Toxicity and prosocial behaviors in massively multiplayer online games: The role of mutual dependence, power, and passion

Abstract

Understanding factors that predict toxic and prosocial behavior in massively multiplayer online (MMO) games has drawn a great deal of scholarly attention. Prior work on this topic has primarily focused on individual and technological factors while overlooking the role of interpersonal dynamics. To fill this gap, this study examines if and how players’ perceptions of mutual dependence and power in MMO games are related to toxicity and prosocial behavior in games. Results from a survey of 782 Chinese game players suggest that players’ perceived power is positively related to prosocial behavior in games. Perceived mutual dependence and power are also indirectly related to prosocial and toxic behaviors through players’ passion for games. This study has theoretical implications for scholarship on toxicity, prosocial behaviors, and gameplay, while also providing design and policy implications for MMO games.

In 2019, the Anti-Defamation League (ADL) released a report that delineated a grim picture of the ubiquity of toxicity, such as abusive language and disruptive play, in massively multiplayer online (MMO) games. The report showed that 78% of MMO gamers experienced some form of toxicity while playing the games (ADL, 2019). Despite the ubiquity of toxicity, multiplayer gamers also engage in prosocial behaviors while playing games. As two sides of social interaction, prosocial and toxic behaviors have been widely examined in previous game research. A majority of studies have focused on individual factors such as players’ personalities (Velez & Ewoldsen, 2013) and technological factors such as game design (Kordyaka & Kruse, 2021) in predicting the behaviors. A few recent studies also showed that toxic game culture and skill disparity within teams were likely to breed toxicity (Cook, 2019; Shen et al., 2020; Kowert, 2020).

However, little is known about how interpersonal factors that emerge from ongoing social interactions in MMO games are related to prosocial and toxic behaviors in the games. In fact, the interpersonal relationships in MMO games are constantly in flux depending on dynamic in-game social interactions. Specifically, a fundamental and defining characteristic of such dynamic interpersonal relationships is interdependence—the extent to which an individual’s own and others’ behaviors affect each other’s behaviors and outcomes (Gerpott et al., 2018). Based on each individual’s own experiences, traits, and situational factors, individuals construct their own sense of interdependence (Columbus & Molho, 2022). Prior research has found that perceptions of interdependence were associated with anti- and prosocial behaviors (Depping & Mandryk, 2017; De Wit et al., 2017), yet this has received little attention in the context of games. It is unclear if and how perceived interdependence is related to toxic and prosocial behaviors in MMO games. This study aims to tackle this question, with a particular focus on two dimensions of interdependence—mutual dependence and power—which are highly relevant in the game context.

Perceptions of independence may be also indirectly related to gamer behaviors through players’ engagement with games, such as players’ passion for games. As an inclination toward activities that people value and spend time on (Vallerand et al., 2003), passion can be categorized as either congruent with one’s goals and values (i.e., harmonious passion) or conflicting with other aspects of their lives (i.e., obsessive passion). A person’s passion for an activity is concurrently shaped by their predisposition (Vallerand, 2010) and situational and interpersonal factors (Vallerand, 2010). When it comes to games, people’s passion for games can be predetermined by their choices of games, but it is also shaped by their feelings of relatedness and closeness in games (Johnson et al., 2021; Perry et al., 2018). Furthermore, research has shown that passion has interpersonal outcomes, such as destructive and reparative behaviors during and following conflicts (Carbonneau & Vallerand, 2013).

This study aims to examine how players’ perceived mutual dependence and power in games relate to their in-game prosocial and toxic behaviors, as well as the role of passion for games in mediating this relationship. We specifically focus on prosocial and toxic behaviors directed toward teammates because both behaviors are more likely to occur among teammates than between opponents (Cook et al., 2019). This is partly because team success entails a high level of collaboration (Verheijen et al., 2019), and failure to meet group expectations may engender toxicity (Kou, 2020). By underscoring the importance of perceived mutual dependence, perceived power, and passion, this research seeks to provide a deeper understanding of people’s constructive and destructive behaviors in games. The findings of this study offer practical implications for promoting players’ prosocial behaviors and curbing toxicity in MMO games.

Literature review

Toxicity and prosocial behavior in MMO games

Characterized by negative behaviors and effects, the concept of toxicity has appeared in various contexts. In organizational settings, for instance, toxicity refers to leadership styles that harm individuals within organizations and result in systemic damage to organizations (Green, 2014). In cyberspace, toxicity takes on the form of hostile, disrespectful expressions or anti-social behaviors (Suler, 2004). Among different contexts, toxicity is mostly examined in multiplayer games, where the concept refers to negative behaviors comprising abusive language and disruptive play, such as griefing and trolling (Adinolf & Turkay, 2018; Türkay et al., 2020). Terms relevant to toxicity include “incivility,” “anti-social behaviors,” “cyberbullying,” and “verbal aggression” (Kou, 2020; Shen et al., 2020). Compared to these well-studied phenomena, toxicity in games seems to be more pervasive and normalized (Shen et al., 2020), because players often see toxicity as acceptable, unavoidable, or simply banter (Beres et al., 2021).

