The spreading impact of playing violent video games on aggression (2017)

Computers in Human Behavior

Available online 17 November 2017

Tobias Greitemeyer^,

https://doi.org/10.1016/j.chb.2017.11.022

Highlights

• Violent video game exposure is related to aggressive behavior in the player.

• This effect in turn spreads across the player’s social network.

• Non-players are more aggressive if they are connected to players.

Abstract

Violent video game exposure has been shown to increase aggression in the player. The present research examines the idea that violent video game play does not only have an impact on the player, but also on the player’s social network. In fact, egocentric social networking analyses showed that playing violent video games is associated with increased aggression, which then spreads among connected individuals. Even participants that do not play violent video games themselves reported more aggression when their social network consists of individuals who do play violent video games. Psychologists and the public alike have been concerned that violent video game exposure has the potential to increase aggression on a societal level. As the present study shows, not only players of violent video games but also their social network may contribute to this phenomenon.

Keywords

video games;
social networks;
aggression;
contagion

1. Introduction

Violent video game exposure (VVE) is associated with aggressive outcomes. Although some studies failed to find significant effects (e.g., Charles et al., 2013 ; Engelhardt et al., 2015), meta-analyses (Anderson et al., 2010 ; Greitemeyer and Mügge, 2014) showed that VVE significantly increases the accessibility of aggressive thoughts, hostile affect, and aggressive behavior and that these effects are consistently found in experimental, cross-sectional, and longitudinal studies (e.g., Anderson et al., 2004 ; Anderson et al., 2007). It thus appears that playing violent video games does affect the player’s social behavior outside the virtual world.

Given the widespread use of violent video games, there has been a debate about their negative impact on a societal level. Although the effect of VVE on aggression is not large (around r = .19, Anderson et al., 2010 ; Greitemeyer and Mügge, 2014), even small effects (and the effect of violent video games is small to medium in its effect size) can have a negative impact on societal level when many people are exposed to it (which certainly applies to VVE). Thus, it has been argued that VVE could lead to societal harm (e.g., Anderson et al., 2010). The present research tests the idea that VVE may not only influence the player but also the player’s social network (“people with whom an individual is directly involved” [Fischer, 1982, p. 2]). If not only players of violent video games but also their social network respond with increased aggression, concern about the harmful effects of playing violent video games is even more warranted.

2. Literature review

As noted above, abundant previous research has addressed the effects of VVE on the player’s aggression (for overviews, Anderson & Gentile, 2014; Krahé, 2015). In contrast, to the best of my knowledge, no study has addressed how VVE affects the player’s social network. It is proposed that VVE is associated with aggression in the player and that this increased aggression spreads across the player’s social network in that the social network also becomes more aggressive. Most importantly, it is further proposed that the increased aggression of the social network emerges even when controlling for the social network’s amount of VVE. That is, even individuals who do not play violent video games may become more aggressive when they are connected to individuals who do play violent video games. Finally, the player’s level of aggression should account for the impact of the player’s VVE on the social network’s aggression.

Previous research has provided overwhelming evidence that psychological constructs can spread across network ties. For example, Christakis and Fowler (2007) found that a person’s chances of becoming obese increased by 57% if they had a friend who became obese. Overall, phenomena as diverse as smoking (Christakis & Fowler, 2008), fertility at the workplace (Pink, Leopold, & Engelhardt, 2014), voting behavior (Nickerson, 2008), cooperative behavior (Rand, Arbesman, & Christakis, 2011), and happiness (Bliss, Klourmann, Harris, Danforth, & Dodds, 2012) have been shown to spread in social networks.

This would suggest that playing violent video games can be contagious in that individuals are more likely to play violent video games if their social network consists of violent video game players. However, the present research aims to go one step further: it will be examined to what extent the effect of VVE spreads across social networks, in that it is proposed that VVE increases the player’s aggression, which then evokes aggression in the player’s social network. This hypothesis is based on the following reasoning.

Perhaps the best predictor of aggression is provocation (e.g., an insult). According to classical models of frustration-aggression (Dollard, Doob, Miller, Mowrer, & Sears, 1939) as well as more modern cognitive neoassociation models (Berkowitz, 1989), aggression increases the risk of aggression (see also the GAM; Anderson & Bushman, 2002). In fact, it is well-known that aggression and violence spread among connected individuals (for a review, Dishion, & Tipsord, 2011). For example, in an organizational setting, Foulk and colleagues (Foulk,Woolum, & Erez, 2016) found that low-intensity negative behaviors like rudeness spread from person to another. Moreover, a nationally representative sample of US adolescents showed that participants were more likely to engage in violent behavior (e.g., pulling a weapon on someone) if a friend had engaged in the same behavior (Bond & Bushman, 2017). Hence, individuals who are connected to (aggressive) players of violent video games could become more aggressive even if they do not play the games themselves. The present study provides the first empirical test of this reasoning, employing an egocentric social networking analysis (Clifton & Webster, 2017).

