Introduction

The COVID-19 pandemic was accompanied by periods of lockdown and physical distancing, which caused substantial shifts in lifestyle habits worldwide (Bailey and Scheuer, 2022). One domain that has drawn interest during this period was video gaming. As people found themselves confined to their homes, questions arose about how the change in circumstances might impact gaming habits and, in turn, players’ well-being. Many concerns were raised about whether players would increase their gaming time and, in turn, decrease their well-being and increase adverse outcomes such as depression or anxiety (Xu et al., 2021). In contrast, some scholars argued that increased gaming during a pandemic could be an effective strategy for maintaining well-being in a restrictive physical environment. It may do so by offering high-quality entertainment, stress relief, and opportunities for social interaction with other players (Boldi et al. 2024; Giardina et al., 2021).

As countries transitioned into the post-pandemic period, it is essential to synthesize the generated knowledge about gaming time and players’ well-being during the COVID-19 pandemic using meta-analytical methods. Thus, we aimed to scrutinize and integrate the available findings. We sought to understand to what extent players changed their gaming time during the COVID-19 pandemic relative to their pre-pandemic levels and whether the amount of time spent playing was related to their mental health, including well-being (e.g., positive affect) and ill-being (e.g., depression) metrics. This provides an empirical-based answer to whether the concerns of adverse (Maraz et al., 2021) or the hopes of protective gaming effects (Giardina et al., 2021) were justified.

This aim is essential because our conclusions might impact various stakeholders, such as players, their parents or partners, educators, and policymakers (European Commission, 2023). For instance, if the increases in gaming time were high and their impact on well-being was negative, more awareness should be raised regarding healthy gaming use during pandemics. In contrast, if the impact on well-being was positive, increased gaming might be advocated as a substantial and endorsed method of sustaining well-being during pandemics. Finally, if the increases were negligible or the impact on well-being neutral, the attention and effort might be redirected to areas other than gaming to optimize well-being and ill-being metrics during pandemics.

The COVID-19 pandemic and video gaming

Video gaming is a global phenomenon that has steadily grown over the last decades. It engages around 70% of the population in developed societies (Statista, 2022). Of the four billion people worldwide with access to gaming technology, approximately three billion play video games (Statista, 2022). This widespread involvement in gaming makes studying gaming relevant to the well-being of a large part of the population. Due to its popularity, any shifts in gaming patterns are likely to have a global impact. This might be particularly important during pandemics, which change daily habits, including work, education, and leisure.

The number of gamers grew before, during, and after the COVID-19 pandemic. It grew by 34% from 2015 to 2023 and is expected to rise a further 7% in the next two years. The video game market generates $347 billion in annual revenue. It is the third-largest entertainment industry in the world (Statista, 2022). There was a 31% increase in consumer spending on video gaming during the COVID-19 pandemic. Moreover, playtime declined by around 20% as the lockdowns ended. This mainly affected hardcore players, suggesting that the pre-pandemic increases were boosted during the pandemic.

Moreover, during the COVID-19 pandemic, social games such as Animal Crossing: New Horizons (Nintendo, Japan) and Among Us (Innersloth, US) surged. For instance, Animal Crossing: New Horizons was used as a social environment for weddings, graduation ceremonies, political campaigning, and virtual protests (Wikipedia, 2023a). Furthermore, the success of Among Us, a social deception game, has been attributed to the COVID-19 pandemic. The game was released in 2018 and received little attention before the pandemic (Wikipedia, 2023b).

However, during the COVID-19 pandemic, there were some challenges for the gaming industry and the players. For instance, due to the breaks in supply chains, it was problematic for manufacturers to produce and distribute video game consoles and hardware. This might have restricted access to individuals who wanted to initiate or improve their gaming experience. Moreover, some developers had problems releasing new games due to the difficulty of working remotely on their games.

Gamers’ well-being

Well-being is a broad concept that overarches a family of diverse metrics (Kaczmarek, 2020) As for any other population, it is critical to distinguish and account for the positive and negative aspects of gamer well-being (Keyes, 2003). It includes high levels of well-being and the absence of ill-being symptomatology. For instance, the theories of well-being often focus on satisfaction and positive emotions (Diener et al., 2002; Kim-Prieto et al., 2005). Ill-being (diminished well-being) includes adverse outcomes such as negative emotions, depression, anxiety, or anger (Schalet et al., 2016).

The necessity of accounting for the positive and negative indicators has a long history (WHO, 1947; Headey et al., 1984; Keyes, 2003). Moreover, both well-being continua have different biological correlates. It suggests that they are not merely mirror reflections of each other but distinct phenomena (Ryff et al., 2006). Consequently, some individuals might increase their ill-being symptomatology during pandemics while maintaining similar levels of well-being. Others might reduce their symptomatology with little benefit for their well-being (Keyes, 2003).

