Effects of Violent Video Games on Aggressive Behaviors Among Children: The Role of Anger and Trait Aggression in China

The current study examined the violent video game effect on aggressive behaviors and the potential role of anger and trait aggression among Chinese children (n = 248; M age = 6.09, SD = 0.21). Half of them were randomly assigned to play a violent or a neutral video game for 20 minutes. Results showed that brief exposure of children to a violent video game produced a small but signicant increases in aggressive behaviors. More specically, males exhibited more aggressive behaviors than females in the violent game condition. Moderation analysis suggested that the violent video game effect on aggression was not moderated by trait aggression, but this effect was mediated by increased anger (especially for males). These ndings indicated that males who were exposed to a violent video game might be regarded as the key group to reduce aggression by training to vent their anger. Results are interpreted within and support the framework of the General Aggression Model (GAM).


Introduction
Aggression refers to a type of behavior directed toward another individual that is carried out with the proximate intent to cause harm (Anderson, 2002). In particular, the perpetrator must ensure that the behavior would harm the target (Geen, 2001). In laboratory settings, arti cial measures of aggression are used for children due to ethical reasons. In the present study, aggressive behavior is operationally de ned as the level of hot sauce powder that children intentionally select for stranger in the picture.

Theories of aggression
To date, General Aggression Model (GAM), Self-Determination Theory (SDT) and Catalyst Model (CM) are common theoretical models to explain the causes and consequences of aggression. The GAM, a useful domain-speci c theoretical framework, considers the in uence of personal (i.e., trait aggression, attitudes, genetic predispositions, sex) and situational factors (e.g., games, lms, songs) on aggressive behavior through mediating variables of cognition, affect, and arousal. All three of these mediating variables (internal states) lead to a decision process, which causes aggressive behaviors . The SDT is a broad theory of human personality and motivation concerned with how the individual interacts with and depends on the social environment. The SDT outlines how these selfdetermined motivations in uence social and cognitive performance, and well-being (Ryan & Deci, 2017).
The CM that details that family community, school, media violence and peer violence may be related to increased aggression, but not cause such aggression (Elson & Ferguson, 2014). In the current article, we focus on one situational variable-exposure to a violent video game-and how it in uences internal state (e.g., anger) to increase the likelihood of aggression. Thus, we attempt to use the GAM as the theoretical framework to explain the violent video game effects on kindergartners.

Children and violent video games
Today, many young children are increasingly playing video games through their parents' smart-phones or tablets (Rideout, 2017). Exploring how violent content used in games for young children provides much-needed information about whether the violent video games that is connected to how young children behave. Further, such information can help improve future video games by highlighting potential gaps in game design as well as provide educators with references to help select high-quality video games for learning. Thus, an ever growing number of kindergartners are playing video games, making our topic of violent video game effects relevant and timely. Speci cally, longitudinal evidence suggests that violent video games exposure is an important independent predictor which contributes to aggression in children and adolescents (Teng, Nie, Guo, Zhang, Q., Liu, & Bushman, 2019; Tian, Gao, Wang, & Gao, 2020). As a child psychology researcher, we are more concerned about kindergartners who often use violent video games, track these kids who play video games, guide them to develop good game habits, and cooperate with parents to manage their gaming behavior reasonably.

Violent video games and aggressive behavior
So far, the relationship between violent video games and aggressive behavior has been debated. A group of researchers who's work indicates a positive game violence-aggression link (Anderson et al., 2017;Chang & Bushman, 2019;Greitemeyer, 2019). However, critics have failed to nd such causal link, and they argue that the conclusion of the above scholars is a publication bias (Ferguson, 2007). The critics contend that violent video game exposure can not promote aggression, or even the violent video game effect is null. Thus, critics state that some of game-aggression works are known to be false positives due to questionable researcher practices that are not replicated when reexamined ( Nonetheless, a few researchers are skeptical of the key hypothesis that anger causes aggression (Geen, 2001). Noticeably, repeated exposure to violent video games enhances hostile feeling and aggressive behavior (Allen, Anderson, & Bushman, 2018). Given that aggression-related knowledge rehearsal structures are primed by anger, aggressive behavior is likely to be energized by increased anger. In many cases, individuals spill out anger and aggressive behavior when he/she is frustrated (Breuer & Elson, 2017;Dollard, Doob, Miller, Mowrer, Sears, & Faris, 1939). As such, frustration may induce anger, which, in turn, may lead to aggressive behavior outcomes. Thus, anger may become a mediator of aggressive behavior under the condition of violent video games.