One potential reason for the prevalence and normalization of toxicity in MMO games is the games’ design features. MMO games allow for the concurrent existence of a large number of players, featuring intense interpersonal communication (Shen, 2014). Players in this type of game normally need to cooperate with others to achieve their goals, thus having a higher chance of attributing failures to others compared to those in single-player games (Adinolf & Turkay, 2018). Competitiveness, which is inherent to these team-based games, further pushes players to place great importance on success and become aggressive to others (Kou, 2020). Making the situation worse, emotional turmoil caused by bad performance in games further undermines players’ performance, resulting in a vicious spiral of toxicity (Türkay et al., 2020).

With frustration and anger, players who are exposed to toxicity will be discouraged from playing these games, perhaps even permanently (Neto et al., 2017; Shores et al., 2014; Türkay et al., 2020). Though toxicity occurs both within groups and between groups, it was found to occur more frequently within groups than between groups (Cook et al., 2019), possibly because players lose tempers when their teammates do not meet their expectations (Kou, 2020), or they seek to urge teammates to play better through performing toxicity (Tan & Chen, 2022). Teammates of perpetrators were also found to be more susceptible to toxicity compared to players from opposing teams (Neto et al., 2017). Along with the adverse impact on player experience, game companies’ revenues are also jeopardized (Kordyaka, 2018; Neto et al., 2017). They need to expend a great deal of effort in tackling the issue, ending up slowing down the development of the game itself (PlayOverwatch, 2017).

However, toxicity is not the sole interpersonal outcome of MMO games. Prosocial behaviors, actions that are intended to benefit others (Carlo & Randall, 2002; Velez & Ewoldsen, 2013), also exist. This is because MMO games require teamwork, which can stimulate positive behaviors among players (Verheijen et al., 2019). Existing research outside the domain of games has revealed a variety of positive psychological outcomes of prosocial behaviors. For instance, prosocial acts can reduce depressive and aggressive symptoms (He et al., 2018), increase meaningfulness in life (Klein, 2017), and enhance subjective well-being (Weinstein & Ryan, 2010). Similarly, existing research in games has demonstrated various positive effects of in-game prosocial behaviors and prosocial video games. For example, in-game prosocial decisions were found to increase people’s subsequent donating behaviors (Iten et al., 2018). Exposure to prosocial video games activates people’s prosocial thoughts and results in further prosocial acts (Greitemeyer & Osswald, 2010).

As two sides of social interaction, toxic and prosocial behaviors can be present concurrently in MMO games. Previous research shows that prosocial behaviors are not necessarily associated with changes in antisocial behaviors (Kavussanu & Boardley, 2009). One potential reason is that while cooperative elements in MMO games can trigger more helping behaviors (Verheijen et al., 2019), competitive elements in MMO games also increase players’ overall aggression (Kou, 2020), as players may perform toxicity to urge teammates to perform well (Tan & Chen, 2022). Given that both cooperative and competitive elements are intense in MMO games, prosocial and toxic behaviors should be examined together to derive a holistic picture of the factors, beyond cooperative and competitive elements, that contribute to interpersonal behaviors in the games. Nevertheless, few studies have examined these two behaviors concurrently.

A plethora of game studies have examined factors that predict prosocial and toxic behaviors, respectively. For instance, players’ experience of toxicity victimization, attitudes toward toxicity, and motivations to play games were found to predict toxic behaviors (Kordyaka et al., 2020; Tan & Chen, 2022). Players who are younger, experienced, play competitive modes, and use damage-dealing characters are also more prone to perform toxicity (Kordyaka et al., 2020; Shen et al., 2020; Shores et al., 2014).¹ In addition, players’ personalities (Velez & Ewoldsen, 2013), motivations (Velez & Ewoldsen, 2013), psychological states (Johnson et al., 2021), and game genres (De Simone, 2013) are associated with prosocial behaviors in games. Certain affordances of games, such as anonymity and asynchronicity, were found to indirectly relate to toxic behaviors (Kordyaka & Kruse, 2021). Types of gameplay (competitive vs. cooperative) were also found to influence players’ helping behaviors in games (Velez & Ewoldsen, 2013; Verheijen et al., 2019).

Though enlightening, these studies primarily focus on individual and technological factors in predicting prosocial and toxic behaviors in games, while overlooking interpersonal factors that emerge from ongoing social interactions in games. Indeed, few studies have examined the impact of social environment and situation on gamer behaviors. For instance, prior work suggests that toxic game culture and skill disparity among teammates are likely to breed toxicity (Cook, 2019; Shen et al., 2020; Kowert, 2020). Yet these factors do not capture the interpersonal dynamics that emerge from ongoing social interactions in games. In fact, various technological factors (e.g., game design) and individual factors (e.g., the roles players take) will influence social relations that players perceive, which further shape how players behave in games (Gerpott et al., 2018). For example, the design of asymmetric game characters’ skills influences players’ perceptions of connectedness, interdependence, and competitiveness (Harris & Hancock, 2019; Shores et al., 2014). Understanding interpersonal factors, such as dependence among players and power dynamics, is thus imperative. Below, we turn to the notions of mutual dependence and power, and discuss them in the context of MMO games.