The importance of the distinction between well-being and ill-being is generalized to gamers’ well-being. For instance, some studies indicated that problematic gaming is more related to increased ill-being (e.g., depressed mood) than to decreased well-being (e.g., life satisfaction) (Ballou and Van Rooij, 2021). Consequently, focusing on both aspects of well-being (positive and negative) leads to a more comprehensive perspective. An exclusive focus on positive well-being might underestimate the gaming risks.

In contrast, a sole focus on negative indicators might overlook that despite increased symptomatology, some gamers are likely to experience less pronounced life satisfaction deterioration and maintain relatively high satisfaction with life. This distinction might be especially relevant to studying dynamic life changes that might provide challenges mixed with new opportunities, as might have been the case for the COVID-19 pandemic.

Video gaming and mental health

Self-determination theory (Ryan and Deci, 2000) is frequently applied to explain how certain activities enhance or diminish well-being. This theory posits that well-being results from satisfying basic psychological needs such as competence, autonomy, and relatedness. This conceptual framework has been effectively used to explain players’ well-being (Ballou and Deterding, 2024; Ryan et al., 2006).

For example, the Basic Needs in Games model (Ballou and Deterding, 2024) proposes that mental health outcomes associated with gaming depend on the motivational quality of play. It applies particularly to the extent to which both quantity and quality of play result in need satisfaction or frustration. This model also emphasizes the role of life fit, i.e., successful management of gaming alongside other life activities. Within the context of the COVID-19 pandemic and associated lockdowns, this model underscores that external factors partly shape gaming behavior. Such external factors may enhance extrinsic motivation, increasing play quantity, even when gaming does not effectively fulfill psychological needs.

Video gaming time measurement

Gaming time can be measured using objective (behavioral) metrics or subjective self-reports. Each method has its advantages and limitations. Behavioral metrics (such as those obtained with telemetry, i.e., automatic log files recorded by the game platform) might be considered the most direct because they involve information logged by the gaming system itself. Such information can be reported by players from their profile information (e.g., Behnke et al., 2020). It can also be provided directly to the researchers by game publishers (e.g., Vuorre et al., 2022). This method provides meaningful results. Nonetheless, several limitations impede its validity, including constraints that might be specific to shifts in gaming consumption during COVID-19.

For some players, the distribution services might overestimate gaming time. For instance, it may be problematic to distinguish active gaming from idling while the game is turned on. This is more likely when players switch between tasks, such as mixing playing with work or education. During the COVID-19 lockdown, many individuals used their computers more frequently for work or learning since they did not attend their usual workplaces or schools. Consequently, they might have their favorite games launched more often, even though they were not actively involved, or at least for some time, their involvement was minimal, e.g., doing less engaging tasks such as character customization.

Moreover, behavioral metrics usually use information from the leading digital games’ distribution services (e.g., Steam). This is problematic because individuals also play video games outside of central gaming ecosystems, e.g., using smartphones. Consequently, during a pandemic, an increase in one ecosystem activity might reflect a change in a platform preference rather than a general change in playing time. For instance, players might play more using desktop PCs rather than smartphones because of longer time spent at home and less outside, e.g., commuting.

Finally, the game systems cannot distinguish whether a specific user uses the account exclusively or whether other persons from their household also use it. Consequently, when more individuals use the same account, the gaming time attributed to a single account might also increase. This can be misleading because the primary account user might not change their gaming time or decrease due to the pressure to share the account with other household members. Such misattribution was probably common during the COVID-19 lockdown, when youth and children stayed home and might have used their parent’s accounts to play games.

Subjective self-reports are another method for measuring time spent playing video games. This method is more straightforward, as players are asked to reflect on and recall the time spent playing video games over a specified period. This method is free of the limitations presented above. Nonetheless, it has limitations resulting from relying on memory and subjective judgments. Subjective and objective metrics of gaming time are correlated. However, they can link with well-being differently (Oka et al., 2024). Self-reports might be the most feasible for meta-analytical work due to their greater availability.

Video gaming time and well-being

Concerns about gaming consumption’s psychological effects are often centered around gaming time (Kaczmarek and Drążkowski, 2014; Vuorre et al., 2022). This might result from several factors, such as the ease of gaming time reporting by gamers, non-gamers (e.g., parents), or digital game distribution services. Indeed, extended gaming durations are characteristic of problematic gaming (Kaczmarek et al., 2022; Severo et al., 2020). However, most players play in a controlled way (Stevens et al., 2021). Thus, the study on problematic gaming does not adequately represent the whole gaming community.

For instance, while problematic players spend more time playing and report more physical symptoms, gaming time does not predict physical symptomatology (Kaczmarek et al., 2022). Several studies indicated that how people play matters more than how much they play (Giardina et al., 2021; Przybylski et al., 2010). For instance, in the COVID-19 pandemic, playing for social compensation was related to lower emotional distress during self-isolation (Giardina et al., 2021).