The present study
One goal of this study is to conduct a conceptual replication of violent video game effects on aggression in children across gender. A second objective is to provide experimental evidences at one potential moderating variable of violent game effect-trait aggression. The nal objective is to look at one potential mediating variable of violent game effect-anger. On the basis of GAM and previous ndings, our study proposes the following hypotheses: Hypothesis 1: Playing a violent video game will lead to more aggressive behavior than playing a neutral video game.
Hypothesis 2: Trait aggression and gender will moderate the violent video game effect on aggressive behavior. Hypothesis 3:Anger will mediate the violent video game effect on aggressive behavior.

Participants
The Research Ethics Committee of Southwest University approved the study. The participants were 248 children (50% females; M age = 6.09, SD = 0.21) recruited from two Chinese kindergartens in Chongqing, Southwest of China. Participants were all of Han ethnicity. Half of them (n = 124; 50% females) were randomly assigned to play a 20-min violent video game (treatment group), and the other half (n = 124; 50% females) were randomly assigned to play a 20-min neutral video game (control group). No participant failed to complete the experiment. We obtained written and informed consent from all participants and their parents. Participants completed a battery of questionnaires in the form of oral selfreport, including the Brief Aggression Questionnaire (BAQ), Anger Sub-scale of BPAQ. Then they completed the Hot Sauce Paradigm (HSP) in the quiet halls of kindergarten.

Experimental design
Participants were randomly assigned to one of 4 conditions in a 2 (game: violent vs. neutral) x 2 (gender: males vs. females) between-subjects experimental design (A randomized control study). The independent variables were game and gender, and the dependent variables were anger and aggressive behavior (setting levels of hot sauce powder).

Video games
We chose two video games "Street Fighter II" and "Tetris" as the materials for children to play. Street Fighter II is a Japanese ghting game. Players can choose one of the characters to ght against the opponents, and use the blood content to represent the lifeline of the game character. The game characters come from different countries (e.g., USA, Russia, China) and all of them have unique killing skills. Tetris is a Russia puzzle game. Different shapes of plates are made up of small squares, which fall from the top of the screen one by one. Players can make them spell one or several pieces at the bottom of the screen by adjusting the position and direction of the plates. These complete bars will disappear immediately, making place for the newly fallen plates, and players will be rewarded with corresponding points. These blocks that have not been eliminated are piling up. Once they reach the top of the screen, players lose the game.
We invited 10 game developers, 10 kindergarten teachers, 10 child psychologists and 10 parents to assess the violent degree of the two games with a scale anchored 5-point score (1 = low, 5 = high) to be carefully matched on variables in terms of the following criteria: (1) minutes. An independent sample t-test was used to compare the violent degree of the two video games. The results showed that the rating score of Street Fighter II was signi cantly higher than that of Tetris in terms of Violent Content and Violent Scene. However, there is no signi cant difference in Pleasure, Di culty, Familiarity, Interest, Excitement, and Action (see Table 1). Based on the criteria that media violence mainly depends violent content and violent scene (Anderson & Dill, 2000), we therefore select Street Fighter II and Tetris as the violent video game and the neutral video game, respectively. We used a self-reported method to assess trait aggression of kindergartners because of their limited literacy (i.e., The experimenter asked questions orally and the subjects answered them orally). BAQ is a 12-item standardized scale, including physical aggression (e.g,. if someone hurts me, I will hurt back), verbal aggression (e.g., when someone hurts me, I will tell him what I think), anger (e.g., I cannot control my temper) and hostility (e.g., I often feel that my partner calls me a nickname behind my back). It was a Likert 5-point scale (1 = very disagree, 5 = very agree). In the present study, the Cronbach's alpha of total scale was 0.98, and the total score was used.

