Over the last forty years, the popularity of video games has significantly increased, making them an essential part of the global culture. At the beginning of the 21st century, the rise of the Internet and mobile phones launched video game use to new heights around the world, and, by 2023, the number of gamers is expected to reach 3.07 billion. In Spain, 18.2 million people played video games regularly in 2022 (Asociación Española de Videojuegos [AEVI], 2022; Newzoo International B.V., 2021). Video games are not only an important source of entertainment but also have a positive influence on several segments of our society. The video game industry plays a crucial role in the global economy (Tripp et al., Reference Tripp, Grueber, Simkins and Yetter2020), the use of video games in educational settings has demonstrated practical value in supporting learning and facilitating the transmission of knowledge (Villani et al., Reference Villani, Carissoli, Triberti, Marchetti, Gilli and Riva2018), and several studies have shown that video game-based interventions can have a positive impact in therapeutic and medical contexts (Halbrook et al., Reference Halbrook, O’Donnell and Msetfi2019; Xu et al., Reference Xu, Liang, Baghaei, Wu Berberich and Yue2020). Video games can also provide a wide variety of psychological benefits (Verheijen et al., Reference Verheijen, Stoltz, van den Berg and Cillessen2019; Wulansari et al., Reference Wulansari, Pirker, Kopf, Guetl, Auer, Hortsch and Sethakul2020), as enhancing intrinsic motivation, and fulfilling basic psychological needs like relatedness, autonomy, and competence (Przybylski et al., Reference Przybylski, Rigby and Ryan2010). Despite this, negative consequences have also been associated with their excessive use, with their alleged addictive potential being a growing concern for mental health professionals and scientific communities (Chen et al., Reference Chen, Oliffe and Kelly2018; Reed et al., Reference Reed, First, Billieux, Cloitre, Briken, Achab, Brewin, King, Kraus and Bryant2022).
In 2013, the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2013) proposed internet gaming disorder (IGD) as a condition in need of further study; since then, numerous studies have investigated its psychological, social, and cultural correlates. More recently, but not without controversy, gaming disorder has been included as a new condition in the eleventh International Classification of Diseases (World Health Organization, 2019), eliminating the term ‘Internet’, as associating the disorder with Internet-related problems was not considered strictly necessary. A recent meta-analysis estimated that problematic gaming affects 1.4%–3.3% of the population worldwide (Kim, Son, et al., Reference Kim, Son, Roh, Ahn, Kim, Shin, Chey and Choi2022). Notably, in Spain, the prevalence ranges from 1.7% to 8.3% among young adults (Beranuy et al., Reference Beranuy, Machimbarrena, Vega–Osés, Carbonell, Griffiths, Pontes and González-Cabrera2020; Buiza-Aguado et al., Reference Buiza-Aguado, Alonso-Canovas, Conde-Mateos, Buiza-Navarrete and Gentile2018) and from 1.8% (International Statistical Classification of Diseases and Related Health Problems, 11th Ed.; ICD–11 framework) to 3.1% (Diagnostic and Statistical Manual of Mental Disorders, 5th Ed.; DSM–5 framework) among adolescents (Nogueira-López et al., Reference Nogueira-López, Rial-Boubeta, Guadix-García, Villanueva-Blasco and Billieux2023). Also, problematic gaming is more prevalent in young men with family and interpersonal problems (Fumero et al., Reference Fumero, Marrero, Bethencourt and Peñate2020; Mihara & Higuchi, Reference Mihara and Higuchi2017; Stevens et al., Reference Stevens, Dorstyn, Delfabbro and King2021). Regarding psychological variables, impulsivity (Ding et al., Reference Ding, Sun, Sun, Chen, Zhou, Zhuang, Li, Zhang, Xu and Du2014; Ryu et al., Reference Ryu, Lee, Choi, Park, Kim and Choi2018) and antisocial traits (Anderson et al., Reference Anderson, Sakamoto, Gentile, Ihori, Shibuya, Yukawa, Naito and Kobayashi2008; Müller et al., Reference Müller, Janikian, Dreier, Wölfling, Beutel, Tzavara, Richardson and Tsitsika2015; T’ng et al., Reference T’ng, Ho, Sim, Yu and Wong2020), among other individual differences factors, have been associated with problematic gaming symptoms. A large body of evidence has also shown that certain patterns of motives underlying gaming behaviors can predict gaming intensity (i.e., frequency and duration) and problems (i.e., addiction symptoms or negative consequences; Cheah et al., Reference Cheah, Shimul and Phau2022; Király et al., Reference Király, Billieux, King, Urbán, Koncz, Polgár and Demetrovics2022; Melodia et al., Reference Melodia, Canale and Griffiths2022).