In egocentric social networking analyses, participants are asked to report on how they perceive their social contacts (called “friends” in the following). It is anticipated that the friends’ VVE is positively associated with aggression in the participants and that this relationship statistically holds when controlling for the participant’s VVE. It is further anticipated that the friends’ level of aggression mediates the impact of the friends’ VVE on the participant’s aggression.

3. Method

Sixty-seven participants did not complete the questionnaire, and were thus not included in the analyses. The final sample included 998 individuals (499 women, 499 men, mean age = 36.8 years, SD = 11.2) who took part on MTurk. Because the questionnaire was very short, no attention checks were included. There were no further data exclusions. All participants were run before any analyses were performed, and all variables analyzed are reported.

At the onset, participants learned that this would be a survey about themselves and their social network. After providing demographics, reported aggressive behavior was assessed. To this end, participants received 10 items (e.g., “I have hit another person” and “I have said nasty things about another person behind his/her back”) that have been successfully employed in previous research (Krahé & Möller, 2010). For each item, participants indicated how often they had shown the respective behavior in the past six months. All items were rated on 5-point scales ranging from 1 (never) to 5 (very often), and scores were averaged to form a composite index (α = .90). They also responded to how often they play violent video games, employing one item: “How often do you play violent video games (where the goal is to harm other game characters)?” (1 = never to 7 = very often).

As in previous egocentric social networking research (e.g., Mötteli & Dohle, in press; Stark & Krosnick, 2017), participants were then asked to answer questions about five individuals they feel closest to and with whom they talked about important matters in the last few months (friends). They learned that these may be friends, coworkers, neighbors, relatives. For each friend, they responded to the same aggression (αs between = .90 and .91) and VVE questions as they responded to themselves. Responses to the five friends were then averaged. Finally, participants were thanked and asked what they thought this experiment was trying to study. Many of the participants noted the link between violent video games and aggression, but none noted the exact hypothesis.

4. Results

Descriptive statistics and intercorrelations of all measures are shown in Table 1. VVE was related to aggression, both for the participant and the friends. Moreover, participant’s and friends’ VVE as well as their level of reported aggression, respectively, were positively related. Most importantly, the relationship between the participant’s aggression and the friends’ VVE was significant.

Table 1.

Means, Standard Deviations, and Bivariate Correlations

	M	SD	1	2	3
1. VVE participant	2.92	2.08
2. Aggression participant	1.39	0.54	.20^∗
3. VVE friends	2.37	1.20	.59^∗	.20^∗
4. Aggression friends	1.38	0.46	.20^∗	.72^∗	.27^∗

Note: ^∗p < .001

Table options

To examine whether the friends’ VVE is associated with the participant’s aggression when controlling for the participant’s VVE, a bootstrapping analysis was performed (which includes findings for the two main effects as well as the interaction). In fact, the impact of the friends’ VVE remained significant (point estimate = .13, SE = .03, t = 4.38, p < .001, 95% CI = 0.07, 0.18). For both, the participant’s VVE (point estimate = .09, SE = .02, t = 4.29, p < .001, 95% CI = 0.05, 0.13) and the interaction (point estimate = -.02, SE = .01, t = 3.09, p = .002, 95% CI = -0.03, -0.01), the confidence interval also did not include 0. At high levels of the participant’s VVE (+1 SD), the friends’ VVE was not related to the participant’s aggression (point estimate = .02, SE = .02, t = 1.21, p = .227, 95% CI = -0.02, 0.06). In contrast, at low levels of the participant’s VVE (-1 SD, equals no VVE in the present data set), the friends’ VVE was related to the participant’s aggression (point estimate = .11, SE = .02, t = 4.44, p < .001, 95% CI = 0.06, 0.15), suggesting that the participants are more aggressive when their social network plays violent video games even though they do not play violent video games themselves ( Figure 1).

Figure 1.

Simple slopes of the interactive effect of the friends’ VVE and the participant’s VVE on the participant’s aggression.

Figure options

Finally, it was examined whether the friends’ level of aggression accounted for the impact of the friends’ VVE on the participant’s aggression. When friends’ VVG and friends’ level of aggression were entered simultaneously, the regression equation accounted for substantial variance in participant’s level of aggression, F(2, 995) = 527.58, R² = .52, p < .001. In addition, friends’ level of aggression received a significant regression weight, t(995) = 31.42, β = .72, 95% CI = .78, .88, p < .001, whereas friends’ VVE did not, t(995) = 0.01, β = .00, 95% CI = -.02, .02, p = .992. This mediation pattern is shown in Figure 2. A Sobel test showed that the indirect effect was significantly different from zero, Sobel test statistic = 8.49, p < .001.

Figure 2.

