Progressing from play to political preferences

Most people lack a cohesive political belief system (Converse, Reference Converse2006; Sears and Funk, Reference Sears, Funk and Zanna1991) but adopt distinct policy positions when personally relevant issues are involved. Membership in “extracurricular” activities – e.g., faith-based groups, labor unions, ethnic minority organizations, and neighborhood associations – is one of the central explanations for political engagement and attitude formation (Coleman, Reference Coleman1988; Djupe and Grant, Reference Djupe and Grant2001; Jacobs and Tillie, Reference Jacobs, Tillie, Azzi, Chryssochoou, Klandermans and Simon2010; Kerrissey and Schofer, Reference Kerrissey and Schofer2013; Lin, Reference Lin2001; Pattie and Johnston, Reference Pattie and Johnston2000). Talò et al. (Reference Talò, Mannarini and Rochira2014) show that a strong sense of community, derived from membership in these “extracurricular” groups, promotes members’ political engagement. Similarly, Fagan et al. (Reference Fagan, McGee and Thomas2019) draw explicit connections between parties and civil society groups, showing that political parties frequently utilize close ties to pull interest group members into political activities.

The traditional forms of sociopolitical engagement dominate the academic discourse due to their apparent connection to politics. Considerably less attention has been directed toward studying the role of leisure-based identities in shaping political behavior, despite Putnam’s (Reference Putnam2000) note about the role of bowling leagues in fostering social capital. Hence, much of this work is still exploratory and unsettled regarding the effects’ exact nature and direction (see Ibsen et al., Reference Ibsen, Elmose-Østerlund, Feiler, Breuer, Seippel, Van der Roest and Scheerder2019 and Van Ingen and Van Eijck, Reference Van Ingen and Van Eijck2009 for competing arguments). This is unfortunate. After all, people internalize the preferences and behaviors of their peer groups and the consumed media, and leisure activities and hobbies are essential sources of informal community building (Riesman et al., Reference Riesman, Glazer, Denney and Gitlin2001; Riley et al., Reference Riley, Griffin, Morey and Demetriou2013). Furthermore, leisure is a universal phenomenon, ubiquitous to all cultures and societies (Chick, Reference Chick1998). My research aims to contribute to this debate by exploring how engagement in the hobby community potentially shapes political behaviors, even in subtle ways.

Hobby and fan culture attachments influence this process through “fan citizenship”Footnote ² (Hinck, Reference Hinck2019), especially among young people (Jenkins et al., Reference Jenkins, Shresthova, Gamber-Thompson, Kligler-Vilenchik and Zimmerman2016; Riley et al., Reference Riley, Griffin, Morey and Demetriou2013).Footnote ³ Fan citizenship manifests across a wide range of hobby identities, and although it is primarily performative (Hinck, Reference Hinck2019, p. 163), evidence of fans’ deep and active engagement exists. Dessewffy and Mezei (Reference Dessewffy and Mezei2020) illustrate that Harry Potter fans are remarkably interested in politics and politically active, even in illiberal states such as Hungary, and participation in interest-driven online activities appears positively related to online and offline political participation (Kahne and Bowyer, Reference Kahne and Bowyer2018). Hence, we must apply our knowledge about political socialization to these new environments.

Research on the interaction between political behavior and sports – especially soccer – fan identity is the most advanced. Alrababa’h et al. (Reference Alrababa’h, Marble, Mousa and Siegel2021) show that soccer fans modified their perceptions of religious minority groups based on favorable experiences with their favorite players. Rosenzweig and Zhou (Reference Rosenzweig and Zhou2021) connect soccer fandom, nationalism, and attitudes toward refugees. Bertoli (Reference Bertoli2017) highlights the link between nationalism fanned by a soccer World Cup and interstate conflict. Soccer hooligans – a group whose identity is strongly tied to their chosen club – influenced events leading up to and during the War in Yugoslavia (Mills, Reference Mills2018), and the Arab Spring protests in Egypt (Zirin, Reference Zirin2013).

