# Causal effect of video gaming on mental well-being in Japan 2020–2022 Received: 22 June 2023 Accepted: 5 July 2024 Published online: xx xx xxxx Check for updates Hiroyuki Egami  1,2 , Md. Shafiur Rahman  3,4, Tsuyoshi Yamamoto  5 , Chihiro Egami  6 & Takahisa Wakabayashi  7 The widespread use of video games has raised concerns about their potential negative impact on mental well-being. Nevertheless, the empirical evidence supporting this notion is largely based on correlational studies, warranting further investigation into the causal relationship. Here we identify the causal efect of video gaming on mental well-being in Japan (2020–2022) using game console lotteries as a natural experiment. Employing approaches designed for causal inference on survey data $( n = 9 7 , 6 0 2 )$ ), we found that game console ownership, along with increased game play, improved mental well-being. The console ownership reduced psychological distress and improved life satisfaction by 0.1–0.6 standard deviations. Furthermore, a causal forest machine learning algorithm revealed divergent impacts between diferent types of console, with one showing smaller benefts for adolescents and females while the other showed larger benefts for adolescents. These fndings highlight the complex impact of digital media on mental well-being and the importance of considering diferential screen time efects. As digital devices and the Internet become integral parts of daily life, concerns about their potential negative impact on human well-being, especially that of prolonged screen time, have become more pronounced. Video games, at the forefront of this debate, increasingly encounter public scepticism1 . Controversial health policy decisions, such as the latest discussions by the World Health Organization regarding gaming disorder, have exacerbated negative perceptions of video gaming2 . The addition of gaming disorder to the International Classification of Diseases (ICD-11) has led to stigmatization among many young people and their carers who consider gaming as a normal part of life3 . Amid escalating concerns about the negative effects of gaming, the coronavirus disease 2019 (COVID-19) pandemic that emerged in 2020 temporarily spotlighted video games as a preferred form of leisure that fit social distancing guidelines. The global number of individuals playing video games has reached nearly three billion4 , accompanied by an increase in gaming time5 . Yet, this surge in video game engagement has renewed concerns about potential negative health impacts6 . Policymakers, researchers and public stakeholders are particularly concerned about game addiction and its possible adverse effects on psychological well-being. However, the current evidence on the effects of video game play is insufficient, not necessarily due to a lack of research but rather the focus and approach of existing studies.

akeholders are particularly concerned about game addiction and its possible adverse effects on psychological well-being. However, the current evidence on the effects of video game play is insufficient, not necessarily due to a lack of research but rather the focus and approach of existing studies. There is extensive research on the effects of video games on users, including their impact on addiction, well-being, cognitive function and aggression. Over the past decades, aggression has received considerable academic attention, but no conclusive evidence on the relationship between gaming and aggression exists. Early studies, predominantly from the 2000s, suggested a connection between digital game violence and heightened aggression7,8 . However, many subsequent studies, including pre-registered ones, have disputed this linkage9–11. While the evidence about aggression remains inconclusive, it is noteworthy that the interest among scholars and policymakers has progressively 1 Research Institute of Economic Science, Nihon University, Tokyo, Japan. 2 Ritsumeikan Center for Game Studies, Ritsumeikan University, Kyoto, Japan. 3 Research Center for Child Mental Development, Hamamatsu University School of Medicine, Hamamatsu, Japan. 4 United Graduate School of Child Development, Osaka University, Kanazawa University, Hamamatsu University School of Medicine, Chiba University, and University of Fukui, Suita, Japan. 5 Department of Policy Studies, National Graduate Institute for Policy Studies, Tokyo, Japan. 6 Office of Audit Support and Innovations, Board of Audit of Japan, Tokyo, Japan. 7 Faculty of Regional Policy, Takasaki City University of Economics, Takasaki, Japan.  e-mail: [email protected]

