We ran an a priori power analyses using the GPower software. With 3 predictors in a regression model we would have 90% power to detect an effect size of f² = 0.05 or larger with at least 288 participants. Using Amazon Mechanical Turk (Mturk), we recruited 310 American video gamers. We excluded participants who failed attention checks or did not complete the survey. After exclusions we had 304 participants (gender: 146 males, 155 females, 3 non-binary; age: 19–77; ethnicity: 22 Asian/Pacific Islander, 29 Black, 3 American Indian, 16 Hispanic, 219 White, 7 Multiracial, 6 Other, 1 Non-response; education: 2 Less than a high school degree, 37 High school degree or equivalent, 57 Some college but no degree, 46 Associate degree, 120 Bachelor degree, 42 Graduate degree).

After providing informed consent, participants indicated whether they played video games. Participants who responded “yes” proceeded to complete all the scales below in random order, with the exception that right-wing identity, white nationalist identity and demographic questions were measured after the other scales to avoid the potential priming of these identities.

Identity fusion was measured with three different, 3-item scales targeting fusion with gamer identity, fusion with gaming culture, and fusion with other gamers. The three items in each fusion scale were adapted from the 7-item verbal identity fusion scale (Gómez et al., 2011b). All three items were measured on 1-7 Likert scales ranging from “Completely Disagree” to “Completely Agree.” An example item from this scale is “I make gaming culture strong.” Total scores on the three fusion scales were highly correlated with each other and mapped onto one factor. Given this strong overlap, we focused our analyses on the scale that had the highest internal consistency: identity fusion with gaming culture (M = 3.14, SD = 1.80, α = 0.95).

A key correlate of identity fusion is the willingness to fight/die for the group (Swann et al., 2009). Participants were asked to respond to a single item “I would fight someone insulting or making fun of gaming culture” on a 1-7 Likert scale ranging from “Completely Disagree” to “Completely Agree” (M = 1.63, SD = 1.02, α = 0.94).

The Dark Triad of personality traits (Machiavellianism, narcissism, and psychopathy) has been linked to antisocial behavior in games (Tang et al., 2020) and the endorsement of extremist beliefs (Pavlović and Wertag, 2021). We measured each of the dark triad traits using a 4-item version of each trait utilized by related research on extremist populations (Forscher and Kteily, 2020). All items were measured on 1 - 7 Likert scales ranging from “Strongly Disagree” to “Strongly Agree.” An example item for Machiavellianism (M = 3.94, SD = 1.12, α = 0.63) is “I like to use clever manipulation to get my way.” An example item for Narcissism (M = 3.10, SD = 1.39, α = 0.78) is “I know that I am special because everyone keeps telling me so.” An example item for Psychopathy (M = 2.20, SD = 1.12, α = 0.72) is “Payback needs to be quick and nasty.” Although the internal reliability of the four Machiavellianism items was less than an ideal of α ≧ 0.70, we still included this measure as it is a key part of the Dark Triad and it had been used successfully in previous work (Forscher and Kteily, 2020).

Social dominance is a trait associated with far-right extremism (Bai, 2019) and antisocial behavior in gaming spaces, including sexism (Fox and Tang, 2014) and harassment (Tang et al., 2020). We included eight items to measure social dominance orientation adapted from the original 12-item version of the scale from Pratto et al. (1994). All items were measured on 1-7 Likert scales ranging from “Strongly Disagree” to “Strongly Agree” (M = 2.54, SD = 1.52, α = 0.93). An example item from this scale is “Some groups of people are simply inferior to other groups.”

Authoritarianism is a belief system commonly associated with extremist action (Wintrobe, 2006). As such, we included six items designed to tap right-wing authoritarianism (Forscher and Kteily, 2020). All items were measured on 1-7 Likert scales ranging from “Strongly Disagree” to “Strongly Agree” (M = 3.38, SD = 1.45, α = 0.83). An example item is “Obedience and respect for authority are the most important virtues children should learn.”