Perceived mutual dependence and power in MMO games

MMO games are identified as “extremely social games,” with social interaction being their most prominent feature (Cole & Griffiths, 2007). Essentially, nearly all social interactions are characterized by interdependence, meaning that each individual’s behavior and outcomes are affected by others (Gerpott et al., 2018). Among various interactions, interdependence is typically examined in close relationships, cooperation, and teamwork (Columbus et al., 2021; Harris & Hancock, 2019), because interactions in these circumstances are more intense. By the same token, interdependence should have been examined widely in the context of MMO games, given the salient interaction features of the games, whereas only a few game studies have examined this issue. With interdependence being recognized as a key mechanism affecting social bonds and the popularity of games, an increasing number of studies have begun to call for work in this area (Depping & Mandryk, 2017; Harris & Hancock, 2019; Rocha et al., 2008).

As a multidimensional construct, interdependence can be divided into conflicts of interests, information certainty, future interdependence, mutual dependence, and power (Columbus et al., 2021; Gerpott et al., 2018). Conflicts of interest refer to behaviors that are good for one individual but bad for others. Information certainty refers to the degree to which a person knows others’ preferred outcomes. Future interdependence refers to situations in which people’s behaviors at present influence their behaviors and outcomes in future interactions (Gerpott et al., 2018). Because collaborations in MMO games normally underscore a shared goal, either to defeat non-player characters or other players, and such collaborations are likely to be “disposable”, meaning that players’ teammates are assigned by systems and they are unlikely to meet again, the constructs of conflict of interest, information certainty, and future interdependence might be less applicable in this context.

In the context of MMO games, the constructs of mutual dependence and power seem more relevant and applicable. Mutual dependence refers to the degree to which group members rely on one another to effectively perform a task (Gerpott et al., 2018). Regarding games, mutual dependence refers to the degree to which players need to interact and cooperate with others to perform game tasks and achieve game goals (Depping & Mandryk, 2017). Common ways of enhancing mutual dependence include building complementary roles (Harris et al., 2016), requiring information exchange (Kutlu et al., 2013), and setting complementary abilities (Harris & Hancock, 2019). Despite the game designs, players also construct their own sense of mutual dependence based on their prior experience, personality traits, and situational factors (Columbus & Molho, 2022; Gerpott et al., 2018). Such perceptions of mutual dependence further influence players’ behaviors in games (Columbus & Molho, 2022). In this light, this study chose to focus on perceived, instead of objective, mutual dependence.

In studies outside games, perceived mutual dependence was found to result in both intrapersonal and interpersonal outcomes. For example, in sports teams, team members’ perceptions of mutual dependence increase team cohesion and satisfaction (Evans & Eys, 2015). In close relationships, those who perceive more mutual dependence are likely to engage in cooperation more often (Columbus et al., 2021). In the context of games, players’ engagement is higher in closely coupled casual games than in loosely coupled ones (Beznosyk et al., 2012). Games with more mutual dependence trigger more connectedness, in-game relatedness, interpersonal trust, behavioral engagement, cooperation, and social capital (Depping & Mandryk, 2017; Depping et al., 2018; Harris & Hancock, 2019). In addition, players whose motivation is to take on specialized roles, a potential indicator of interdependence, are more likely to engage in ingroup helping behaviors (Velez & Ewoldsen, 2013). Grounded in these studies, we propose the following hypothesis:

H1: Perceived mutual dependence is (a) positively related to prosocial behaviors and (b) negatively related to toxicity.

Power refers to an asymmetric dependence, in which one’s actions have more influence than others (Galinsky et al., 2015). The power differential in games may stem from asymmetrical game roles. In multiplayer online battle arena games, for instance, some characters have more damage-dealing abilities and are more likely to be regarded as keys in the games (Shores et al., 2014). Players’ gaming experience, the money and time spent on the game, and rankings are also potential indicators of power (Ballard & Welch, 2017; Shen et al., 2020). As with mutual dependence, players also construct their own sense of power based on individualized experiences and traits despite their objective status. Such perceived power, unlike objective power featuring “asymmetric control” (Maner & Mead, 2010), allows all people in one situation to be concurrently high or low in the variable (Brick et al., 2021). Given that pertinent game research mainly focuses on objective power (Ballard & Welch, 2017; Shen et al., 2020; Shores et al., 2014), this study focuses on perceived power and defines it as the belief that one’s actions have more influence on the cooperation and outcomes of the games than others.