Anger
We used the sub-scale of anger from Buss-Perry Aggression Questionnaire (BPAQ; items 1,5,7,8,9,15,19,23,28) to measure levels of anger (Buss & Perry, 1992). We also used a self-reported method to assess anger of kindergartners after they played the assigned game. The sub-scale of anger was anchored ve points (1 = very disagree, 5 = very agree). The internal consistency reliability coe cient of anger sub-scale of BPAQ was 0.83. In the present study, the Anger Sub-scale showed a good internal consistency reliability (Cronbach's α = 0.97).

Aggressive behavior
The hot sauce paradigm (HSP), an arti cial/proxy measure of aggression, was used to measure kindergartner's aggressive behavior in laboratory settings due to ethical reasons. Speci cally, the experimenter showed participants a picture of a child who was very uncomfortable when eating peppers.
However, participants were required to choose spoons of hot sauce powder from zero spoon (no aggressive behavior = zero points) to ve spoons (aggressive behavior = 5 points) for the child in the picture to eat. The speci c spoon of hot sauce participants chose for the child was a measure of aggressive behavior. The more spoons the participants chose for the child in the picture, the more aggressive the participants were. Previous research has shown that this experimental paradigm is positively correlated with the trait aggression scores measured by Buss-Perry Aggression Questionnaire and has good reliability and validity (Adachi, 2015; Sun & Liu, 2019).

Procedure
Following the selection of the participants, the subjects and their parents signed the informed consent forms and agreed to participate in the experiment. Participants were then asked to complete BAQ to measure trait aggression. Afterwards, participants were randomly assigned to play an assigned violent or a neutral game for 20 minutes. 62 males played a violent video game, while 62 males played a neutral video game; 62 females played a violent video game, while 62 females played a neutral video game.
Then experimental assistants interviewed with the participants about their anger using anger sub-scale of BPAQ. Finally, each participant completed the HSP and received a nice gift for participation. No participants expressed any suspicion during debrie ng process by a post-experimental interview. Table 2 summarizes the means, standard deviations, and the cell sample sizes of aggressive behavior across 4 conditions.

ANOVA on aggressive behaviors
To test Hypotheses 1 and 2, we used a 2 x 2 ANOVA to test the violent video effects on aggressive behavior. There main effect of game on aggressive behavior was small but signi cant. Playing a violent video game led to more aggressive behavior than playing  (Fig. 1).

Trait aggression as a moderator of aggressive behavior
To test Hypothesis 2 that trait aggression is a potential moderator of violent video game effect on aggressive behaviors, we run a moderation analysis by conducting the PROCESS macro 3.0 (Model 1) of SPSS 21.0 with all data standardized. In this model, violent video game is entered as a predictor, trait aggression (continuous variable) is the moderator, aggressive behavior is the outcome variable, and gender is controlled as a covariate (because of signi cant game x gender interaction). The results show that the moderating effect of trait aggression is not signi cant (F(1,243) = 0.14, β = 0.02; SE = 0.06, 95% CI = [-0.10, 0.15])(see Table 3).