Right-wing ideologies have been specifically associated with hate and harassment in gaming spaces, specifically the #GamerGate hate campaigns of 2014 (Aghazadeh et al., 2018). To assess the relationship between gamer fusion and right-wing identity, we created a single-item measure. First, we gave participants a brief text defining what we meant by right-wing identity: “The right-wing nationalist movement is characterized by a rejection of mainstream politics and media, strong identification with one's own nation and support for its interests, and the belief that your own country is better than all others without question or doubt. Then, participants were asked to indicate their level of agreement with the statement “I identify with the right-wing nationalist movement.” Participants indicated their level of agreement to this statement using a 1-7 Likert scale ranging from “Strongly Disagree” to “Strongly Agree” (M = 2.48, SD = 1.94).

While extremist beliefs can take many forms, White Nationalist ideologies have been found to be particularly prevalent in gaming spaces (Anti-Defamation League, 2019; Institute for Strategic Dialogue, 2021). To assess the links between fusion and white nationalism, we created a single-item measure of identity with this movement. We first gave participants a brief text defining what we meant by white nationalism (“The alt-right movement is a right-wing, primarily online political movement or grouping based in the U.S. whose members reject mainstream conservative politics and espouse extremist beliefs and policies typically centered on ideas of white nationalism.”). Then participants were asked to indicate their level of agreement with the statement “I identify with the alt-right movement.” Participants indicated their level of agreement to this statement using a 1-7 Likert-scale from “Strongly Disagree” to “Strongly Agree” (M = 1.98, SD = 1.59). Note that although we referred to this identity as “alt-right identity,” we later discovered this is a common term used by white nationalists to make their identity and ideology seem more palatable (Gallaher, 2020). As the definition presented to participants was explicitly framed as an endorsement of white nationalism, we will refer to it as that throughout the rest of the paper.

Sexism is a cornerstone of “toxic gamer cultures” (Consalvo, 2012) and white nationalist ideologies (Forscher and Kteily, 2020). To assess the links between fusion and sexism, we used a three-item measure of both benevolent and hostile sexism taken from Forscher and Kteily (2020) that measures sentiments on 1-7 Likert scales from “Strongly Disagree” to “Strongly Agree.” Benevolent sexism (M = 4.15, SD = 1.70, α = 0.82) is characterized by a paternalistic, condescending view toward women. An example item of benevolent sexism is “Women should be cherished and protected by men.” Hostile sexism (M = 2.84, SD = 1.53, α = 0.83) is characterized by general dislike and disregard of women. An example item of hostile sexism is “Once a woman gets a man to commit to her, she usually tries to put him on a tight leash.”

Racism is another cornerstone of “toxic gamer cultures” (Gray, 2012) and underpins extremist ideologies (Michael, 2003; Fuchs, 2016). We measured racism with two scales adapted from the motivations to express and inhibit bias scales used by Forscher and Kteily (2020). We changed the target of the items from “Black people” to “minorities” to target general racism beyond antipathy toward Blacks. Both measures used 1-7 Likert scales ranging from “Strongly Disagree” to “Strongly Agree.” Extrinsic Racism (M = 2.42, SD = 1.31, α = 0.77) was measured using four items drawn from the external motivation to express bias subscale. Intrinsic Racism (M = 2.06, SD = 1.27, α = 0.78) was measured using four items drawn from the internal motivation to express bias subscale from the same measure.

To examine connections between fusion, extremist ideologies, and extreme pro group behavior, we also included a behavioral measure of recent aggressive behaviors (within the last month) using a six-item self-report scale from Forscher and Kteily (2020). All items were measured on a 1-7 Likert scale ranging from “Not at all” to “Frequently” (M = 1.74, SD = 1.12, α = 0.90). Participants were asked to report their aggressive behaviors regardless of whether they occurred online or offline. An example item from this measure is “Made a statement because others find it offensive.”