Notably, existing views on the effect of power on prosocial and antisocial behaviors tend to split. A large body of research suggests that power leads to antisocial behaviors. People who have more power or perceive themselves as having more power are less likely to engage in cooperative behaviors and more likely to make selfish choices (Righetti et al., 2015). In the context of games, players who have higher rankings or play characters with mostly damage-dealing abilities commit toxicity more often than others (Ballard & Welch, 2017; Shen et al., 2020; Shores et al., 2014). Potential explanations for these phenomena include the notion that powerful people lack perspective-taking and empathy (Galinsky et al., 2006) and care more about their dispositions and intrapsychic processes than situational factors (Galinsky et al., 2008).

On the other hand, an increasing number of studies have drawn attention to the positive impact of power on prosocial behaviors. It is argued that, when power holders construe power as a responsibility rather than an opportunity, they become responsive to others (De Wit et al., 2017; Tost, 2015). People with more power also have a higher tendency toward forgiveness, even in face of others’ negative manners (Karremans & Smith, 2010). Particularly, Maner and Mead (2010) showed that, unless powerful people perceived their dominant role to be threatened, they wielded their power to help groups achieve common goals. When intergroup competition is present, powerful people even place group success over their self-interest, even though they perceive their power as being threatened (Maner & Mead, 2010).

Given the conflicting views on the influence of power on prosocial and antisocial behaviors and the fact that research on the influence of perceived power on players’ behaviors in MMO games is scant, we pose the following research question:

RQ1: How does perceived power relate to prosocial and toxic behaviors in MMO games?

The mediating role of passion

While the direct link between perceived interdependence and prosocial and toxic behaviors seems plausible, it is important to explore mechanisms that explain this relationship. Johnson et al. (2013) suggest that players’ experience of play, especially their passion, serves as a significant mechanism explaining the intrapersonal outcomes of gameplay. Passion is an inclination toward activities that people like, value, and spend time on (Vallerand et al., 2003). The dualistic model of passion suggests that there are two types of passion: harmonious and obsessive (Vallerand et al., 2003), distinguished based on how activities are internalized into people’s identities. Harmonious passion results from autonomous internalization. People with such a passion engage in activities freely, without feeling compelled to. These activities, therefore, work in harmony with other aspects of people’s lives, including their pre-existing values and goals. On the contrary, obsessive passion results from controlled internalization. People pursue activities for the sake of certain contingencies attached to them, such as social acceptance. Therefore, people feel that they cannot help but engage in the activity, leading to conflict between the activities and other aspects of people’s lives (Vallerand et al., 2003).

Seemingly opposites, harmonious passion and obsessive passion are not dichotomous. They have been found to be positively correlated (e.g., Przybylski et al., 2009; Puerta-Cortés et al., 2017). This is because both passions underpin the enhanced value placed and time spent on the activity, as well as the inclusion of the activity into one’s identity (Vallerand et al., 2003). Despite the overlap between these two types of passion, one type of passion normally takes precedence over the other (Vallerand, 2010). Such a predominance is often shaped by people’s activity selection, valuation, and internalization (Vallerand, 2010), but is also shaped by situational and interpersonal factors while people engage in the activity. For instance, whether companies emphasize cooperation or competition influences employees’ passion for work (Vallerand, 2010). When it comes to games, although people’s passion for games can be predetermined by their choices of games, it is also shaped by their feelings of relatedness and closeness in games (Johnson et al., 2021; Perry et al., 2018). Specifically, with respect to obsessive passion, researchers have linked it to the notion of game addiction (Przybylski et al., 2009), which has been found to be predicted by interpersonal dependency, such as players’ attachment to others (Škařupová & Blinka, 2015). Therefore, people’s passion for games is likely to be shaped by their interpersonal relationships in games.

As an interpersonal factor, perceived mutual dependence and power in MMO games have the potential to influence players’ passion for games. Such an influence could be inferred from the link between the self-determination theory (SDT) and the dualistic model of passion (DMP). According to SDT, people will be intrinsically motivated to undertake activities if their basic needs are satisfied (relatedness, competence, autonomy) (Deci & Ryan, 2000). Over time, these need satisfactions are thought to relate to passion (Vallerand et al., 2003). Drawing on SDT and DMP, Johnson et al. (2021) found that people’s satisfaction with relatedness, competence, and autonomy in games was associated with harmonious passion and that satisfaction with relatedness was associated with obsessive passion. Holding et al. (2021) also found that need satisfaction in games was positively related to both types of passion, though the relationship with harmonious passion was much stronger. To illustrate, relatedness occurs when players recognize that they are important to others, and competence occurs when players feel that they are good at games (Rigby & Ryan, 2011). Previous research took relatedness as an important metric of mutual dependence (Arthur Jr et al., 2012) and suggested that perceived competence underlay power (Karakowsky et al., 2004). In this vein, perceived mutual dependence and power, as potentially characterized by relatedness and competence, may relate to players’ passion for games.