I focus on the video game hobby identity for several reasons. One, gaming is one of the most ubiquitous pastimes, and the size of the video game market has long since eclipsed both the film and music industries (Entertainment Software Association, 2022; Morris, Reference Morris2016; Nath, Reference Nath2016). Two, video games have increasingly been used as platforms for political speech and campaigning (Bogost, Reference Bogost2006; Otenyo, Reference Otenyo and Hendricks2010; Steiner, Reference Steiner2020). Three, a burgeoning literature exists on the relationship between playing video games and political engagement. Some scholars suggest gamers become more prosocial due to their pastime (Molyneux et al., Reference Molyneux, Vasudevan and Gil de Zúñiga2015; Steinkuehler and Williams, Reference Steinkuehler and Williams2006), while others are less optimistic (Bacovsky, Reference Bacovsky2021b). Meanwhile, Zhao and Wu (Reference Zhao and Wu2021) demonstrate that Chinese gamers are politically active and efficacious, and Bacovsky (Reference Bacovsky2021a) suggests that Swedish gamers show signs of being a “primordial” issue public with political preferences different from non-gamers.Footnote ⁴ Finally, a better understanding of the political attitudes of the gaming community has normative implications due to the increasing incidence of politically extreme actors recruiting followers in the context of the gaming spaces (Al-Rawi, Reference Al-Rawi2018; Koehler et al., Reference Koehler, Fiebig and Jugl2022; Lakomy, Reference Lakomy2019; Massanari, Reference Massanari2017).

If gamers really are an emerging issue public with distinct patterns of political behaviors and preferences (Bacovsky, Reference Bacovsky2021a; Licari, Reference Licari2020), news and events that merge their hobby and politics should inevitably pique their level of political engagement. This expectation rests on political psychology research. Affect-free hot cognition in the context of political information processing is exceedingly rare and even apolitical primes may significantly impact political evaluations (Taber and Young, Reference Taber, Young, Huddy, Sears and Levy2013, pp. 538–540). Once a prime activates a specific concept in a person’s mind, this concept will influence all downstream information processing and subsequent decision-making (Förster and Liberman, Reference Förster, Liberman, Higgins and Kruglanski2007). Research on football fans’ shows that mixing sports fandom and political primes can shift attitudes toward marriage equality (Harrison and Michelson, Reference Harrison and Michelson2016). In the gaming realm, Shackford and Ekins (Reference Shackford and Ekins2014) report results of an opinion survey in which gamers displayed a high degree of suspiciousness of anyone who interferes with how they live their lives and enjoy their hobbies. Therefore, I propose and test the following hypothesis:

Data and methodology

The data for this study were collected via the Lucid Marketplace and Qualtrics platforms. A total of 1,175 research participants were recruited via the Lucid Marketplace platform in early February 2020. Built-in Lucid quota tools were utilized to produce a sample that was as close as possible to a representative sample of the U.S. public.Footnote ⁵ The sample size of roughly 1,000 respondents was chosen since this is the gold standard in survey-based research that aims to produce results that are generalizable to the wider population.Footnote ⁶ The sample resembles the U.S. population regarding age, gender, ethnicity, education attainment, employment status, and annual income. This matches the observation of Coppock and McClellan (Reference Coppock and McClellan2019) about the properties of the Lucid Marketplace sample. My sample departs from the national averages in partisanship and ideological leaning measures, where it leans more liberal and Democratic.

Research participants were randomly assigned one of two vignettes (see Figure 1). Both vignettes were based on a moderately altered version of a GamesInsdustry.biz article (Sinclair Reference Sinclair2019). The article, focusing on video game goods impacted by the tariffs placed on China, was chosen to represent a piece of media that the users (especially gamers) might randomly encounter while browsing the Internet. The specific goods in question represent the “most likely” research scenario under which I will most likely see the hypothesized relationship (Levy, Reference Levy2008). In future iterations of this project, more “neutral” goods will be used to explore the limits of the theorized relationship. The respondents were then given three manipulation check questions to ensure that research participants received the treatment. Footnote ⁷

Figure 1.