QAnon is a conspiracy theory characterized by the belief that the world is run by a sinister cabal of child-molesting devil-worshiping elites and that Donald Trump is the only one who can thwart the designs of this sinister group. Belief in QAnon was included as an outcome of interest as it has been found to be strongly associated with white nationalist ideologies and beliefs (Anwar et al., 2021). To measure one's belief in QAnon conspiracies, we developed a seven-item measure. All items were measured on Likert scales ranging from “Completely Disagree” to “Completely Agree” (M = 1.89, SD = 1.18, α = 0.92). An example item from this scale was “Donald Trump is secretly working to overthrow the power of child molesters and devil worshipers who run this country.” Given that we developed this scale ourselves, we ran an exploratory factor analysis on the seven items. We found that all 7-items mapped strongly to 1 factor (factor loadings >0.60). The strong loadings to a single factor combined with the high Cronbach's alpha (α = 0.92) suggests that this scale works well as a measure of a single construct.

To determine whether experience with gaming moderated our findings, we assessed three relevant variables. Participants were asked to report how many hours per week they currently played video games. They were also asked how many years they have been playing video games. These two measures were included to determine whether the relationships between fusion and the various outcomes could be related to prolonged exposure to the gaming community in the short-term (i.e., weekly play time) or long-term (i.e., number of years playing games). Responses to both of these open response questions were manually converted to numerical values to create two continuous measures where larger numbers indicated more weekly game play time (M = 9.91, SD = 10.43) or more years playing video games (M = 20.14, SD = 10.15), respectively. Finally, participants indicated which game genres they played the most in a multiple-choice question. Participants chose one option from a list of genres, including sandbox, real-time strategy, shooters, multiplayer online battle arena, role-playing, simulation and sports, puzzlers and party games, action-adventure, survival and horror, platformer, and Other (fill-in-the-blank). We measured this categorical variable to assess whether relationships between identity fusion and the outcome measures were limited to, or magnified within, players of a specific genre.

Participants were also asked to indicate their gender, age, ethnicity, nationality, education level, political identity, political orientation, religion, and religiosity.

We recruited 304 American participants from Amazon Mechanical Turk who indicated that they play video games. We excluded participants who failed attention checks or failed to complete the survey. After exclusions we had 294 participants (gender: 132 males, 161 females, 1 non-binary; age: 19-74; ethnicity: 27 Asian/Pacific Islander, 25 Black, 1 American Indian, 19 Hispanic, 210 White, 9 Multiracial, 3 Other; education: 3 Less than high school degree, 42 High school degree or equivalent, 69 Some college but no degree, 48 Associate degree, 98 Bachelor degree, 34 Graduate degree).

Study 2's procedure was identical to Study 1 except for the addition of the new measures. Participants completed a consent and screener question, then completed most measures in randomized order followed by the right-wing and white nationalist identity questions, then demographic questions and finally the debriefing form.

Study 2 included all the measures from Study 1 that demonstrated a significant relationship between fusion and gaming culture. We also added several additional measures to assess individual differences that may impact the nature of the relationships between identity fusion and extreme outcomes: insecure attachment, loneliness, and social motivations for play. Finally, we measured two aspects of play modality (whether play was online/offline and social/alone).

To measure insecure attachment styles we used the 12-item Experiences in Close Relationship scale created by Wei et al. (2007). All items were measured on a 1-6 Likert scale ranging from “Strongly Disagree” to “Strongly Agree.” The scale has 6 items tapping Avoidant Attachment Style (M = 2.38, SD = 1.14, α = 0.87) and 6 items measuring Anxious Attachment Style (M = 2.95, SD = 1.21, α = 0.83). An example of an Avoidant item is “I try to avoid getting too close to my partner”; a sample Anxious item is “I need a lot of reassurance that I am loved by my partner.” To avoid the issue of whether participants were in an active romantic relationship or not, we asked them to respond based on their likely typical experiences in romantic relationships.