Furthermore, players’ passion might also be related to their interpersonal behaviors in games. Previous studies showed that having a harmonious passion was associated with reparative behaviors following conflict (Carbonneau & Vallerand, 2013), whereas having an obsessive passion was associated with aggressive behaviors towards others (Donahue et al., 2009). This is possibly because harmonious passion denotes autonomous internalization. People with such passion are likely to engage in activities with an open mind and feel less judged by others (Donahue et al., 2009; Mageau et al., 2005). In contrast, as an overpowering space in identity, obsessive passion will lead people to perform aggressive behaviors to defend their identities (Donahue et al., 2009). People with such passion experience more negative affect during and after activities (Lafreniere et al., 2009; Mageau et al., 2005), and such negative affect may further trigger negative behaviors in games. Additionally, as an important predictor of prosocial behaviors, empathy was found to be positively associated with harmonious passion, whereas it was negatively related to obsessive passion (Johnson et al., 2021).

Drawing on these studies, it is plausible to speculate that passion may mediate the relationships between perceived mutual dependence and power and prosocial and toxic behaviors in games. Thus, we postulate the following hypotheses:

H2: Harmonious passion mediates the relationship between perceived mutual dependence and power and prosocial and toxic behaviors in MMO games such that (a) perceived mutual dependence and power increase harmonious passion, which will in turn (b) increase prosocial behaviors, and (c) reduce toxicity in MMO games.

H3: Obsessive passion mediates the relationship between perceived mutual dependence and power and prosocial and toxic behaviors in MMO games, such that (a) perceived mutual dependence and power increase obsessive passion, which will in turn (b) reduce prosocial behaviors, and (c) increase toxicity in MMO games.

Method

Procedure and sample

We collected data via an online survey in October 2021. Respondents were recruited from a nationally representative sample of Chinese people via an online survey platform, Sojump (http://www.wjx.cn). Eligible participants were 18 years or older, played MMO games that allowed for in-game communication with teammates, participated in teamwork in MMO games in the past six months, and played MMO games for at least three hours each week. We chose three hours as a threshold for basic engagement because China limits adolescents’ gameplay time to three hours per week (Huang, 2021). Each participant was paid RMB14 (USD 2.19) for completing the survey. The questionnaire was originally written in English. Three bilingual speakers translated the questionnaire into Chinese through forward and backward translation to ensure its accuracy. Institutional ethics approval and participants’ consent were obtained before data collection.

To calculate the sample size, we carried out a power analysis using the pwr package in R (power = .80, r = .10, sig.level = .05). The result showed that a minimum of 782 participants were required to obtain sufficient statistical power. A total of 989 respondents completed our survey. To ensure scale validity, we employed an attention check by reversing items in the measures of perceived power and mutual dependence. People who chose the same positive/negative extreme answers for the reversed item and the remainder of the items in one construct failed the attention check and were removed from the survey. As a result, 207 cases were removed, yielding 782 responses. Overall, participants were aged between 18 and 55 years (M = 27.2, SD = 4.72), mostly possessed a bachelor’s degree (n = 685, 87.6%), and were comprised of 243 (31.1%) females and 533 (68.2%) males. Around half of the participants (45%) reported playing MMO games 4–7 hours a week. Also, 40.4% of participants reported having played MMO games for four to five years. Our surveys, data, R script, and data analyses are publicly available at: https://osf.io/nfkz3.

Measures

We first asked participants to recall the MMO game that they played most frequently in the past six months. We chose six months as the timeframe for two reasons. First, toxicity does not occur on a regular basis, so a short timeframe (e.g., one month) might not capture the behavior. Second, if we cast a wider net and focus on a longer timeframe (e.g., one year), the data may suffer from severe recall bias. Thus, we decided to limit the timespan to six months, following prior cyberbullying studies (e.g., Pabian & Vandebosch, 2014; Sourander et al., 2010). After recalling the game, participants were required to report mutual dependence, power, passion, toxicity, and prosocial behavior for the specific game they identified. All measures were rated on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree) unless otherwise stated.

Perceived mutual dependence

Six items from the scale developed by Depping et al. (2018) were used to measure perceived mutual dependence. Example items include “the actions of me and my teammates were strongly linked” (α = .8, M = 5.29, SD = 0.9).

Perceived power

Four adapted items from Gerpott et al. (2018) were employed to measure perceived power. Examples include “compared with my teammates, I had more impact on what happened in the game” (α = .77, M = 4.56, SD = 0.96).

Passion

Passion was measured through the scale developed by Vallerand (2003). Harmonious passion was measured with five items, such as “This game reflected the qualities I like about myself.” The construct showed acceptable reliability with the item “this game was in harmony with the other activities in my life” being dropped (α = .77, M = 5.27, SD = 0.91). Obsessive passion was measured with five items, such as “I could not live without this game” (α = .85, M = 3.24, SD = 1.10).

Toxicity

We used measures developed by Tang and Fox (2016) (e.g., “I said curse or swear words toward my teammates”). We also developed new items such as “I taunted my teammates” to capture toxicity in games in a more well-rounded manner. Ten items were rated on a 5-point Likert scale ranging from 1 (never) to 5 (always). The construct showed good validity and reliability after four items were dropped (α = .84, M = 1.72, SD = 0.58). The final construct targeted both verbal and behavioral actions.