Determining the strength of the link between gaming time and well-being is consequential. If the link in the general players’ population is substantial, limiting the amount of game consumption would be a feasible factor in preventing deterioration or improving well-being among players. Suppose the link yields weak or limited to specific subgroups. In that case, the global efforts should be redirected toward pursuing more significant risk and protective factors, e.g., game choice (Greitemeyer and Mügge, 2014), gaming motivation (Przybylski et al., 2010), gamers’ psychosocial resources such as self-regulation skills (Luxford et al., 2022), or gamers’ clinical status (Cheng et al., 2018).

Moreover, players’ well-being might be considered in a broader context of life circumstances because, in some cases, extended gaming times reflect players’ attempts to escape from problems in real life. This can lead to psychosocial resource depletion and well-being deterioration (Kaczmarek and Drążkowski, 2014).

Cultural differences in gaming

Gaming behaviors and their psychological impacts vary substantially across cultural contexts. This reflects differences in social norms, cultural values, and regulatory frameworks (Anderson et al., 2010; Cheng et al., 2018). For instance, in some East Asian countries, such as China and South Korea, youth gaming is heavily regulated due to concerns related to academic performance and addiction risks. Conversely, Western countries generally impose fewer formal restrictions. This reflects broader social acceptance of gaming (Lee and Wohn, 2012). However, recent studies challenge the notion of Asian exceptionalism by revealing similarities in gaming playtime between East Asian and Northern European countries (Zendle et al., 2023).

Cultural variations also extend to adverse gaming behaviors. For instance, significant differences in tendencies toward toxic behavior among Multiplayer Online Battle Arena players in North America compared to India underscore the influence of offline cultural environments on online interactions (Kordyaka et al., 2023). Additionally, the prevalence of gaming disorder exhibits considerable cross-national variation. Higher rates have been observed in Asian and Middle Eastern nations relative to South and Central American countries (Thomas et al., 2024). Cross-cultural research consistently links problematic gaming to adverse psychological outcomes, including increased depression and anxiety. However, the strength of these associations differs by region. This likely reflects varied cultural perceptions and social functions attributed to gaming (Cheng et al., 2018).

Motivations for gaming also differ notably across cultures. Eastern cultures often prioritize gaming for social connectedness. Western cultures emphasize individual achievement and competitive play (Anderson et al., 2010; Lee and Wohn, 2012). Furthermore, beliefs regarding gaming as compensation for social relationships differ markedly between Eastern and Western contexts (Colwell and Kato, 2003). These motivational differences moderate gaming’s psychological and emotional impacts, reinforcing the importance of cultural context.

Behavioral patterns within games further illustrate cultural distinctions. German and Swedish gamers demonstrate significantly more cooperative behaviors than players from the United Kingdom and the United States (Bialas et al., 2014). National cultural dimensions, particularly individualism, heavily influence playing styles and behavioral expressions. This highlights substantial differences based on players’ cultural backgrounds (Pan et al., 2024).

These findings underscore that cultural contexts strongly influence the relationship between gaming behaviors and psychological well-being. This relationship is shaped by national regulations, cultural norms, and social perceptions of gaming (Lee and Wohn, 2012; Neut et al., 2023). Thus, understanding the cultural context is crucial for accurately synthesizing and interpreting research on gaming behaviors and psychological well-being

Overview of the present investigation

We aimed to integrate and evaluate findings that measured gaming time and the well-being of video game players during the pandemic. First, we aimed to determine the size of the change in gaming time (pandemic to pre-pandemic or pre-lockdown levels). Second, we examined whether gaming time was related to well-being during the pandemic. Furthermore, we aimed to test possible moderators that might reveal significant effects in some groups or contexts.We focused on longitudinal studies that involved two measurement points because this method is the most robust in determining change.

We expected significant increases in gaming time during the COVID-19 pandemic relative to pre-pandemic levels. However, building upon previous theories and research on well-being and video gaming, we expected negligible-to-weak effects of gaming time on well-being. Examining these effects quantitatively is essential to provide an empirical perspective on the effects of the COVID-19 pandemic on the gaming population. This might help understand the past pandemic events’ psychological impact and provide an informed basis for possible future pandemics or other events involving lockdowns.

Methods

Selection of studies

We conducted a systematic literature search using EBSCO, PsycINFO, PubMed, ProQuest, and Google Scholar, covering the period from March 2020 to December 2021 Fig. 1. We used the following terms: [“games” OR “gameplay” OR “video games” OR “playing video games” OR “increased gameplay” OR “video game time” OR “hours played” OR “minutes played” AND “covid” OR “well-being” OR “negative emotions” OR “negative affect” OR “depression” OR “anxiety” OR “hostility” OR “loneliness” OR “positive affect” OR “positive emotions” OR “happiness” OR “flourishing” OR “quality of life”]. We also cross-checked the references in the searched studies. We contacted 41 authors who had published articles or dissertations on gaming, well-being, and COVID-19 and asked for any unpublished materials. Of the 22 authors responding to the request (53.65%), none provided unpublished materials.

