Introduction
Violent Video Games and Aggression
The relationship between violent video games and adolescent aggression has become a hot issue in psychological research [Wiegman and Schie, 1998; Anderson and Bushman, 2001; Anderson et al., 2010; Ferguson et al., 2012; Greitemeyer, 2014; Yang et al., 2014; Boxer et al., 2015]. Based on the General Aggression Model [GAM], Anderson et al. suggested that violent video games constitute an antecedent variable of aggressive behavior, i.e., the degree of exposure to violent video games directly leads to an increase of aggression [Anderson and Bushman, 2001; Bushman and Anderson, 2002; Anderson, 2004; Anderson et al., 2004]. Related longitudinal studies [Anderson et al., 2008], meta-analyses [Anderson et al., 2010; Greitemeyer and Mugge, 2014], event-related potential studies [Bailey et al., 2011; Liu et al., 2015], and trials about juvenile delinquents [DeLisi et al., 2013] showed that exposure to violent video games significantly predicts adolescent aggression.
Although Anderson et al. insisted on using the GAM to explain the effect of violent video games on aggression, other researchers have proposed alternative points of view. For example, a meta-analysis by Sherry [2001] suggested that violent video games have minor influence on adolescent aggression. Meanwhile, Ferguson [2007] proposed that publication bias [or file drawer effect] may have implications in the effect of violent video games on adolescent aggression. Publication bias means that compared with articles with negative results, those presenting positive results [such as statistical significance] are more likely to be published [Rosenthal and Rosnow, 1991]. A meta-analysis by Ferguson [2007] found that after publication bias adjustment, the related studies cannot support the hypothesis that violent video games are highly correlated with aggression. Then, Ferguson et al. proposed a Catalyst Model [CM], which is opposite to the GAM. According to this model, genetic predisposition can lead to an aggressive child temperament and aggressive adult personality. Individuals who have an aggressive temperament or an aggressive personality are more likely to produce violent behavior during times of environmental strain. Environmental factors act as catalysts for violent acts for an individual who have a violence-prone personality. This means that although the environment does not cause violent behavior, but it can moderate the causal influence of biology on violence. The CM model suggested that exposure to violent video games is not an antecedent variable of aggressive behavior, but only acts as a catalyst influencing its form [Ferguson et al., 2008]. Much of studies [Ferguson et al., 2009, 2012; Ferguson, 2013, 2015; Furuya-Kanamori and Doi, 2016; Huesmann et al., 2017] found that adolescent aggression cannot be predicted by the exposure to violent video games, but it is closely related to antisocial personality traits, peer influence, and family violence.
Anderson and his collaborators [Groves et al., 2014; Kepes et al., 2017] suggested there were major methodological shortcomings in the studies of Ferguson et al. and redeclared the validity of their own researches. Some researchers supported Anderson et al. and criticized Ferguson’s view [Gentile, 2015; Rothstein and Bushman, 2015]. However, Markey [2015] held a neutral position that extreme views should not be taken in the relationship between violent video games and aggression.
In fact, the relation of violent video games to aggression is complicated. Besides the controversy between the above two models about whether there is an influence, other studies explored the role of internal factors such as normative belief about aggression and external factors such as family environment in the relationship between violent video games and aggression.
Normative Beliefs About Aggression, Violence Video Games, and Aggression
Normative beliefs about aggression are one of the most important cognitive factors influencing adolescent aggression; they refer to an assessment of aggression acceptability by an individual [Huesmann and Guerra, 1997]. They can be divided into two types: general beliefs and retaliatory beliefs. The former means a general view about aggression, while the latter reflects aggressive beliefs in provocative situations. Normative beliefs about aggression reflect the degree acceptance of aggression, which affects the choice of aggressive behavior.
Studies found that normative beliefs about aggression are directly related to aggression. First, self-reported aggression is significantly correlated to normative beliefs about aggression [Bailey and Ostrov, 2008; Li et al., 2015]. General normative beliefs about aggression can predict young people’s physical, verbal, and indirect aggression [Lim and Ang, 2009]; retaliatory normative beliefs about aggression can anticipate adolescent retaliation behavior after 1 year [Werner and Hill, 2010; Krahe and Busching, 2014]. There is a longitudinal temporal association of normative beliefs about aggression with aggression [Krahe and Busching, 2014]. Normative beliefs about aggression are significantly positively related to online aggressive behavior [Wright and Li, 2013], which is the most important determining factor of adolescent cyberbullying [Kowalski et al., 2014]. Teenagers with high normative beliefs about aggression are more likely to become bullies and victims of traditional bullying and cyberbullying [Burton et al., 2013]. Finally, normative beliefs about aggression can significantly predict the support and reinforcement of bystanders in offline bullying and cyberbullying [Machackova and Pfetsch, 2016].