To assess whether players were seeking out online games for emotional comfort, we included 6 items to measure participants' emotional motivations for playing video games (Kowert, 2015). A list of emotions was preceded by the statement “I play video games when I feel…” All items were measured on a 1-5 Likert scale ranging from “Strongly Disagree” to “Strongly Agree “The six emotions were: Stressed, Anxious, Sad, Lonely, Happy, and Excited. This measure was used because it differentiates between positive gameplay motivations (happy, excited) and negative gameplay motivations (stressed, anxious, sad, lonely). We averaged the first 2 items to create a composite measure of Positive Gameplay Motivations (M = 2.98, SD = 1.05, α = 0.88) and the last 4 items to create a composite measure of Negative Gameplay Motivations (M = 3.75, SD = 0.91, α = 0.84).

In addition, we included a measure of loneliness (ULS-8; Hays and DiMatteo, 1987) to determine the degree to which individual gamers report feelings of loneliness in their daily lives. All 8 items were measured on a 1-6 Likert scale ranging from “Strongly Disagree” to “Strongly Agree” (M = 2.75, SD = 1.22, α = 0.88). An example loneliness item is “There is no one I can turn to.”

To gain insight into the social environments the players are engaging in, we created a single-item measure of play modality to tap the degree to which participants played games offline vs. online (“Which statement best describes your play style?”). The scale was measured on a 1-5 Likert scale ranging from “Offline Exclusively” to “Online Exclusively” (M = 2.98, SD = 1.18).

We created a single-item measure of gameplay companions to tap the degree to which participants played games by themselves vs. with other people (“Do you primarily play alone or with others?”). The scale was measured on a 1-5 Likert scale ranging from “Exclusively Alone” to “Exclusively with Others” (M = 2.46, SD = 1.05).

We aimed to recruit at least 300 American participants from Amazon Mechanical Turk who indicated that they play video games and played either Minecraft or Call of Duty for at least a few hours per week. These games were chosen as contrasting gaming spaces as Call of Duty is a first-person shooter battlefield game known to have high levels of community toxicity (Anti-Defamation League, 2019) whereas Minecraft is a sandbox video game known to be highly social and cooperative (Wald, 2020). We collected data from 338 Minecraft gamers and 327 Call of Duty players. We excluded participants who failed attention checks or failed to complete the survey. After exclusions we had 330 participants who play Minecraft (gender: 163 males, 160 females, 7 non-binary; age: 18–80; ethnicity: 20 Asian/Pacific Islander, 25 Black, 1 American Indian, 33 Hispanic, 233 White, 14 Multiracial, 4 Other; education: 4 Less than high school degree, 35 High school degree or equivalent, 79 Some college but no degree, 43 Associate degree, 118 Bachelor degree, 51 Graduate degree) and 315 participants who play Call of Duty (gender: 184 males, 129 females, 2 non-binary; age: 19–80; ethnicity: 18 Asian/Pacific Islander, 38 Black, 2 American Indian, 37 Hispanic, 207 White, 7 Multiracial, 6 Other; education: 1 Less than high school degree, 42 High school degree or equivalent, 74 Some college but no degree, 47 Associate degree, 110 Bachelor degree, 41 Graduate degree).

The procedure for Study 3 was virtually identical to the previous two studies except for the addition of the new measures as well as a new screener question. After consenting to take the survey, participants were asked the initial screening question “Do you play video games?” followed by an additional screening question asking, “Do you play any of the following games for at least a few hours a week?.” They were shown a list of similar games and were required to select Minecraft or Call of Duty (depending on which version of the survey they were taking) to be able to proceed. Then they completed most measures in randomized order followed by the right-wing and white nationalist identity questions, then demographic questions and finally the debriefing form.