Fig. 1
figure 1

Flow Diagram of the Search Procedure.

Full size image

We selected potentially eligible studies in two phases. First, PC reviewed the titles and abstracts. Then, PC, MB, and DD reviewed the full text of the articles with the appropriate title or abstract. All studies identified as potentially eligible during the first selection phase were re-evaluated in the second selection phase.

Inclusion criteria for eligible studies were as follows: (1) the study was conducted during the COVID-19 lockdown; (2) gaming time and well-being or ill-being were measured; and (3) non-clinical participants took part in a study. Exclusion criteria included: (1) gaming time was not reported, and (2) available data did not allow for calculating effect sizes. Inter-rater agreement for the initial selection on which studies met the eligibility criteria and could be included was acceptable (Krippendorff’s α = 0.68). Disagreements between coders were resolved through discussion. Table 1, 2.

Table 1 Characteristics of studies included in the meta-analysis on video gaming time change.
Full size table
Table 2 Characteristics of Studies on Video Gaming Time and Well-being Included in the Meta-Analysis.
Full size table

Data extraction

Coding

We coded data collection time (i.e., a quarter of 2020), data collection location (i.e., continent), type of measure (ill-being = 0, well-being = 1), participants’ age, gender distribution (i.e., percentage of females), sample size, and the study quality index. The study quality index included scores from six criteria: (a) provision of exclusion criteria (yes = 1, no = 0); (b) report of artifacts, outliers, and missing data (yes = 1, no = 0); (c) availability of dataset and (d) code for analysis (yes = 1, no = 0); (e) presence of sample size justification (yes = 1, no = 0); and (f) pre-registration of the study (yes = 1, no = 0).

Effect size extraction

We coded or calculated the difference in gaming time before and after the COVID-19 lockdown to determine the effect size. Second, we coded the correlations between gaming time and well-being or ill-being. We reversed the sign for ill-being so that higher values represented more favorable outcomes, i.e., more well-being and absence of ill-being. In most studies, the authors reported the means and standard deviations of gaming time and correlations of well-being or ill-being measures with gaming time. For studies reporting other indicators (e.g., regression coefficients), we sent the authors a request for the averages of the relevant variables. Of the forty-one authors we contacted, twenty-two responded to our request (53.65%), and of these, two authors did not grant us access to the requested data (9.09%). The present meta-analytic review used (1) Cohen’s d and (2) Pearson’s r correlation coefficients as the effect size indexes. We interpreted effect sizes using conventional standards, i.e., small, d = 0.20; medium, d = 0.50; large, d = 0.80 (Cohen, 1992). For correlations (Pearson’s r), we considered effect sizes as small (r = 0.10), medium (r = 0.30), and large (r = 0.50) (Cohen, 1992).

Meta-analytic procedures

We employed random-effects models with robust variance estimation and the Subgroup Correlated Effects working model using the metafor and clubSandwich packages in R (Pustejovsky and Tipton, 2022; Viechtbauer, 2010; R Core Team, 2020). We used random-effects models because we expected the true effect sizes to vary across studies (Borenstein et al., 2010). For instance, in our project, the variation might have been caused by regional differences in mental health indicators (Diener and Tay, 2015), video game use (Stevens et al., 2021), or COVID-19 impact (Olff et al., 2021). We used robust variance estimation to address the non-independence among multiple effect sizes from the same study (Pustejovsky and Tipton, 2022). Selection of these tests provided prediction intervals that better reflect the expected range of true effects in future studies. We used similar meta-analytic procedures for changes in gaming time and the association between gaming time and well-being or ill-being. We converted the effect sizes to Fisher’s z-transformed and weighted them by the inverse of their variance before combining them. This weighting gave more importance to larger studies in the pooling process, ensuring that each study’s effect size contributes proportionately to the overall effect size based on its sample size and precision.

Additionally, we examined outliers by computing studentized residuals and through leave-one-out sensitivity analyses, which identified effect sizes that notably contributed to the overall heterogeneity and results (Gleser and Olkin, 2009; Viechtbauer and Cheung, 2010).

Many articles provided more than one relevant effect size for our analyses, including multiple types of well-being or ill-being measures. As articles reported multiple effect sizes, the assumption of independence between effect size point estimates was unmet. To address this, we calculated robust variance estimates and confidence intervals (CIs). We also employed the Subgroup Correlated Effects working model, which allows for dependencies across sub-groups (well-being or ill-being measures) while maintaining the clarity of sub-group analysis. This model combines the principles of separate meta-regression analyses in a full-data working model. As a result, we ran a single meta-analytic model for both changes in gaming time and the association between gaming time and well-being or ill-being, taking into account the dependence of the effect sizes (Hedges et al., 2010).

We calculated prediction intervals for the mean correlations to account for the heterogeneity of the pooled effect sizes (Higgins et al., 2009). These intervals indicate where the true effect size would lie in 95% of all similar populations. Lastly, we transformed Fisher’s z back into correlation coefficients.

