’t know if they are really aroused. But I can say something about the other measures. State aggression, well given the middle score in the stories is three, the mean scores after playing Time Crisis 3 is around three. Nevertheless, the baseline mean score is below three. So we’re at the middle aggression range? But it does point the evidence that it does increase our state aggression. Between controller types, it just the same, except the light gun produced somewhat higher scores.
Across time, mean scores for state hostility (1.11 (baseline) vs. 1.44 (time 2) vs. 1.52 (time 3)) isn’t really that high, given the max number is 5. So yeah, we’re more hostile, well slightly I guess. Between controller type, a difference of .1, a significant difference nevertheless.
I have some issues on how they operationalized game time, it differs from my idea of game time. My idea is a single game session of varying lengths of time, say sessions of 15, 30 and 60 minutes. What they did was given participants 15 minutes of play time, then interrupt, switch controller type and resume play for another 15 minutes. Why did they interrupt the game time? Can they take another kind of heart rate measure that continuously monitor their heart rate? Does the interruptions and switching of game controller interfere participants’ play experience? Well it certainly does and what was probably meant to analyze the effects between controller type, but I believe it confounds the game time analyses. So, I see this experiment, not as a within-subject design, rather a between-subject design. Barlett et al. addressed this concern, but I got lost when they say it is possible to do this kind of experiment.
Several limitations were mentioned:
- Their measure for physiological arousal is not reliable.
- Did not control participants’ playing experience with the game used (Time Crisis 3) or whether the participants had previous experience with real guns.
- Did not include video game experience and performance as a control factor.
- The validity for assessing frustration and hostility in this study is a problem .
- Aggression measure has not been previously used, so validity and reliability is a problem. Although in previous studies, researchers had used similar measures.
Barlett, C. P., Harris, R. J., & Baldassaro, R. (2007). Longer you play, the more hostile you feel: Examination of first person shooter video games and aggression during video game play. Aggressive Behavior, 33(6), 486-497.
Anderson, C.A. (2004). An update on the effects of playing violent video games. Journal of Adolescence, 27, 113-122.