This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
The rapid expansion of online video gaming as a leisure time activity has led to the appearance of problematic online gaming (POG). According to the literature, POG is associated with different psychiatric symptoms (eg, depression, anxiety) and with specific gaming motives (ie, escape, achievement). Based on studies of alcohol use that suggest a mediator role of drinking motives between distal influences (eg, trauma symptoms) and drinking problems, this study examined the assumption that there is an indirect link between psychiatric distress and POG via the mediation of gaming motives. Furthermore, it was also assumed that there was a moderator effect of gender and game type preference based on the important role gender plays in POG and the structural differences between different game types.
This study had two aims. The first aim was to test the mediating role of online gaming motives between psychiatric symptoms and problematic use of online games. The second aim was to test the moderator effect of gender and game type preference in this mediation model.
An online survey was conducted on a sample of online gamers (N=3186; age: mean 21.1, SD 5.9 years; male: 2859/3186, 89.74%). The Brief Symptom Inventory (BSI), the Motives for Online Gaming Questionnaire (MOGQ), and the Problematic Online Gaming Questionnaire (POGQ) were administered to assess general psychiatric distress, online gaming motives, and problematic online game use, respectively. Structural regression analyses within structural equation modeling were used to test the proposed mediation models and multigroup analyses were used to test gender and game type differences to determine possible moderating effects.
The mediation models fitted the data adequately. The Global Severity Index (GSI) of the BSI indicated that the level of psychiatric distress had a significant positive direct effect (standardized effect=.35,
The results suggest that psychiatric distress is both directly and indirectly (via escape and competition motives) negatively associated with POG. Therefore, the exploration of psychiatric symptoms and gaming motives of POG can be helpful in the preparation of prevention and treatment programs.
The large-scale expansion of the video game industry and online video gaming as a leisure time activity [
The attraction of online games appears to lie in their potential to satisfy different psychological needs that can be conceptualized as motives for gaming. In one earlier study, 7 motivational dimensions for online gaming were identified [
Studies into alcohol use have shed light on the importance of motives in both drinking and problem drinking (eg, [
It has also been reported that psychiatric symptoms can be related to POG. For instance, depression, anxiety, and social anxiety are positively related to problematic gaming [
To date, psychiatric symptoms have been viewed as being directly related to POG. This approach is understandable because—as in other addictions—it is highly probable that psychiatric symptoms accompany the problem behavior. However, an indirect link might also exist between the 2 entities via the mediation of specific motivations. Such findings have been reported in the alcohol and gambling literature. Drinking motives have been shown to be the most proximate factors that precede alcohol use, the gateway via which more distal influences (eg, alcohol expectancies, anxiety sensitivity, trauma symptoms, and personality) are mediated [
Gender differences may also play an important role in problematic gaming. Research has consistently shown that males are more likely to play online games [
Given that different game types (eg, role-playing games, first-person shooters, and real-time strategy games) vary in their basic structural characteristics and gameplay [
Consequently, the present study had 2 aims. The first aim was to test the mediating role of online gaming motives between psychiatric symptoms and problematic use of online games. Based on the literature, it was assumed that escape, coping, and competition motives would act as mediators in the model. The second aim was to test the moderator effect of gender and game type preference in this mediation model. Based on gender differences reported in the online gaming literature, differences in the paths related to competition, social, and escape motives have been expected between male and female players. On the other hand, no assumptions have been made regarding game type-related differences in the lack of findings in this domain.
The study was approved by the Institutional Review Board of the Eötvös Loránd University, Budapest, Hungary.