Publication bias

We separately evaluated potential publication bias for changes in gaming time and the association between gaming time and well-being or ill-being. Publication bias was assessed using funnel plots and the adapted Egger’s regression test, adjusted for the dependency of multiple effect sizes per study (Rodgers and Pustejovsky, 2021). With funnel plots, we visually examined possible asymmetry. Egger’s regression provided statistical evidence for asymmetry, indicating potential publication bias.

Moderator analysis

Finally, we ran six moderation analyses for changes in gaming time and seven moderator analyses for the association between gaming time and well-being or ill-being (inversely coded by multiplying by −1). We used the omnibus F test to determine whether factors (e.g., usage of well-being vs. ill-being measures) had a moderating effect. The omnibus F tests the null hypothesis that the predictor is unrelated to the effect sizes. To address the risk of Type I error resulting from multiple comparisons (e.g., testing seven different moderators on the same dataset), which is common in meta-analyses (Cafri et al., 2010), we adjusted probability values using Bonferroni correction (Abdi, 2007).

We determined whether a specific moderator affected the mean effect sizes by evaluating the significance of the regression coefficients. A significant omnibus F test for a continuous variable indicates a significant correlation between the continuous variable (e.g., percentage of females) and the mean effect of gaming time and well-being or ill-being. A significant omnibus F test for categorical variables (e.g., data collection location) suggests a significant difference from the comparison category. In tables presenting moderation analyses, ‘RC’ (Reference Category) denotes the baseline group against which other categories are compared. We deemed the mean effect sizes significantly different in one sub-group from other sub-groups if the CIs of the difference between effect sizes in a given moderator sub-group did not include zero. We interpreted the mean effect of the moderator category when there were at least three studies per category. Finally, for the continuous variables (e.g., participant age and sample size), we evaluated whether the pooled effect sizes were correlated with those measures. To examine whether the observed effects were sensitive to the inclusion of any single study, we conducted a sensitivity analysis using the leave-one-out method. This approach recalculates effect sizes by sequentially omitting each individual study, allowing us to determine whether the results remain robust irrespective of the exclusion of any specific study.

Results

Changes in gaming time

We found that individuals spent more time playing video games during the COVID-19 pandemic relative to the pre-pandemic period, d = 0.26, 95% CI [0.14, 0.37], p = 0.0004, k = 26 (Fig. 2). The effect size for the increase was small. Moreover, the variance of the effect τ = 0.196 and the wide prediction intervals PI [−0.330, 0.700] underscore the substantial variability that future studies may observe, ranging from possible decreases to increases in gaming time.

Fig. 2: Forrest plot of the effect sizes for changes in gaming time before and after the COVID-19 pandemic.
figure 2

The square boxes represent Cohen’s d and sample sizes (the larger the box, the larger the sample size, the larger the contribution to the total effect size). The lines represent 95% confidence intervals. The diamond represents the pooled effect size and the 95% confidence intervals.

Full size image

Gaming time and well/Ill-being

Video gaming duration was unrelated to well-being, r = −0.032, 95% CI [−0.084, 0.020], p = 0.22, k = 100 (Fig. 3). The variance of the effect τ = 0.090 and the prediction intervals PI [−0.445, 0.393] indicate that positive and negative associations between gaming time and well-being cannot be ruled out for future studies.

Fig. 3: Forrest plot of the effect sizes for gaming time and well-being included in the meta-analysis.
figure 3

The square boxes represent Cohen’s d and sample sizes (the larger the box, the larger the sample size, the larger the contribution to the total effect size). The lines represent 95% confidence intervals. The diamond represents the pooled effect size and the 95% confidence intervals.

Full size image

Sensitivity analysis

For changes in gaming time, the pooled effect size ranged from 0.23 to 0.29 across leave-one-out analyses. The largest change was observed upon removing the study by Schmidt et al. (2020), which lowered the overall effect to 0.23. In all cases, the direction of the effect remained positive. Regarding the relationship between gaming time and well-being, the pooled effect size ranged from −0.05 to −0.02 across leave-one-out analyses, with the largest change observed when removing the study by Giardina et al. (2021), which lowered the overall effect to −0.05. The direction of all effects remained negative. No individual study’s removal changed the summary effect beyond a negligible amount, indicating robust findings.

Publication bias

We examined the possibility of publication bias for changes in gaming time and the association between gaming time and well-being (Figs. 4–5). The funnel plots for both analyses appeared symmetrical, suggesting no clear indication of bias. Consistent with these visual inspections, Egger’s regression tests confirmed that neither changes in gaming time (Intercept = 0.331, Slope = −1.606, SE = 2.308, p = 0.753) nor the association between gaming time and well-being or ill-being (Intercept = −0.020, Slope = −0.271, SE = 0.991, p = 0.608) demonstrated statistically significant evidence of publication bias.