According to Bandura’s social cognitive theory [Bandura, 1989], violent video games can initiate adolescents’ observational learning. In this situation, not only can they imitate the aggressive behavior of the model but also their understanding and acceptability about aggression may change. Therefore, normative beliefs about aggression can also be a mediator between violent video games and adolescent aggression [Duan et al., 2014; Anderson et al., 2017; Huesmann et al., 2017]. Studies have shown that the mediating role of normative beliefs about aggression is not influenced by factors such as gender, prior aggression, and parental monitoring [Gentile et al., 2014].
Family Environment, Violence Video Games, and Aggression
Family violence, parenting style, and other family factors have major effects on adolescent aggression. On the one hand, family environment can influence directly on aggression by shaping adolescents’ cognition and setting up behavioral models. Many studies have found that family violence and other negative factors are positively related to adolescent aggression [Ferguson et al., 2009, 2012; Ferguson, 2013], while active family environment can reduce the aggressive behavior [Batanova and Loukas, 2014].
On the other hand, family environment can act on adolescent aggression together with other factors, such as exposure to violent video games. Analysis of the interaction between family conflict and media violence [including violence on TV and in video games] to adolescent aggression showed that teenagers living in higher conflict families with more media violence exposure show more aggressive behavior [Fikkers et al., 2013]. Parental monitoring is significantly correlated with reduced media violence exposure and a reduction in aggressive behavior 6 months later [Gentile et al., 2014]. Parental mediation can moderate the relationship between media violence exposure and normative beliefs about aggression, i.e., for children with less parental mediation, predictability of violent media exposure on normative beliefs about aggression is stronger [Linder and Werner, 2012]. Parental mediation is closely linked to decreased aggression caused by violent media [Nathanson, 1999; Rasmussen, 2014; Padilla-Walker et al., 2016]. Further studies have shown that the autonomy-supportive restrictive mediation of parents is related to a reduction in current aggressive behavior by decreasing media violence exposure; conversely, inconsistent restrictive mediation is associated with an increase of current aggressive behavior by enhancing media violence exposure [Fikkers et al., 2017].
The Current Study
Despite GAM and CM hold opposite views on the relationship between violent video games and aggression, both of the two models imply the same idea that aggression cannot be separated from internal and external factors. While emphasizing on negative effects of violent video games on adolescents’ behavior, the GAM uses internal factors to explain the influencing mechanism, including aggressive beliefs, aggressive behavior scripts, and aggressive personality [Bushman and Anderson, 2002; Anderson and Carnagey, 2014]. Although the CM considers that there is no significant relation between violent video games and aggression, it also acknowledges the role of external factors such as violent video games and family violence. Thus, these two models seem to be contradictory, but in fact, they reveal the mechanism of aggression from different points of view. It will be more helpful to explore the effect of violent video games on aggression from the perspective of combination of internal and external factors.
Although previous studies have investigated the roles of normative beliefs about aggression and family factors in the relationship between violent video games and adolescent aggression separately, the combined effect of these two factors remains unstudied. The purpose of this study was to analyze the combined effect of normative beliefs about aggression and family environment. This can not only confirm the effects of violent video games on adolescent aggression further but also can clarify the influencing mechanism from the integration of GAM and CM to a certain extent. Based on the above, the following three hypotheses were proposed:
Hypothesis 1: There is a significant positive correlation between exposure to violent video games and adolescent aggression.
Hypothesis 2: Normative beliefs about aggression are the mediator of exposure to violent video games and adolescent aggression.
Hypothesis 3: The family environment can moderate the mediation effects of normative beliefs about aggression in exposure to violent video games and adolescent aggression; exposure to violent video games, family environment, normative beliefs about aggression, and aggression constitute a moderated mediation model.