All the most frequently visited Hungarian online gaming websites having more than 100 registered users were identified (N=18). The majority of the sites (n=11) were game-specific: (massively) multiplayer online role playing games ([M]MORPGs) (eg, Lineage II, World of Warcraft, Guild Wars, Diablo II, Lord of the Rings; Thrillion kincsei [The Treasures of Thrillion])
Before filling out the questionnaire, all participants were informed about the general goals of the study (ie, to obtain an objective and real picture of online games and the online gamer community in contrast with the simplified and rather negative picture the media disseminates) and the time needed to complete it. They were assured about confidentiality and anonymity, and their informed consent was obtained by ticking a box if they agreed to continue and participate in the study. No personal information was collected or stored and no incentives were offered. Data collection took place between April and July 2009. The advertisements were displayed for approximately 3 months, whereas the newsletters including the call for participation were sent twice. According to the voluntary nature of participation, answering all the questions was not mandatory. The number of unique website visitors was not assessed at the time of the recruitment period, and the number of overlapping visitors between the 18 websites was also unknown; therefore, the view rate could not be calculated. In total, 7520 gamers visited the first page of the questionnaire and 4374 completed at least some of it. Consequently, the participation rate was 58.16%. Of these 4374 gamers, 3186 completed the entire survey including psychiatric symptoms, online gaming motives, and problematic online game use resulting in a completion rate of 72.84%. Consequently, all analyses were carried out on this latter subsample. Only completed questionnaires were analyzed. However, skipping answers was allowed; therefore, missing data were treated with full information maximum likelihood (FIML) method with MPlus 6.0.
Data relating to major sociodemographics were collected including age, gender, marital status, and education.
Data were collected regarding weekly game time and preferred game type. Game type preference was obtained using the results of a latent profile analysis on the amount of time spent playing different game types described in a previous study that used the same sample (which was slightly bigger than the current sample because it also included those respondents who did not complete the Brief Symptom Inventory [BSI] placed at the end of the questionnaire) [
These were assessed using the Hungarian version of the BSI [
These were assessed using the Motives for Online Gaming Questionnaire (MOGQ) [
This was assessed using the Problematic Online Gaming Questionnaire (POGQ) [
Structural regression analyses within structural equation modeling (SEM) were used to test the proposed mediation models. Because the scales were not normally distributed, maximum likelihood estimation robust to nonnormality (MLR) was used in all SEM analyses [
The mean age of the sample (N=3186) was 21.1 years (SD 5.9 years, range 14-54 years), and the majority of the sample (2859/3186, 89.74%) were male. As previously mentioned, 1466 of 3186 gamers (46.01%) had a clear preference for MMORPGs and 872 of 3186 (27.37%) for MOFPSs. Among the MMORPG players, 240 gamers (16.37%) were female, whereas among MOFPS players only 20 gamers (2.29%) were female. Information regarding weekly game time and the proportion of gamers at high risk of problematic use is presented in
It was assumed that the parameters of the proposed model might be moderated by players’ game type preference and gender. Therefore, descriptive statistics and group differences regarding the variables included in the model are presented in
Results demonstrated that MOFPS players reported significantly higher scores on competition, skill development, and coping motives, whereas MMORPG players scored significantly higher on fantasy, escape, POG, the GSI, and recreation.
Weekly game time and proportion of gamers at high risk of problematic use for the overall sample, for males and females, and for multiplayer online first-person shooter (MOFPS) and massively multiplayer online role-playing games (MMORPGs) gamer types.
Gaming-related variables | Total sample, n (%) |