Fig. 4
figure 4

Funnel plot of the effect sizes for the change in gaming time before and after the COVID-19 pandemic included in the meta-analysis.

Full size image
Fig. 5
figure 5

Funnel plot of the effect sizes for the relationship between gaming time and well-being during the COVID-19 pandemic.

Full size image

Moderator analyses

The results above indicate that the mean effect sizes should not be treated as estimates of one common effect size. Thus, moderator analyses are justified to search for variables that could explain the heterogeneity of the overall increases in gaming time during the COVID-19 pandemic and the association between gaming time and well-being.

Changes in gaming time

Table 1 presents the results of the omnibus test of the moderator analyses and effect sizes. The moderator analysis showed that, under most conditions, gaming time increased during the COVID-19 pandemic; that is, the increases in gaming time during the COVID-19 pandemic were not affected by most variables tested as moderators. We found that the time of data collection (a quarter of 2020) moderated the size of increases in gaming time, with higher increases in the second quarter. Furthermore, we found that during the pandemic, people played more in Europe but not Asia.

Gaming time and well-being

The second moderator analysis showed that, under most conditions, gaming time was not associated with well-being. The association between gaming time and well-being was not affected by most variables tested as moderators. Table 2 presents the results of the omnibus test of the moderator analyses. We found that gamers in Asia had lower well-being if they played more. Nevertheless, the effect size was negligible. Moreover, the association between gaming time was significant for ill-being rather than well-being. This effect size was also negligible Tables 3 and 4.

Table 3 Results of moderator analyses for the increases in gaming time during the COVID-19 pandemic.
Full size table
Table 4 Results of moderator analyses for the association between gaming time and well-being.
Full size table

Discussion

We determined the extent to which players changed their gaming time during the COVID-19 pandemic relative to their pre-pandemic levels and examined whether individuals who played more had different levels of well-being than their peers who played less. Our findings indicated that individuals spent slightly more time on video gaming during the COVID-19 pandemic, but gaming time was unrelated to their well-being. We found some instances where the links were stronger but had negligible magnitudes. Our findings suggest that despite considerable attention from policymakers, caretakers, and players, COVID-19 had a negligible effect on the amount of gaming consumption. Furthermore, more playing was not related to well-being risks or benefits. Sensitivity analyses indicated that our findings are robust and not overly dependent on any single study. These findings are vital for further consideration of the role of gaming from broader public health and individual perspectives.

We found that players spent more time gaming during the pandemic than pre-pandemic levels. This complements findings from sources that used different methods (Barr et al., 2022). It supports the Basic Needs in Games model by documenting that global changes in external factors lead to changes in gaming behavior (Ballou and Deterding, 2024). However, despite the statistical significance of our findings, the small effect size and wide prediction intervals call for careful interpretation. The increases were small, contradicting some less stringent surveys and anecdotal evidence, arguing for dramatic increases in gaming time. Moreover, our findings contradict subjective gamers’ reports, in which they were asked to what extent their gaming habits changed during the pandemic. In such surveys, players typically believed their gaming durations increased substantially (Barr et al., 2022). This also points to the limited information derived from asking about gaming time increases without determining whether these increases were significant or just barely noticeable. For instance, in a survey, most gamers reported increasing gaming times (Barr et al., 2022). However, this approach does not differentiate whether their increases were small, moderate, or large. Our analytical approach provided a more detailed quantitative description of the increases, allowing their interpretation using well-established effect size conventions. We determined that the growth was relatively weak and sometimes negligible. However, wide prediction intervals suggest that future studies might observe a broader range of effects.

Our findings emphasize the need for further research to increase the understanding of the impact of global crises on gaming behavior. The link between the pandemic and gaming time might be more complex and involve other factors unaccounted for by the available studies. Moreover, integrating our findings with qualitative methods might provide more complex observations. The qualitative and quantitative methods often converge (Sheals et al., 2016). However, they can also contradict each other (Wagner et al., 2012). Moreover, qualitative findings can suggest more pronounced effects than those reported using scales (Slonim-Nevo and Nevo, 2009). Several studies used qualitative methods to address questions regarding players’ well-being during COVID-19 (Barr and Copeland-Stewart, 2022; Elis et al., 2020; Yee and Sng, 2022). For instance, in response to open-ended questions, approximately three-quarters of the participants reported that playing video games had benefited their well-being during the COVID-19 pandemic (Elis et al., 2020). In addition to seeking an escape during the pandemic and as entertainment, participants reported they used video games for emotional coping and to lower stress, relax, and alleviate well-being conditions. This indicates that qualitative methods might provide further insights within this study domain.