Mediation of the impact of the friends’ VVE on the participant’s aggression by the friends’ aggression. All paths are significant unless otherwise noted. β^∗ = the coefficient from friends’ VVE to participant’s aggression when controlling for friends’ aggression.

Figure options

5. Discussion

Video games have gained wide popularity (Lenhart et al., 2008) and most of the top-selling video games contain violence (Dill, Gentile, Richter, & Dill, 2005). Hence, the question of whether VVE makes the players aggressive has been a topic of hot debate. What has been missing from this debate, however, is the impact of VVE on people with whom the player is involved. As the present study shows, even individuals who do not play violent video games may have higher levels of aggression if they are connected to violent video game players.

As in previous research (for meta-analyses, Anderson et al., 2010 ; Greitemeyer and Mügge, 2014), VVE was associated with increased aggression in the player. The friends’ level of aggression, in turn, was closely related to the participant’s aggression. In fact, the friends’ level of aggression accounted for the impact of the friends’ VVE on the participant’s aggression. Overall, this pattern suggests that increased aggression as a consequence of VVE may spread across individuals, in that the social network of a violent video game player becomes more aggressive because of increased aggression in the player.

It is noteworthy, however, that the correlational design does not allow any causal conclusions. Longitudinal or experimental studies are needed to provide evidence of the hypothesis that the impact of VVE on the player’s aggression indeed spreads across the social network, in that individuals become more aggressive who are connected to the player. It should be also kept in mind that self-reported data rather than actual behavior were assessed. This is particular important insofar as egocentric network data rely on the participant’s perception of the friend’s characteristics. It is thus conceivable that participants overestimated the extent to which they are similar to their friends. Moreover, participant’s perception of the friend’s VVE and level of aggression may have been driven by what participants believe how VVE and aggression are associated. For example, they may perceive a friend as being aggressive and thus (wrongly) infer that the friend plays violent video games. Please also note that the correlation between participant’s and their friends’ aggression levels was very high (which also applies for the correlation between participant’s and friends’ VVE), suggesting that participants judge their friends by using their self-ratings as a starting point. However, such a tendency would reduce the unique effect of friends’ VVE scores on the participant’s aggression when controlling for the participant’s VVE. Likewise, the finding of the interaction that participants that do not play violent video games are more aggressive when their friends play violent video games also suggests that participant’s ratings of their friends do not simply mirror their self-ratings. Nevertheless, future research that employs sociocentric networks in which the information about the friends is provided by the friends themselves would be very welcome.

One could also examine whether players of violent video games do not only promote aggression in their direct social networks, but also in their friends’ friends. Psychological constructs appear to spread to three degrees of separation (Christakis and Fowler, 2007 ; Christakis and Fowler, 2008). That is, players of violent video games may increase aggression in their social network, who will respond with increased aggression against a third person, who, as a consequence, will also become more aggressive.

On an applied level, the present research will prove useful to anyone whose goal it is to decrease aggression. For example, recent findings suggest that cooperatively playing a violent video game in a team (relative to playing the same video game alone) counteracts (at least in part) the negative effects of VVE on aggression (Mihan et al., 2015 ; Velez et al., 2016). Hence, convincing the video game industry to create cooperative video games that are highly attractive to consumers may ameliorate negative effects of VVE in everyday life. Importantly, any policy intervention that alters the player’s VVE does not only influence the player’s aggression but may also have an indirect effect on the player’s social network.

There are also important implications for researchers interested in social network effects. Previous work has focused on how certain characteristics spread across a social network. In contrast, the present research addresses how an effect (i.e., aggression) of a characteristic (i.e., violent video game exposure) spreads in social networks. Even when the characteristic does not spread (i.e., the social network does not play violent video games), the social network may still exhibit the same consequence (increased aggression) as the player.

It should be stressed that video game exposure does not necessarily harm interpersonal relations. To the contrary, research has shown that playing prosocial video games (in which the aim is to benefit other game characters) increases prosocial outcomes (for reviews, Greitemeyer, 2011 ; Greitemeyer and Mügge, 2014). Hence, future research may reveal that prosocial video game play does not only positively influence the player but spreads across the player’s social network in that the player’s friends become more helpful.

To conclude, aggression and violence have been shown to spread across network ties (Dishion, & Tipsord, 2011). As Huesmann (2012) put it: “One of best-established findings in the psychological literature on aggressive and violent behavior is that violence begets violence” (p. 63). Likewise, the present study suggests that VVE makes the player more aggressive, which then spreads through their social networks. It has been proposed that VVE has the potential to increase aggression on a societal level (Anderson et al., 2010). It appears that not only players of violent video games but also their social network may contribute to this phenomenon.

Acknowledgements:

This research was supported by grant P28913 from the Austrian Science Fund.

References

Anderson and Bushman, 2002
C.A. Anderson, B.J. Bushman
Human aggression
Annual Review of Psychology, 53 (2002), pp. 27–51
CrossRef

Quick Links