shifted to the connection between video gaming and mental well-being over the past decade12. The examination of the association between gaming and psychological well-being in existing literature has yielded mixed and inconsistent findings13,14. First, negative associations are mainly documented in observational studies6,15–20, while some experimental studies on violent games also indicate adverse relationships21. A substantial part of such observational studies consists of the examination of the effect of screen time, including time spent on video games6,17,19. Second, positive associations are documented in both observational studies and experimental studies22–27. Some experimental studies have specifically employed video games of genres like casual games and exergames as therapeutic tools, indicating an underlying assumption of their beneficial impact on mental well-being before conducting these studies24,28. In contrast, some recent observational studies have found neither positive nor negative associations29,30. Two primary methodological challenges might underlie the conflicting findings in the literature. First, the scarcity of evidence regarding the causal relationship between playing video games and well-being has been a problem; most observational studies depend on association analysis with either cross-sectional or longitudinal data31,32. Second, many experimental studies lack tests for external validity and have faced criticisms26,30,33. Typical experimental studies that invite participants to play video games for a limited time fail to replicate the natural gaming environment. For example, examining the effect of ‘heavy gaming’ (or dysregulated gaming)—a notion lacking consensus34 or even ‘moderate gaming’ on mental well-being in laboratory settings is challenging. Identifying the causal relationship of habitual behaviours like video gaming, much like other lifestyle habits, such as the health effects of moderate alcohol consumption35,36, presents inherent challenges. The difficulties in conducting randomized controlled trials with habitual behaviours, the multifaceted nature of these behaviours coupled with numerous confounders, and the bidirectional relationship between behaviours and health outcomes collectively complicate the identification of causal relationships. Given the substantial disagreement within the literature and the methodological challenges, how video gaming may actually impact psychological well-being remains elusive. A promising approach is to apply causal inference to observational data37 by exploiting a natural experimental design. However, finding a suitable natural experimental situation has been challenging. This study addresses the gap by leveraging a gaming console lottery, which is as close to an ideal natural experiment as possible. By identifying a unique and fitting situation and collecting relevant data in a timely manner, the study moves beyond the constraints of correlation analysis and controlled laboratory settings, offering a more authentic examination of the effect of playing video games on mental well-being in daily life.

ntifying a unique and fitting situation and collecting relevant data in a timely manner, the study moves beyond the constraints of correlation analysis and controlled laboratory settings, offering a more authentic examination of the effect of playing video games on mental well-being in daily life. To this end, we applied a natural experimental study design to the original survey data containing information on video gaming activity and mental well-being indicators for individuals aged 10–69 in Japan, collected at various points between 2020 and 2022 during the COVID-19 pandemic. Supply chain disruptions and surged demands during this time limited the availability of two major gaming consoles: Nintendo Switch (Switch) and PlayStation 5 (PS5). To address these shortages, Japanese retailers used lotteries to assign these gaming consoles to consumers, inadvertently creating a plausibly random distribution of opportunities to play video games. Winning a lottery became the primary determinant of whether one could purchase these consoles; details of the lotteries and the two game consoles are provided in Supplementary Method 1. Leveraging this unique circumstance and using original survey data, we drew causal inferences grounded in this pandemic context. Additionally, by applying a causal forest machine learning algorithm—an algorithm for estimation of heterogeneous treatment effects—to our diverse sample, we investigated the moderating role of sociodemographic factors in the causal link between video gaming and well-being. While this specific intersection remains relatively understudied, there is a growing consensus among digital media scholars about the value of a person-specific approach, viewing it as a pathway towards tailored mental health interventions38. # Results # Participant characteristics Table 1 presents the study participants’ background characteristics. Out of the 97,602 included survey respondents, 8,192 took part in the lottery. Approximately one-fourth of the 97,602 respondents were between 10 and 25 years old, while $3 9 \%$ were between 45 and 69. Around $2 1 \%$ were students, $1 0 . 7 \%$ were unemployed and $3 9 \%$ were full-time employees. Over one-third of the 8,192 lottery participants $( 3 5 \% )$ were hardcore gamers, and around $20 \%$ were core gamers. Time spent playing games was closely related to video gaming preference; hardcore gamers spent more than 1 h 30 min per day (Supplementary Table 1).