We found that the amount of time devoted to gaming during the COVID-19 pandemic was unrelated to well-being. This result indicates that gamers had similar levels of well-being regardless of whether they played more or less than their counterparts. Our evidence converges with a recent large-scale behavioral data analysis that found negligible links between gaming time and well-being (Vuorre et al., 2022). Nonetheless, these findings contradict the prevailing subjective belief among gamers that playing video games during COVID-19 positively impacted their well-being (Barr et al., 2022; Elis et al., 2020). This discrepancy between subjective beliefs and empirical data needs further research to determine if other factors we did not account for in our analyses might support or moderate the validity of these beliefs. For instance, there might be misattributions or a subjective bias resulting from the focusing illusion, which overestimates the influence of single factors on well-being (Kaczmarek et al., 2016). We determined that the growth was relatively weak and sometimes negligible. However, wide prediction intervals suggest that future studies might observe a broader range of effects.

Moreover, our findings might be interpreted as suggesting that the quality of the gaming experience may be more significant than the amount of time spent gaming. For instance, this would support previous studies indicating that gaming time is not predictive of well-being when gaming satisfies psychological needs and provides enjoyment (Ballou and Deterding, 2024; Przybylski et al., 2010). This, again, points to the fact that during the pandemic, focusing on gaming time (while neglecting other aspects of gaming consumption) might tell little about whether gaming benefited or harmed individuals’ well-being. It is an important finding because gaming time is a gaming parameter that is the easiest to observe and measure for the player, their peers, service providers, or any other external observers (parents, partners, etc.). In contrast, other gaming aspects (e.g., motivations) might be more challenging to recognize and evaluate. This may facilitate the formulation of misconceptions about the impact of gaming time on well-being in general and during pandemics.

Along similar lines, gaming time is among the most feasible parameters to account for in a meta-analysis because it is reported in most studies on gaming as part of descriptive statistics or as a control variable in models. In contrast, other parameters are often dispersed among several studies, making it difficult to establish a substantial empirical base for evaluating their impact on well-being. However, our findings suggest that this commonly reported parameter might be among the least important in determining the effect of video game use on well-being. Factors that are more nuanced and less widely reported might play a more significant role. It might be challenging to establish their role using a large-scale meta-analytic approach.

Another aspect of our analyses was evaluating potential publication bias, as unreported null or negative findings can distort meta-analytic results. We found no evidence that might suggest the publication bias for either the increases in gaming time during the pandemic or the associations between gaming time and mental health metrics. This finding implies that the available literature provided a balanced representation of both significant and non-significant findings, providing robustness to our conclusions. The lack of observed publication bias supports the neutrality of the current scientific reporting in this research domain, reassuring policymakers, researchers, and stakeholders that our synthesis accurately reflects the current state of empirical evidence.

We found several moderators that indicated situations when players spent more time playing video games under specific circumstances. For instance, we found considerably greater increases in gaming time for the second quarter of 2020 than in other periods. The effect size was small-to-moderate, whereas others were negligible-to-small and insignificant for the fourth quarter. With the recent design, it is difficult to distinguish whether these effects were due to the development of the pandemic or the time of year. However, these findings indicate that it is essential to determine the time of year in future studies. Moreover, we found that the increases we observed in Europe did not occur in Asia. This is a worthwhile finding indicating potential regional or cultural differences. However, they might also reflect different progress of the COVID-19 pandemic, different histories of previous pandemics (especially in Eastern countries), or differences in formulated national regulations. This finding brings to attention that gaming findings based on samples from one region cannot be easily generalized to other areas, as the gaming cultural context might differ.

The present findings contribute to the discussion regarding the cultural context of gaming. Noteworthy results lie in the cultural variations - or, in most cases, the lack thereof - in gaming behavior and its relationship to well-being. We found that gaming time increased in Europe but not in Asia. These findings align with previous literature emphasizing how cultural norms and regulatory environments shape gaming behaviors differently across regions (Cheng et al., 2018; Lee and Wohn, 2012; Zendle et al., 2023). However, this effect was negligible, suggesting limited practical implications. Furthermore, in our moderator analyses, Asian gamers displayed a statistically significant negative correlation between gaming time and well-being. Nevertheless, the effect size was negligible, in line with studies highlighting only modest cross-national differences in how gaming behaviors influence mental health (Zendle et al., 2023).

Self-Determination Theory (Ryan and Deci, 2000) and the Basic Needs in Games model (Ballou and Deterding, 2024) offer frameworks to explain these cultural differences. In Asian cultures, which often prioritize collectivism and relatedness over autonomy (Sheldon, 2012), solitary gaming during the COVID-19 pandemic may have frustrated the need for relatedness, potentially driving the negative correlation with well-being. Conversely, in European cultures, which value individualism, gaming likely supported autonomy and competence, even with reduced social interaction. The Basic Needs in Games model’s “life fit” concept further suggests that in high-expectation Asian contexts, gaming might conflict with obligations like academic achievement, reducing autonomy, and impacting well-being. Though effect sizes were small, these theoretical insights lay a foundation for future theory and research into gaming’s cultural dynamics. These outcomes underscore that while gaming has global relevance and certain universal aspects, cultural factors should be considered as possible moderators of its psychological effects. Future studies should further explore these cultural similarities and differences to better inform global and regional guidelines and policies for digital engagement in times of crisis.