Discussion
As expected, our results support Hypothesis 1, which suggests that brief exposure to a violent video game produces a small but signi cant increase in aggressive behavior. As such, the result of a positive game violence-aggression linkage is consistent with previous researches that violent video game exposure was Zhen, Xie, Zhang, Wang, & Li, 2011). The nding suggests that policy makers may consider establishing a policy to assess and classify the level of violence in video games played by kindergartners. Educators and parents may prevent kindergartners from being exposed to violent video games to prevent their possible aggressive behaviors.
Our results support Hypothesis 2, which suggests that gender is a moderator of aggressive behavior. The nding replicates previous study that boys are more sensitive to aggression than girls (Smith & Waterman, 2005). Males are more likely to solve interpersonal con icts by resorting to violence (e.g., bullying, perpetration), whilst females have more empathetic concern and sensitivity of moral cognition than males in the face of violence (Jolliffe & Farrington, 2006;Toussaint & Webb, 2005). In addition, males are more aggressive than females because males involve more physical activity than females during childhood (Archer, 2009 Our results support Hypothesis 3, which suggests that the violent game effect on aggressive behavior is mediated by increased anger. Thus the anger-aggression link may reinforce the positive game violenceaggression link. This nding is consistent with prior cognitive-neoassociationistic analysis that anger is related to aggression (Berkowitz, 1990;Giumetti & Markey, 2007;Petikas, 2013). The possible reason is that violent video games arouse aggressive affect to cause aggression, at least when being provoked (as they are in this experimental design). In particular, the mediation effect of anger on aggressive behavior is especially signi cant for males, indicating that males may be more easily aroused anger than females under the context of violent video games. Given that the game x gender interaction and mediating effect were signi cant for males, educators may teach males in kindergarten how to vent anger in order to alleviate aggressive behavior after violent video game exposure.
The ndings of the experimental study provides some useful information on the question of how playing a violent video game increases aggression across gender by recruiting a non-Western children sample. The experimental method and randomization of participants does allow us to draw causal inferences. The causal relationship can be claimed as a strong point of our study. In addition, the mediating role of anger indicates that reducing exposure to violent video games and reasonably venting anger may be effective intervention measures to reduce aggressive behavior of kindergartners (especially for males). Anger (internal state) mediates the violent video game effects on aggression in Chinese kindergartners, which also supports the GAM as an effective theoretical model. Despite the strengths discussed above, the present study has several limitations as follows. First, the similar age sample (6-year-olds) may limit the generalizability of the nding to other age groups. Future experimental study may consider expanding the age range to improve the external validity. Second, we have made claims that play a violent video game leads to increased aggression, especially for males.
However, the label of aggression is often associated with acts of physical aggression, verbal aggression, and relational aggression. While the hot sauce paradigm is interesting and is an ethical way of measuring "aggression" in the lab, it really only infers aggressive behavior. So far, there are hardly studies can suggest that this task is related to observed or self-reported physical or relational aggression in the real world. Thus perhaps it is an overstatement to suggest that playing a violent video game leads to aggression in real life settings. More caution is needed when interpreting the ndings related to "aggressive behavior" in the hot sauce paradigm. The potential lack of ecological validity between the hot sauce paradigm in the laboratory and aggressive behavior in the real world should be speci cally addressed in future research. Third, although the BAQ and Anger Sub-scale have good reliability, they may not be suitable for the measurement of kindergartners in China. Researchers may be skeptical of the reliability of six-year-olds lling out self-report measures of aggression and anger or, for that matter, understanding the hot sauce paradigm may be questioned. Therefore, we may consider the development of Chinese kindergartner's anger and trait aggression scale in the form of oral report to avoid potential cultural differences. Besides, perhaps the novelty to theoretical contribution is limited in this eld because dozens of studies have addressed the issue concerning the impact of violent video games on aggressive behavior in Western adults, although these ndings have been controversial.

Conclusion
In closing, our ndings suggest that short-term exposure to violent video games may constitute an important situational contributor to increases in aggressive behavior, especially for males. To some extent, violent video game exposure may be a risk factor for aggression in Chinese kindergartners. In addition, the violent video game effect on aggression was mediated by increased anger, especially for males. This study further expands the understanding of the mechanisms underlying violent video game effect on aggressive behavior in Chinese kindergartners by the role of anger, and its ndings are novel and interesting. Our experimental work provides an important piece of this knowledge. Findings support the general aggression model (GAM) and provide useful information for policy and practice. For example, educators and parents may prompt children to train to vent their anger in an appropriate way to decrease aggression. Speci cally, males may be regarded as the key group in the reduction and prevention of aggression under the context of violent video games. To this end, the current experiment offers a foundation for future work in this eld.

Declarations
Author Contributions QZ, YC, LC and JT conceived and designed the experiments. QZ, LC and JT performed the experiments. QZ and YC analyzed the data. QZ, YC, LC and JT wrote the draft. All authors reviewed the manuscript.