Prosocial behaviors

We measured prosocial behaviors through measures developed by Shoshani et al. (2021) (e.g., “I shared resources with my teammates”) and Velez and Ewoldsen (2013) (e.g., “I gave my teammates’ suggestions to improve their play”). We also developed new items such as “I encouraged my teammates” to holistically capture prosocial behaviors in games. Participants responded to eight questions using a 5-point Likert scale (1 = never, 5 = always). The construct showed acceptable reliability and validity with five items being dropped (α = .73, M = 3.64, SD = 0.58). The final construct targeted only verbal actions.

Control variables

Demographics (i.e., age, gender, education background, and income), game experience, and perceived gameplay competitiveness were controlled for in this study. Game experience was measured through “How many hours did you spend playing MMO games each week in the past six months?” and “How many years have you been playing MMO games?” Since previous research suggests that competitive gameplay might influence people’s prosocial and toxic behaviors in games (Tan & Chen, 2022; Verheijen et al., 2019), we also controlled players’ perceived gameplay competitiveness through the item “I needed to compete with other teams” (1 = Strongly disagree, 7 = Strongly agree).

Analysis and results

We performed the analyses using R (Version 4.1.1). To ensure the validity of the measures, we performed confirmatory factor analyses (CFA) with robust maximum likelihood estimation (lavaan package). Based on common goodness-of-fit (GOF) criteria (Kline, 2015), we found that the validity of all measures was good. The psychometric properties of the study variables are presented in Table 1. Bivariate correlations (Hmisc package) between the latent variables are presented in Table 2.

We examined the hypotheses and RQ using structural equation modeling (SEM) (lavaan package). The estimation used a 1,000 bootstrapped resampling procedure to determine 95% confidence intervals for the indirect effects. Missing data were handled through maximum likelihood estimation. Drawing on goodness-of-fit (GOF) criteria (Kline, 2015), our overall model showed a good fit (χ2(543) = 1216.711, p < .001, CFI = 0.92, TLI = 0.90, SRMR = 0.050, RMSEA = 0.040). We also ran reversed models,² and two SEM models with prosocial and toxic behaviors as dependent variables separately. Models with prosocial behaviors (χ2(342) = 863.967, p < .001, CFI = 0.92, TLI = 0.90, SRMR = 0.054, RMSEA = 0.044) and toxicity (χ2(456) = 1045.248, p < .001, CFI = 0.92, TLI = 0.91, SRMR = 0.056, RMSEA = 0.041) showed a good fit. Figure 1 displays the pathway coefficients for the measurement model. Table 3 shows the indirect effect, retrieved at a 95% confidence interval, indicated in the two models.

For H1, our results showed that perceived mutual dependence was not related to prosocial behaviors (β = .03, SE = .04, p = .41) and toxicity (β = −.04, SE = .03, p = .27). Thus, H1 was not supported. For RQ1, the results showed that perceived power was positively related to prosocial behaviors (β = .24, SE = .04, p < .001) but not toxicity (β = −.01, SE = .03, p = .66).

For H2, the results showed that perceived mutual dependence was positively associated with harmonious passion (β = .29, p < .001), which in turn was positively associated with prosocial behaviors (β = .17, p < .001) and negatively associated with toxicity (β = −.17, p < .001). Harmonious passion fully mediated the link between perceived mutual dependence and prosocial behaviors (b = .05, SE = 0.01, 95% CI = 0.025–0.079) and toxicity (b = −.05, SE = 0.01, 95% CI = −0.072 to −0.030). Our results also showed that perceived power was positively associated with harmonious passion (β = .18, p < .001), which partially mediated the link between perceived power and prosocial behaviors (b = .03, SE = 0.01, 95% CI = 0.016–0.056), as well as between perceived power and toxicity (b = −.03, SE = 0.01, 95% CI = −.054 to −.019). Thus, H2 was supported.

For H3, our results showed that perceived mutual dependence (β = .14, p < .01) was positively associated with obsessive passion, which in turn was positively associated with toxicity (β = .14, p < .001) but not associated with prosocial behaviors (β = .02, p = .42). Obsessive passion fully mediated the link between perceived mutual dependence and toxicity (b = .02, SE = 0.01, 95% CI = 0.007–0.036) and did not mediate between perceived mutual dependence and prosocial behaviors (b = .003, SE = 0.004, 95% CI = −.004 to .013). Our results also showed that perceived power was positively associated with obsessive passion (β = .18, p < .001), which mediated the relationship between perceived power and toxicity (b = .03, SE = 0.01, 95% CI = 0.007–0.036), and did not mediate between perceived power and prosocial behaviors (b = .004, SE = 0.005, 95% CI = −0.006 to 0.016). Thus, H3a and H3c were supported.

We also classified the games based on whether they were cooperative (e.g., role-playing games) or competitive (e.g., battle arena games) and ran a post-hoc analysis. The result showed that there was no significant difference (F(12) = 11.828, p = .46). We ran the analysis in which a sum index rather than a mean was computed for prosocial and toxic behaviors. The results were in line with the above where H2, H3a, and H3c were supported and H1 and H3b were not supported.