Gender, n (%) | Game type preference, n (%) | |||
|
|
Males |
Females |
MOFPS players |
MMORPG players |
|
|
|
|
|
|
|
|
|
<7 hours | 382 (12.00) | 328 (11.48) | 54 (16.5) | 125 (14.4) | 137 (9.36) |
|
7-14 hours | 772 (24.25) | 688 (24.09) | 84 (25.7) | 266 (30.5) | 298 (20.36) |
|
15-28 hours | 1102 (34.62) | 1001 (35.05) | 101 (30.9) | 308 (35.4) | 509 (34.77) |
|
29-42 hours | 639 (20.08) | 583 (20.41) | 56 (17.1) | 136 (15.6) | 355 (24.25) |
|
>42 hours | 288 (9.05) | 256 (8.96) | 32 (9.8) | 36 (4.1) | 165 (11.27) |
Gamers at high risk of problematic usea | 2.43 | 2.39 | 2.76 | 1.96 | 2.95 |
a The proportion of gamers at high risk of problematic use was calculated using the established cut-off point (ie, 66) suggested in a previous article [
Means, standard deviations (SD), and confidence intervals (CI) for multiplayer online first-person shooter (MOFPS) and massively multiplayer online role-playing games (MMORPGs) gamer types examined and for all players (MOFPS and MMORPG)a and effect sizes (Cohen’s
Psychopathology- and gaming-related variablesb | All (MOFPS & MMORPG) players (n=2338) | MOFPS players (n=872) | MMORPG players (n=1466) | Comparison of MOFPS and MMORPG players | ||||||||||
|
Mean (SD) | 95% CI | Mean (SD) | 95% CI | Mean (SD) | 95% CI |
|
|
|
|||||
Global Severity Index | 0.61 (0.61) | 0.59-0.64 | 0.57 (0.57) | 0.53-0.61 | 0.64 (0.64) | 0.61-0.67 | 2.82 (1986.8) | .005 | 0.12 | |||||
POGQ Total score | 36.13 (11.93) | 35.65-36.62 | 34.99 (11.73) | 34.21-35.77 | 36.82 (12.00) | 36.20-37.44 | 3.59 (2326) | <.001 | 0.15 | |||||
|
|
|
|
|
|
|
|
|
|
|||||
|
Escape | 1.93 (1.01) | 1.89-1.97 | 1.83 (0.93) | 1.76-1.88 | 2.00 (1.06) | 1.95-2.05 | 4.09 (2019.5) | <.001 | 0.18 | ||||
|
Coping | 2.51 (1.08) | 2.47-2.55 | 2.57 (1.12) | 2.50-2.64 | 2.47 (1.05) | 2.42-2.52 | 1.99 (1740.8) | .047 | 0.09 | ||||
|
Fantasy | 2.33 (1.13) | 2.28-2.38 | 2.05 (1.00) | 1.98-2.12 | 2.49 (1.17) | 2.43-2.55 | 9.67 (2057.9) | <.001 | 0.40 | ||||
|
Skill development | 2.23 (1.14) | 2.18-2.28 | 2.54 (1.21) | 2.46-2.62 | 2.04 (1.05) | 1.99-2.09 | 9.99 (1635.4) | <.001 | 0.44 | ||||
|
Recreation | 4.18 (0.87) | 4.15-4.22 | 4.12 (0.90) | 4.06-4.18 | 4.22 (0.85) | 4.18-4.26 | 2.75 (1737.7) | .006 | 0.11 | ||||
|
Competition | 2.39 (1.19) | 2.34-2.44 | 2.75 (1.22) | 2.67-2.83 | 2.17 (1.11) | 2.11-2.23 | 11.41 (1693.5) | <.001 | 0.50 | ||||
|
Social | 3.07 (1.20) | 3.02-3.12 | 3.07 (1.21) | 2.99-3.15 | 3.07 (1.19) | 3.01-3.13 | 0.15 (2332) | .88 | 0.00 |
a Only MOFPS and MMORPG gamers have been included because the MMORTS gamer group was very small and the rest of the players could not be differentiated regarding their game type preference.
b POGQ: Problematic Online Gaming Questionnaire; MOGQ: Motives for Online Gaming Questionnaire.
Means, standard deviations (SD), and confidence intervals (CI) for both genders and for the total sample and effects sizes (Cohen’s
Psychopathology- and gaming-related variablesa | Total sample (N=3186) | Males (n=2859) | Females (n=327) | Gender comparison | |||||||
|
Mean (SD) | 95% CI | Mean (SD) | 95% CI | Mean (SD) | 95% CI |
|
|
|
||
Global Severity Index | 0.62 (0.62) | 0.60-0.64 | 0.60 (0.61) | 0.58-0.62 | 0.77 (0.69) | 0.70-0.85 | 4.05 (387.4) | <.001 | 0.26 | ||
POGQ Total Score | 35.89 (11.85) | 35.48-36.30 | 35.86 (11.83) | 35.43-36.30 | 36.12 (12.31) | 35.78-37.46 | 0.37 (3171) | .71 | 0.02 | ||
|
|
|
|
|
|
|
|
|
|
||
|
Escape | 1.91 (0.99) | 1.88-1.94 | 1.87 (0.97) | 1.84-1.90 | 2.28 (1.14) | 2.16-2.40 | 6.22 (380.2) | <.001 | 0.39 | |
|
Coping | 2.50 (1.07) | 2.46-2.54 | 2.50 (1.08) | 2.46-2.54 | 2.56 (0.99) | 2.45-2.67 | 1.05 (3177) | .30 | 0.06 | |
|
Fantasy | 2.33 (1.12) | 2.29-2.37 | 2.28 (1.10) | 2.24-2.32 | 2.77 (1.23) | 2.64-2.90 | 6.80 (386.5) | <.001 | 0.42 | |
|
Skill development | 2.25 (1.14) | 2.21-2.29 | 2.26 (1.15) | 2.22-2.30 | 2.17 (1.00) | 2.06-2.28 | 1.54 (430.9) | .13 | 0.08 | |
|
Recreation | 4.18 (0.88) | 4.15-4.21 | 4.16 (0.88) | 4.13-4.19 | 4.29 (0.82) | 4.20-4.38 | 2.61 (416.5) | .009 | 0.15 | |
|
Competition | 2.41 (1.18) | 2.37-2.45 | 2.47 (1.18) | 2.43-2.51 | 1.80 (0.93) | 1.70-1.90 | 12.06 (456.4) | <.001 | 0.63 | |
|
Social | 3.04 (1.20) | 3.00-3.08 | 3.01 (1.20) | 2.97-3.05 | 3.32 (1.18) | 3.19-3.45 | 4.42 (3177) | <.001 | 0.26 |
a POGQ: Problematic Online Gaming Questionnaire; MOGQ: Motives for Online Gaming Questionnaire.