Experimental manipulations.

To avoid the danger of the results being skewed by bots that populate online survey environments (Dennis et al., Reference Dennis, Goodson and Pearson2019) or by inattentive survey-takers (Abbey and Meloy, Reference Abbey and Meloy2017), I include two additional “safeguard” questions that help facilitate response quality (see Online Appendix 1, pages A1.4 and A1.7). Then, I calculate a response quality index, which is a mean index derived from the correctness of the answer to the manipulation check and safeguard check questions. The value of this index in the whole sample was 0.760 (s.d. = 0.009, N = 1,124). I then filtered out respondents with response index of less than 0.600, which resulted in the working N of 898 that was used as the sample for analysis reported in this manuscript.Footnote ⁸

Independent variables

I utilize a four-question battery to measure respondents’ gamer identity based on a novel approach stemming from social identity research (Bačovský, Reference Bačovský2024). Table 1 presents the survey questions, and Figure 2 illustrates the distribution of respondents’ answers. Next, I utilize the three variables coded on the 4-point scale to construct a mean index of gaming identity (Cronbach’s α = 0.940). The mean index value for the control group is 2.085 (s.d. = 0.896, N = 443), and the treatment group records 2.049 (s.d. = 0.949, N = 422). As an additional measure of gamer identity, I include time spent playing video games as well.

Table 1.

Gamer identity measures

Figure 2.

Gamer identity measures.

Dependent variablesFootnote ⁹

The relevant dependent variable in this study is the interest in U.S. foreign trade with China, where most of the video game hardware is manufactured. Therefore, the topic can become salient for gamers. A series of three prompts measures this concept. The first prompt was presented before the experimental manipulation and instructed respondents to do the following: “On a scale from 1 to 10, please indicate your general interest in the U.S. foreign trade policy.” After reading the randomly assigned vignette, the respondents were asked the remaining two prompts. The question stem was: “Think about the article you just read. On a 10-point scale, where 1 = Strongly Disinterested and 10 = Strongly Interested, please indicate ….” Then the respondents were queried about “[their] general interest in the U.S. foreign trade policy” and “[their] interest in the specific U.S. foreign trade policies discussed in the article.” Figure 3 visualizes the distribution of the respondents answers.

Figure 3.

Interest in the U.S. Foreign Trade Policy, breakdown.

ResultsFootnote ¹⁰

The analyses presented herein support the hypothesis that gamers who received the treatment manipulation would be more interested in U.S. foreign trade policy than gamers who received the control manipulation and non-gamers.Footnote ¹¹

General interest in the U.S. foreign trade policy

The average pre-manipulation general interest score for gamers is 7.043 (s.d. = 2.314, N = 300), while the average post-manipulation general interest score for gamers is 7.168 (s.d. = 2.176, N=304). The average pre-manipulation score for non-gamers equals 5.958 (s.d. = 2.545, N = 571), and the post-manipulation score is 5.952 (s.d. = 2.500, N = 582). Focusing only on treated non-gamers, we see that the average post-treatment general interest equals 6.202 (s.d. = 2.380, N = 297), a figure that drops to 5.691 (s.d. = 2.598, N = 285) among treated gamers. Control group gamers recorded an average post-treatment general interest of 7.089 (s.d. = 2.343, N = 146), while the average general interest among the treatment group gamers increased to 7.241 (s.d. = 2.014, N = 158). Table 2, in conjunction with Figure 4, demonstrates that treated gamers became more interested in U.S. foreign policy in general than gamers who were sorted into the control group. Among non-gamers, the direction of the effect was the opposite.

Table 2.

Receiving the treatment vignette increases respondent’s general interest in the U.S. Foreign Trade Policy among gamers but not among non-gamers. Models demonstrate robustness to the operationalization of gamer identity

Note: Ordinary Least Squares regression estimates.