Discussion

Understanding factors predicting toxic and prosocial behaviors in online games has attracted considerable attention. While previous research has primarily focused on personal and technological factors (Kordyaka et al., 2020; Kordyaka & Kruse, 2021; Shen et al., 2020; Shores et al., 2014), this study examines how interpersonal factors (i.e., perceived mutual dependence and power) are related to toxic and prosocial behaviors in games. The results show that perceived power is positively related to prosocial behaviors in MMO games and perceived mutual dependence and power are indirectly related to toxic and prosocial behaviors in MMO games through harmonious and obsessive passion. Notably, the results are consistent after taking into account gameplay competitiveness, a factor that has been found to influence players’ in-game behaviors (Verheijen et al., 2019). This study makes important contributions to the literature on toxicity, prosocial behavior, and game addiction. Below, we discuss these potential contributions in more detail.

Interpersonal dynamics and passion in MMO games

This study contributes to game studies by uncovering significant interpersonal factors associated with toxic and prosocial behaviors in MMO games. Previous studies primarily focused on individual factors (e.g., personality) and technological factors (e.g., game genres; De Simone, 2013; Kordyaka et al., 2020; Velez & Ewoldsen, 2013). This study goes beyond these commonly studied factors by examining the most prominent yet understudied characteristic of MMO games: interpersonal interaction (Cole & Griffiths, 2007). It shows that the more influence players feel that they can exert on teamwork compared to their teammates, the more likely they are to carry out prosocial behaviors. Though players’ perceptions of mutual dependence are not directly related to prosocial and toxic behaviors, they are indirectly correlated with these behaviors through passion.

This study also highlights passion as a mechanism through which perceived mutual dependence and power are associated with prosocial and toxic behaviors in MMO games. Previous studies on passion mainly focused on its intrapersonal outcomes, such as players’ affect, wellbeing, and time spent on games (Johnson et al., 2013; Przybylski et al., 2009; Puerta-Cortés et al., 2017). This study unveils the explanatory power of passion on interpersonal outcomes in games. One potential explanation for this is that players with harmonious passion play games without any attached contingencies (e.g., self-esteem; Vallerand, 2010), thus they are less likely to feel hurt and lose tempers when encountering bad performance. In contrast, players with obsessive passion hold gameplay close to their identity, and thus might rage in the face of bad performance due to identity violation.

In addition, our study shows that mutual dependence and power are positively associated with both harmonious and obsessive passion, which is in line with previous work showing that interpersonal factors in games may lead to increases in both passions for games (Johnson et al., 2021). One plausible explanation is that positive experiences in games increase the values players attach to these activities and will elevate their overall passion (i.e., both harmonious and obsessive passion) for games (Vallerand, 2010). However, perceived mutual dependence and power are more strongly related to harmonious passion than obsessive passion. This may be because a secured mutual dependence and power relieves players’ pursuit of the contingencies attached to games, such as social acceptance, thus players are more likely to develop harmonious passion. Future studies can further explore the factors that lead to the divergence of passions, such as personality traits (Dalpé et al., 2019) and autonomy support (Bonneville-Roussy et al., 2013).

Explicating toxicity and cyberbullying: the role of power dynamics

In addition to extending the game literature, this study also sheds light on the differences between toxicity and cyberbullying by highlighting their different power dynamics. Given the commonalities between toxicity and cyberbullying (e.g., anonymity and aggression), researchers tend to understand toxicity through the lens of cyberbullying (e.g., Kordyaka & Kruse, 2021; Shen et al., 2020). However, this study reveals notable differences between toxicity and cyberbullying. In general, the power imbalance is recognized as a core component of cyberbullying, with perpetrators being the stronger party and victims being the weaker ones (Kowalski et al., 2014). However, this study finds that players who perceived themselves as having more power carry out more prosocial behaviors and do not necessarily perform toxicity. This corroborates prior work suggesting that powerful people showcase more forgiveness and leveraged their power to help groups achieve goals (Karremans & Smith, 2010; Maner & Mead, 2010).

These different findings regarding power suggest a need to go beyond the cyberbullying literature to understand toxicity, especially the contextual and psychological factors presumably shaping the differences. On one hand, cyberbullying is not commonly examined in a collaborative context (Kowalski et al., 2014), whereas collaboration is a defining characteristic of MMO games (Cole & Griffiths, 2007), which potentially nudges powerful people to carry out prosocial behaviors. On the other hand, MMO games are different from general platforms in the way that they are characterized by inter-group competition, which heightens the necessity to cooperate. In this case, powerful people may place group success over their self-interests and perform acts that are desirable for the group (Maner & Mead, 2010).