Results also demonstrated that female gamers scored significantly higher on fantasy, escape, social, and recreation motives as well as on the GSI of psychiatric symptoms, whereas male gamers reported significantly higher scores only on competition motive. The zero-order correlations between the components of the mediation model along with internal consistencies (Cronbach alpha) for all scales and subscales are presented in
All correlation coefficients were significant at least
Because both the BSI and the POGQ are multidimensional scales, a comprehensive correlation matrix is also provided as a multimedia appendix containing all subscales of the BSI, the MOGQ, and the POGQ (see
Zero-order correlations and Cronbach alphas (N=3186).
Psychopathology- and gaming-related variables | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Cronbach α |
1. Global Severity Index | .55 | .51 | .29 | .36 | .11 | –.01 | .20 | .09 | .97 |
2. POGQ Total Score |
|
.51 | .39 | .40 | .19 | .15 | .37 | .26 | .91 |
3. MOGQ Escape |
|
|
.60 | .61 | .24 | .18 | .27 | .29 | .87 |
4. MOGQ Coping |
|
|
|
.50 | .41 | .41 | .38 | .42 | .84 |
5. MOGQ Fantasy |
|
|
|
|
.30 | .29 | .28 | .34 | .83 |
6. MOGQ Skill development |
|
|
|
|
|
.23 | .38 | .45 | .89 |
7. MOGQ Recreation |
|
|
|
|
|
|
.23 | .35 | .78 |
8. MOGQ Competition |
|
|
|
|
|
|
|
.32 | .90 |
9. MOGQ Social |
|
|
|
|
|
|
|
|
.91 |
It was hypothesized that general psychiatric distress has both a direct and indirect effect (via the mediating effect of the 7 online gaming motives) on POG. Psychiatric distress was measured by the GSI and introduced in the model as a continuous observed variable. Problematic online gaming was measured by the summarized score of the POGQ and also introduced in the model as a continuous observed variable. Gaming motives were measured by the 7 factors of the MOGQ and introduced in the model as continuous latent variables. The proposed mediation models were tested with SEM. Because significant gender and gamer type differences were found in many of the variables included in the model, 3 different models were tested: (1) an overall model, (2) a separate multigroup analysis for the 2 gamer types, and (3) another multigroup analysis for males and females. The third model was carried out on the MMORPG gamer subsample because this group was the only one that had enough female players for comparison.