Finally, we also found that ill-being was more affected by gaming time than well-being, reaching a statistically significant threshold, suggesting a possible influence on reducing symptoms of psychological distress. However, the size of the effect was, again, negligible, suggesting little practical meaning. These findings suggest that there might be minor differences between cultures and measures, which might be further dissected to better examine the link between gaming time and well-being and ill-being.

Limitations and future studies

Several limitations of our analysis should be considered. First, our findings are based on self-reports, which are subject to limitations stemming from reliance on memory and subjective judgments (Bonke, 2005; Dockray et al., 2010). While this is the most popular method, more diverse criteria to evaluate time spent playing video games or the magnitude of clinical symptomatology might reflect more comprehensive evidence, such as using behavioral metrics obtained via telemetry.

Moreover, we focused on the increases in average time among individuals who played before and after the pandemic outbreak. Thus, we did not account for individuals who started to play video games after the lockdown. For instance, games like Animal Crossing or Among Us attracted considerable attention during the lockdown. These games might have attracted new players, not accounted for by our analysis.

As the COVID-19 pandemic was global, our study lacks a comparison group, i.e., studies with individuals from areas unaffected by the pandemic. Consequently, it is difficult to determine the extent to which the changes in gaming levels between pre-pandemic vs. post-pandemic reflected or interfered with more natural, yearly gaming duration patterns or the global increase in gaming popularity observed over the last decades (Statista, 2022). For instance, the time of the lockdown and its termination interfered with other events in some countries, such as the end-of-semester examination (Balhara et al., 2020). This is important because 15% of students report increasing or decreasing their gaming time due to examination-related stress. Future studies might focus on in-depth comparisons between countries with different lockdown policies. This has become feasible given that pandemic policies are less uniform, in contrast to the initial phase in 2020–2021, where most countries introduced lockdowns and advocated for minimal physical social contact.

We focused on synthesizing the quantitative evidence, while several studies used qualitative methods to address similar questions and found more pronounced effects (Elis et al., 2020). This suggests that metasyntheses, i.e., integration of findings from qualitative studies, could further expand the perspective (Wagner et al., 2012).

Many included studies provided cross-sectional data, which limits causal interpretations. Future research should prioritize longitudinal methodologies that address the problem of change more explicitly. Such studies would significantly enhance our understanding of causality and temporal dynamics between gaming behaviors and well-being.

Finally, our meta-analysis would benefit from more diversity among included samples, primarily from low-income, Eastern, African, and South American countries, increasing our findings’ reliability and generalizability. For instance, only one study originated from South America. In another sample, just a small fraction was from Africa. Since this depends on data availability, additional studies exploring the role of gaming across various global regions are necessary to inform international policy decisions better and reveal local specificities. Additionally, none of the studies reported an average participant age exceeding 40 years, suggesting that older populations, among whom gaming is increasingly prevalent, may be underrepresented in the analyzed research. Conversely, the youngest participants surveyed were 6 years old. However, an increasingly earlier age of first exposure to computer games is observed among children as young as 3–6 years old (Wang et al., 2024). Future research should prioritize the inclusion of diverse populations, encompassing participants from varied cultural, socioeconomic, and geographic backgrounds. Strategies such as targeted recruitment efforts, cross-cultural collaborations, and multilingual surveys could significantly broaden demographic representation.

Practical implications

Our findings have several practical implications, each referring to a different group of stakeholders involved in the social debate on gaming during pandemics. First, for the general public, understanding that gaming behavior increased during the pandemic with little effect on well-being could facilitate acceptance of video gaming as a healthy leisure. Second, policymakers might account for these findings while establishing policies promoting leisure via video game consumption during extended periods of isolation. Third, our findings might be considered by parents and educators and incorporated into their guidance. The results can help parents, teachers, and caregivers understand that the amount of time spent gaming is, on average, neither harmful nor helpful for well-being. Although not addressed directly in our analysis, other studies complement our findings by suggesting that such stakeholders could focus more on the gaming experience’s quality, the play’s context, and the motivation behind gaming (Giardina et al., 2021; Przybylski et al., 2010). Scrutinizing what the players do rather than focusing on how much time they devote to doing this might be more informative and helpful.

Conclusions

The COVID-19 pandemic has provided a unique opportunity to observe changes in gaming behavior and a variety of well-being metrics during a global crisis. Our findings have challenged popular beliefs about the influence of gaming duration on gamers’ lives. These results added a robust scientific perspective on gaming globally when shifts in gaming time were observed. The conclusions underscore the importance of carefully examining and interpreting the strength of relationships between gaming and well-being. In an increasingly digital world, several stakeholders might capitalize on these findings to better understand the COVID-19 pandemic and guide their decisions during similar disruptive social events in the future.