Materials and Methods
Participants
All subjects gave informed written consent for participation in this investigation, and their parents signed parental written informed consent. The study was reviewed and approved by the Professor Committee of School of Psychology, Nanjing Normal University, which is the committee responsible for providing ethics approvals. A total of 648 Chinese middle school students participated in this study, including 339 boys and 309 girls; 419 students were from cities and towns, and 229 from the countryside. There were 277 and 371 junior and high school students, respectively. Ages ranged from 12 to 19 years, averaging 14.73 [SD = 1.60].
Measures
Video Game Questionnaire [VGQ]
The Video Game Questionnaire [Anderson and Dill, 2000] required participants to list their favorite five video games and assess their use frequencies, the degree of violent content, and the degree of violent images on a 7-point scale [1, participants seldom play video games, with no violent content or image; 7, participants often play video games with many violent contents and images]. Methods for calculating the score of exposure to violent video games: [score of violent content in the game + score of violent images in the game] × use frequency/5. Chen et al. [2012] found that the Chinese version of this questionnaire had high internal consistency reliability and good content validity. The Chinese version was used in this study, and the Cronbach’s α coefficient of the questionnaire was 0.88.
Aggression Questionnaire [AQ]
There were 29 items in AQ [Buss and Perry, 1992], including four dimensions: physical aggression, verbal aggression, anger, and hostility. The scale used 5-point scoring criteria [1, very incongruent with my features; 5, very congruent with my features]. Scores for each item were added to obtain the dimension score, and dimension scores were summed to obtain the total score. The Chinese version of AQ had good internal consistency reliability and construct validity [Ying and Dai, 2008]. In this study, the Chinese version was used and its Cronbach’s α coefficient was 0.83.
Family Environment Scale [FES]
The FES [Moos, 1990] includes 90 true-false questions and is divided into 10 subscales, including cohesion, expressiveness, conflict, independence, achievement-orientation, intellectual-cultural orientation, active-recreational orientation, moral-religious emphasis, organization, and control. The Chinese version of FES was revised by Fei et al. [1991] and used in this study. Three subscales closely related to aggression were selected, including cohesion, conflict, and moral-religious emphasis, with 27 items in total. The family environment score was the sum of scores of these three subscales [the conflict subscale was first inverted]. The Cronbach’s α coefficient of the questionnaire was 0.75.
Normative Beliefs About Aggression Scale [NOBAGS]
There are 20 items in the NOBAGS [Huesmann and Guerra, 1997], which includes retaliation [12 items] and general [8 items] aggression belief. A 4-point Likert scale is used [1, absolutely wrong; 4, absolutely right]. The subjects were asked to assess the accuracy of the behavior described in each item. High score means high level of normative beliefs about aggression. The revised Chinese version of NOBAGS consists of two factors: retaliation [nine items] and general [six items] aggression belief. Its internal consistency coefficient and test-retest reliability are 0.81 and 0.79. Confirmative factor analysis showed that this version has good construct validity: χ2 = 280.09, df = 89, χ2/df = 3.15, RMSEA = 0.07, SRMR = 0.04, NFI = 0.95, NNFI = 0.96, and CFI = 0.96 [Shao and Wang, 2017]. In this study, the Cronbach’s α coefficient of the Chinese version was 0.88.
Procedures
Group testing was performed in randomly selected classes of six middle schools. All subjects completed the above four questionnaires.
Data Analysis
IBM SPSS Statistics 22 was used to analysis the correlations among study variables, the mediating effect of normative beliefs about aggression on the relationship between exposure to violent video games and aggression, and the moderating role of family environment in the relationship between exposure to violent video games and normative beliefs about aggression. In order to validate the moderated mediation model, Mplus 7 was also used.
Results
Correlation Analysis Among Study Variables
In this study, self-reported questionnaires were used to collect data, and results might be influenced by common method bias. Therefore, the Harman’s single-factor test was used to assess common method bias before data analysis. The results showed that eigenvalues of 34 unrotated factors were greater than 1, and the amount of variation explained by the first factor was 10.01%, which is much less than 40% of the critical value. Accordingly, common method bias was not significant in this study.
As described in Table 1, the degree of exposure to violent video games showed significant positive correlations to normative beliefs about aggression and aggression; family environment was negatively correlated to normative beliefs about aggression and aggression; normative beliefs about aggression were significantly and positively related to aggression. The gender difference of exposure to violent video games [t = 7.93, p