The overall model had an adequate fit to the data (χ2
343= 3217.8,
The model comparing MOFPS and MMORPG players also had an adequate fit to the data (χ2
726=3120.7; MOFPS: χ2
726=1343.6; MMORPG: χ2
726=1777.0,
The overall mediation model with standardized path coefficients and the explained variance of the endogen variables (
The mediation model and standardized path coefficients. Results of multigroup analysis and path coefficients across the 2 gamer types, multiplayer online first-person shooter (MOFPS) and massively multiplayer online role-playing games (MMORPGs), and the explained variance of the endogen variables (
Given that the proportion and sample size of female players were only considerable among MMORPG players, the third model was carried out only on this group (
Results relating to gender differences showed a significant difference in the escape → POG direct link between male and female MMORPG players. The standardized direct effect size of this link was .20 (
The mediation model and standardized path coefficients. Results of multigroup analysis and path coefficients across both genders (males/females) and the explained variance of the endogen variables (
The results of the present study suggest that psychiatric symptoms are both directly and indirectly (via escape and competition motives) negatively associated with POG. The mediator effect of gaming motives amounts to approximately 35% of the total effect. To the authors’ knowledge, this is the first study to statistically unravel the complex association between psychiatric symptoms, online gaming motivations, and problematic use using a SEM framework. The results relating to the associations between motives and problematic use are similar to previous findings in the psychological gaming literature. For instance, Yee [
Although advancement and competition motives are not the same, they are related to each other via a common connection to achievement and performance. Progressing rapidly and gaining power eventually become a way to be competitive, a way to complete goals successfully, and a way to defeat others. However, the advancement motive was developed in studies focusing on MMORPGs where “leveling up” is of a particular importance due to the persistence of the virtual world. On the other hand, the competition motive used in the present study is related to online games in general including games such as multiplayer online first-person shooter games or strategy games, in addition to MMORPGs. Other studies [
In addition to confirming previous findings in the gaming literature, as predicted the present study showed that the same 2 motives (escape and competition) mediated between psychiatric symptoms and problematic gaming. Playing games to escape everyday difficulties appears to be a motivating behavior that can ease psychiatric distress, and thus extends self-medication theory [
The second mediator variable in the present study was competition. Despite the fact that competition is usually considered a healthy and adaptive behavior, our findings suggest that in some cases it can also be a pathological factor. This has also been reported in the literature on problem gambling in which problem gamblers have been shown to be more competitive than nonproblem gamblers [
In contrast to prestudy expectations, coping did not mediate between psychiatric distress and problematic gaming. Earlier motivational research yielded the surprising finding that although highly correlated (.60), coping and escape are distinct motives [
The recreational use of online games was related neither to psychiatric symptoms nor to problematic use of games. This suggests that playing online games can be a healthy form of entertainment if it is used moderately and balanced with other leisure time activities (ie, sports). This result also serves as counterweight for media scaremongering that often exaggerates the potential dangers of video games [
This model in the present study was found to be invariant across game type preference (ie, MOFPS or MMORPG), but varied significantly between males and females in the case of MMORPG players. As expected, females were characterized by a stronger link between escape and POG, and also had higher escape scores than males. This latter result is in-line with Yee’s findings [
In addition, it is important to point out that the proportion of female players in the present sample is much lower than the proportion of women that play video games in the general population (ie, approximately 40% [
Despite the advantage of a large sample size, the self-selected and self-reported nature of the Hungarian-only data need to be taken into consideration when generalizing the results (especially because recent research has shown that self-selection of MMORPG players affects the sample’s representativeness [
The present study has some direct implications for prevention and treatment. There is little reason for parents, educators, and health professionals to be concerned or worried about the recreational use of online games. Neither should they necessarily be concerned about playing as a way to cope with day-to-day stress or tension. However, playing excessively as a way to avoid real-life problems or to defeat other players should receive attention because such motivations may lead to negative (addiction-like) real-life consequences. Therefore, exploring gaming motivations both on the individual and group level are likely to be helpful in the preparation of prevention and treatment programs concerning problematic gaming.
Correlation matrix including the nine Brief Symptom Inventory (BSI) subscales, the seven Motives for Online Gaming Questionnaire (MOGQ) subscales and the six Problematic Online Gaming Questionnaire (POGQ) subscales.
Brief Symptom Inventory
comparative fit index
full information maximum likelihood
Global Severity Index
massively multiplayer online role-playing games
multiplayer online first-person shooter
massively multiplayer online real-time strategy
Motives for Online Gaming Questionnaire
nonnormed fit index
problematic online gaming
Problematic Online Gaming Questionnaire
root mean square error approximation
structural equation modeling
standardized root mean square residuals
Tucker-Lewis fit index
Present work was supported by the Hungarian Scientific Research Fund (grant numbers: 83884 and 111938). Zsolt Demetrovics and Gyöngyi Kökönyei acknowledge financial support of the János Bolyai Research Fellowship awarded by the Hungarian Academy of Science.
KN, KG, and ZD designed the study; CA was responsible for data collection; OK and RU run the analysis; and OK, MDG, and ZD drafted the manuscript. All authors commented on the draft and contributed to the final version.
None declared.