%$ were full-time employees. Over one-third of the 8,192 lottery participants $( 3 5 \% )$ were hardcore gamers, and around $20 \%$ were core gamers. Time spent playing games was closely related to video gaming preference; hardcore gamers spent more than 1 h 30 min per day (Supplementary Table 1). # Preliminary association analysis Before our main analysis, we conducted multivariate regression without leveraging our natural experiment, expecting confounded results. This traditional approach shows associations rather than establishing causal relationships. A statistically significant positive correlation between video gaming and psychological distress (PD) (Kessler Psychological Distress Scale, K6) was found for two out of five estimates by a regression model controlling for a comprehensive set of covariates (Supplementary Table 2, model 3). Conversely, a significant positive association between gaming and life satisfaction (Satisfaction With Life Scale, SWLS) was found. Further details are presented in Supplementary Result 1. # Multivariate regression and propensity score matching Moving to our core analysis (Table 2), which aims to establish causal relationships, we present the results derived from our natural experimental framework. The intention-to-treat (ITT) effects of winning game console lotteries, estimated by multivariate regression and propensity score matching approach (PSM), are statistically significant and comparable (Fig. 1). Winning a Switch lottery reduced PD by approximately 0.2 standard deviations (s.d.) (0.18 s.d. with $9 5 \%$ confidence interval (CI) 0.13–0.24, $P { < } 0 . 0 0 1$ by regression; 0.16 s.d. with $9 5 \%$ CI 0.06–0.25, $P { = } 0 . 0 0 1$ by PSM). Similarly, winning a PS5 lottery led to a distress reduction of around 0.1 s.d. (0.08 s.d. with $9 5 \%$ CI 0.02–0.13, $P { = } 0 . 0 1 4$ by regression; 0.07 s.d. with $9 5 \% \operatorname { C l } 0 . 0 1 \substack { - 0 . 1 4 , P = 0 . 0 3 2 }$ ${ P / } = 0 . 0 3 2$ by PSM). Furthermore, winning a PS5 lottery enhanced life satisfaction by around 0.2 s.d. (0.15 s.d. with $9 5 \% \mathrm { C l } 0 . 1 0 – 0 . 2 0 , P < 0 . 0 0 1$ by regression; 0.18 s.d. with $9 5 \% \mathrm { C l } 0 . 1 0 – 0 . 2 5 , P – 0 . 0 0 1$ $P { < } 0 . 0 0 1$ by PSM). Additionally, lottery winners increased their daily video game play time by around $0 . 5 \mathsf { h }$ (0.53 h with $9 5 \% \mathrm { C l } 0 . 4 3 \mathrm { - } 0 . 6 3 , P < 0 . 0 0 1$ $P { < } 0 . 0 0 1$ by regression; 0.57 h with $9 5 \% \mathrm { C l } 0 . 3 5 \mathrm { - } 0 . 7 9 , P \mathrm { < } 0 . 0 0 1$ $P { < } 0 . 0 0 1$ by PSM) but did not increase smartphone game play time (Supplementary Fig. 1).

9 5 \% \mathrm { C l } 0 . 4 3 \mathrm { - } 0 . 6 3 , P < 0 . 0 0 1$ $P { < } 0 . 0 0 1$ by regression; 0.57 h with $9 5 \% \mathrm { C l } 0 . 3 5 \mathrm { - } 0 . 7 9 , P \mathrm { < } 0 . 0 0 1$ $P { < } 0 . 0 0 1$ by PSM) but did not increase smartphone game play time (Supplementary Fig. 1). The lottery winners and non-winners exhibited minor differences in the background characteristics: only one—number of times that the respondents joined the lotteries (as expected, see ‘Statistical analysis’ section in Methods)—out of 30 variables had standardized differences exceeding 0.10 in absolute value for PS5 lottery participants (Supplementary Table 3) and three for Switch lottery participants (Supplementary Table 4 with more details in Supplementary Result 3). Additionally, the ‘ITT effect’ on pseudo-outcomes was small and not statistically significant, confirming unconfoundedness (Supplementary Table 5). Moreover, the covariates’ balance after matching and common support indicated that our PSM analysis was successful in achieving a balance between the treatment and control groups (Supplementary Figs. 2–5 with further explanations in Supplementary Result 4). Additional analyses mitigate concerns over an unadjusted potential confounder: non-winning lottery participations (elaborated in ‘Assessing natural experiment validity’ subsection in Methods), thereby