We included all the measures from Studies 1 and 2 that were uniquely predicted by fusion with gaming culture in multiple predictor models controlling for right-wing and white nationalist identity (fight/die for gaming culture, Dark Triad personality traits, hostile sexism, extrinsic racism, recent aggressive behaviors). In Study 3 we also included a measure of identity fusion with the game itself (i.e., fusion with Call of Duty or fusion with Minecraft), in-game toxicity, game competitiveness, and prosocial measures (e.g., life satisfaction, self-esteem, online bonding, and relatedness). These new measures are discussed in more detail below. Overall means and standard deviations are given below but see Table 4 for descriptive statistics separated by gaming community (Minecraft and Call of Duty).

To measure whether fusion with one's game of choice (Minecraft or Call of Duty) had a unique impact different from fusion with gaming culture, we added a measure of fusion with game using the same 3-item fusion scale but targeting Minecraft or Call of Duty (M = 3.68, SD = 1.69, α = 0.91). All three items were measured on a 1-7 Likert scales ranging from “Completely Disagree” to “Completely Agree.” An example item from this scale is “I make [Minecraft/Call of Duty] strong.”

To determine whether gamers who played Minecraft or Call of Duty perceived their gaming community as socially toxic we included a measure of in-game toxicity (Depping et al., 2018). This measure was included to quantify any differences in the nature of the social communities of these games. All items were measured on a 1-5 Likert scale from “Strongly Disagree” to “Strongly Agree.” The scale has 8 items preceded by the statement “The people I play with are sometimes...” followed by the following options: angry, offensive, mean, good-natured, sympathetic, friendly, hurtful, and toxic. The three positive items (good-natured, sympathetic, friendly) were reverse coded then all 8 items were averaged (M = 2.55, SD = 0.85, α = 0.88).

To determine whether gamers who played Minecraft or Call of Duty perceived their gaming community as more competitive or cooperative, we included a single 1-7 Likert-type item where participants responded to the statement “I would say that the gaming environment of [Minecraft/Call of Duty] is...” on a scale from “Cooperative” to “Competitive” (M = 4.39, SD = 1.94).

Life satisfaction was measured using the Riverside Life Satisfaction Scale (Margolis et al., 2019). We included this measure to see if fusion with games/gaming culture predicted overall satisfaction with life. All items were measured on a 1-7 Likert scale from “Strongly Disagree” to “Strongly Agree” (M = 4.33, SD = 1.43, α = 0.89). An example item is “I like how my life is going.”

Self-esteem was measured with the Rosenberg Self-Esteem Scale (Rosenberg, 1965). We included this measure to see if fusion with games/gaming culture were related to self-esteem. All items were measured on a 1-7 Likert scale ranging from “Strongly Disagree” to “Strongly Agree” (M = 5.11, SD = 1.34, α = 0.93). An example item is “On the whole, I am satisfied with myself.”

Bonding social capital is indicative of close, tight-knit, and long-lasting friendship bonds (Putnam, 2000). To measure players' degree of online bonding social capital, we used the bonding subscale of the social capital scale developed by Williams (2006b). We included this measure to see if fusion with games/gaming culture predicted a perception of social bonding among gamers. All ten items were measured on a 1-7 Likert scale ranging from “Strongly Disagree” to “Strongly Agree” (M = 3.64, SD = 1.54, α = 0.93). An example item was “There are several people online I trust to help solve my problems.”

As social connectedness is a critical component to extremist action (Horgan, 2008), we included a measure of relatedness to see if fusion with games/gaming culture predicted a sense of relatedness to other players. This was assessed using the relatedness items of the Basic Psychological Need Satisfaction Scale (Gagne, 2003). All seven items were measured on a 1-7 Likert-type scale ranging from “Not True” to “Very True,” with a midpoint value at 4-Somewhat True (M = 5.02, SD = 1.03, α = 0.81). An example item is “I get along with people I come into contact with.”