Also, while previous research suggested that players with power were more likely to commit toxicity (Ballard & Welch, 2017; Shen et al., 2020; Shores et al., 2014), this study draws the opposite conclusion. This may be because, in previous research, power was measured through objective status (e.g., rankings; Ballard & Welch, 2017; Shen et al., 2020), whereas this study measured players’ perceptions of power. When people perceive more power in games, their sense of responsibility may increase and thus become more likely to help others (De Wit et al., 2017; Tost, 2015). While previous research suggests that outcome expectancies influence aggressive behaviors (Pornari & Wood, 2010), our study reveals that regardless of the evaluation of outcomes, the mere awareness of outcomes may shape players’ behavior in games. Explicating these postulations may be a promising endeavor for future research.

Gameplay as a double-edged sword

Another contribution of this study lies in its illumination of the intricate social implications of technology. While prior work on ICT has mainly focused on either positive or negative effects of technology usage (e.g., Peng et al., 2021; Qin et al., 2021), this study brings to light the dual effects of technology in the context of games. This finding suggests that when people perceive more mutual dependence and power in games, their pre-existing values and goals may work in harmony with games (i.e., harmonious passion), but, in the meantime, they may also develop an uncontrollable urge to play games (i.e., obsessive passion). In line with prior work that shows a strong link between harmonious passion and obsessive passion (Przybylski et al., 2009; Puerta-Cortés et al., 2017), this finding supports the contention that the influence of technology is “not confined to an either/or framework” (Weinstein, 2018). People experience both positive and negative effects of social technology, even from a single functional dimension (Sum et al., 2008; Weinstein, 2018). Acknowledging that technology is a double-edged sword may shatter the moral panics around technologies, which arise with the advent of almost every new technology, be it television, game, or social media (Orben, 2020).

Our findings also complicate the studies on Internet addiction. Previous studies have regarded Internet addiction as a consequence of obsessive passion (Lafreniere et al., 2009; Przybylski et al., 2009), while our study shows that obsessive passion coexists with harmonious passion, an indicator of positive experiences and behaviors. In this vein, coercively reducing the time players spend on games may wash out the potential positive effects of harmonious passion. As suggested by Vallerand (2010), the predominance of passion depends on the weight of elements such as autonomy support (Bonneville-Roussy et al., 2013) and relationship types (Perry et al., 2018). Prevention strategies for game addiction should attend to these factors to help people sustain a healthy relationship with technology while still obtaining enjoyment.

Practical implications

Our study also provides practical implications. First, the direct and indirect relationships between perceived mutual dependence and power, harmonious passion, and prosocial behaviors suggest that efforts aimed at creating a healthy and inclusive gaming environment should consider cultivating mutual dependence and power among players. Game designers can enhance players’ perceived mutual dependence and power through certain affordances in games, such as building complementary roles and abilities (Harris et al., 2016; Harris & Hancock, 2019), requiring information exchange (Kutlu et al., 2013), and allowing players to realize the greater impact they have on gameplay.

Second, the mediating role of passion suggests that practitioners need to be acutely cognizant of the effects of players’ relationships with games on toxic and prosocial behaviors. Because obsessive passion denotes an uncontrollable urge to play games (Vallerand et al., 2003), game designers should consider in-game interventions that help people strike a balance between play and non-play, such as rest systems (Rest, n.d.) that heuristically nudge players to take a break. In a similar vein, practitioners from other domains, such as social media, can also consider setting up interventions, possibly in the form of time setting (Hiniker et al., 2016), to help people regulate technology usage and mitigate the uncivil behaviors engendered by obsessive passion.

Limitations and future research

Despite the significance of the findings, several limitations should be noted. First, the major limitation of this study is its lack of causality. Due to the cross-sectional nature of the survey, no causal relationships could be inferred from this research. It is possible that players’ behaviors in games influence their passion for games or players’ passion for games influences their perceived mutual dependence and power in games. Further longitudinal or experimental studies are needed to scrutinize the causality among these variables.

Second, this study uses self-report measures, which may be subject to recall bias and social desirability bias, especially when people’s negative behaviors are measured. Though social desirability bias may be lessened in the context of games because players assume toxicity to be acceptable and normal (Beres et al., 2021), future studies should consider more direct measurements such as game log or behavioral data (Shen et al., 2020).

Third, the scale of prosocial behaviors in this study only targets verbal actions and its reliability is borderline. Future research should consider developing novel scales to capture behavioral actions of prosocial behavior and increase the reliability of the scale of prosocial behaviors in games. In addition, the receivers of toxicity and prosocial behaviors in this study are limited to teammates. Though these behaviors are more common among teammates (Adinolf & Turkay, 2018; Neto et al., 2017; Shen et al., 2020), future research can explore how players from different teams interact with each other, which may provide more insights into fostering vigorous gaming environments.

Last but not least, mutual dependence and power are two key dimensions of interest in this study. Future studies are encouraged to explore other dimensions of interdependence, such as conflict of interest, to provide more insights into players’ interdependence in games.

Data Availability

The data underlying this article are available in OSF, at https://osf.io/nfkz3.

Funding statement

This research is supported by the National University of Singapore Start-up Grant (R-124-000-128-133). The authors are grateful for the helpful feedback from the reviewers and editors.

Conflicts of interest: None declared.

Notes

References