Figure 4.

Predicted general interest in U.S. foreign trade policy based on experimental treatment status and moderated by gamer identity measures. Gamers who received the treatment are more interested than control group gamers and non-gamers.

Note: Visualization based on the linear Ordinary Least Squares model results from Table 2.

Shaded regions represent 95% confidence intervals.

Interest in the U.S. foreign trade policy discussed in the article

Non-gamers record a mean specific interest score of 5.529 (s.d. = 2.609, N = 573). The average interest score in the particular policy for gamers equals 7.207 (s.d. = 2.137, N = 299). When looking only at treated gamers, this figure increases to 7.529 (s.d. = 1.900, N = 157). Meanwhile, control group gamers only recorded an interest value of 6.852 (s.d. = 2.328, N = 142). Treated non-gamers record a mean specific interest score of 5.691 (s.d. = 2.598, N = 285), and control group non-gamers record 6.202 (s.d. = 2.381, N = 297). In brief, gamers are more interested in the article than non-gamers, treated gamers are more interested than control group gamers, and treated non-gamers are less interested in the topic than control group non-gamers.

Regardless of the gamer identity measure, the effects lean in the expected direction (see Table 3 and Figure 5). Treated gamers were likelier to say they were interested in the policy. Conversely, control group gamers were less interested in the policy. Using the “importance” operationalization of the independent variable translates to the predicted increase in interest of 1.345 (13.45 percentage points). Moreover, the effect of receiving the treatment version of the vignette among those who do not see themselves as a gamer is equally strong. Seeing an article that discusses the impact of the trade war on the video game industry decreases the respondent’s likelihood of saying that they are interested in the U.S. foreign trade policy. As a result of receiving the treatment, non-gamers’ estimated interest score dropped by 1.477 (or 14.77 percentage points).

Table 3.

Receiving the treatment vignette increases respondent’s interest in the specific U.S. Foreign Trade Policy discussed in the vignette among gamers but not among non-gamers. Models demonstrate robustness to the operationalization of gamer identity

Note: Ordinary Least Squares regression estimates.

Figure 5.

Predicted interest in the U.S. foreign trade policy mentioned in the vignette based on experimental treatment status and moderated by gamer identity measures. Gamers who received the treatment are more interested than control group gamers and non-gamers.

Note: Visualization based on the linear Ordinary Least Squares model results from Table 3.

Shaded regions represent 95% confidence intervals.

The gaming-foreign trade frame triggers a greater degree of interest in the topic among gamers but not among non-gamers. Conditional treatment effect plots presented as Figures A3.7 and A3.8 in the online appendix further substantiate these results. As noted, gamers have previously been targeted by political advertising on social media and in the context of the games themselves (Carras, Reference Carras2020; Gallucci, Reference Gallucci2020; Otenyo, Reference Otenyo and Hendricks2010; Steiner, Reference Steiner2020). Political entrepreneurs who engage in targeted advertising campaigns to reach gamers, sports fans, and other hobbyists may have the right idea. It appears that committed hobbyists can isolate policy issues from the vignettes and connect them to gaming, which sparks their interest in those topics. However, more work must be done to uncover whether interest translates into tangible forms of political engagement, such as voting.

One somewhat unexpected finding from this set of results concerns the negative effect of treatment manipulation on non-gamers. Although I expected that non-gamers would be the least interested, I did not hypothesize that their interest would decrease. I suspect that non-gamers who received the treatment manipulation interpreted the impact of the U.S.-China trade war as affecting only the gaming industry, lowering the issue’s personal relevance. I am reluctant to go further in this line of reasoning at the risk of overindulging in post-hoc theorizing. However, these results present the exciting possibility that politicians who cannot correctly isolate the targets for their leisure-laden advertising might harm their appeal instead of boosting it. I invite other scholars to explore this intuition further.