Motives are the declarative goals people perceive as guiding their instrumental behaviors (Volkow et al., Reference Volkow, Wise and Baler2017). They encompass intrinsic, extrinsic, and experiential motivation (Banyte & Gadeikiene, Reference Banyte and Gadeikiene2015), and include pleasure seeking, curiosity, interest (Gómez-Maureira & Kniestedt, Reference Gómez-Maureira and Kniestedt2019; Gunnell & Gaudreau, Reference Gunnell and Gaudreau2015), experiencing intense emotions (Hemenover & Bowman, Reference Hemenover and Bowman2018), escaping from reality, dealing with negative feelings such as boredom or loneliness (Larche & Dixon, Reference Larche and Dixon2021), socializing with other people (Nebel & Ninaus, Reference Nebel and Ninaus2022), or obtaining economic rewards or prizes (Johnson et al., Reference Johnson, Klarkowski, Vella, Phillips, McEwan and Watling2018). In the past, a large body of work has shown that the study of motives can help understand a variety of human behaviors, including substance and non-substance addictive behaviors (Barrada et al., Reference Barrada, Navas, Ruiz de Lara, Billieux, Devos and Perales2019; Bennett & Holloway, Reference Bennett and Holloway2017).
Process-based models of behavioral addiction have examined the dynamics of the relative utility attributed to positive and negative reinforcers in various stages of problematic gaming, including exposure, behavior consolidation, and dysregulated use (Perales et al., Reference Perales, King, Navas, Schimmenti, Sescousse, Starcevic, van Holst and Billieux2020). In this vein, a preponderant model such as the I-PACE (Interaction of Person-Affect-Cognition-Execution) includes motives among the moderating or mediating variables that explain the progression from recreational to dysregulated use of the Internet, including problematic video gaming (Brand et al., Reference Brand, Young, Laier, Wölfling and Potenza2016). The onset of problematic gaming is more likely to occur in individuals who display a heightened motivation linked to gaming (Young & Brand, Reference Young and Brand2017). Notably, although the I-PACE model does not explicitly delineate specific motives, it considers the way the experience of positive and negative rewards during gaming aligns with distinct motives. This alignment is proposed to underlie the temporal progression of the disorder. In the initial stages, certain motives are postulated to be implicated in reward anticipation, while in later stages, other motives are hypothesized to be associated with coping with negative affect (Brand et al., Reference Brand, Young, Laier, Wölfling and Potenza2016).
In view of this, numerous studies have investigated the motives why individuals engage in video game play (Király et al., Reference Király, Billieux, King, Urbán, Koncz, Polgár and Demetrovics2022; Reid, Reference Reid2012), and the circumstances in which these motives could be related to negative outcomes (Ng & Wiemer-Hastings, Reference Ng and Wiemer-Hastings2005; Wan & Chiou, Reference Wan and Chiou2006). Demetrovics et al. (Reference Demetrovics, Urbán, Nagygyörgy, Farkas, Zilahy, Mervó, Reindl, Ágoston, Kertész and Harmath2011) developed the most widely used measurement instrument to assess gaming motives: The Motives for Online Gaming Questionnaire (MOGQ). This instrument includes 27 items categorized into seven dimensions (social, escape, competition, coping, skill development, fantasy, and recreation motives). The use of this instrument has revealed that escape and fantasy motives are correlated with problematic gaming symptoms (Laconi et al., Reference Laconi, Pirès and Chabrol2017). To date, although translations of the MOGQ exist in other languages, to our knowledge, it has only been systematically validated in Turkish, Italian, Chinese, Korean, and Persian. Table 1 summarizes the main findings of these studies.
Validation Studies of the Motives for Online Gaming Questionnaire and Their Main Outcomes.
Studies examining the psychometric properties of MOGQ across languages and samples have yielded slightly different factor structures, with specific motives such as coping, sometimes being identified as a unique factor, and sometimes being merged with other motives (Melodia et al., Reference Melodia, Canale and Griffiths2022). Most of these studies have however disregarded connections between motives, preferences, and gaming-related behaviors such as loot-box purchasing, which have been linked to gaming-related harms in the context of progressive gaming-gambling convergence (King & Delfabbro, Reference King and Delfabbro2020; Wardle & Zendle, Reference Wardle and Zendle2021; Zendle & Cairns, Reference Zendle and Cairns2019).
Additional research suggests that some cultural factors (e.g., individualistic vs. collectivist cultures, or culture-specific expressions of achievement motives) have an impact on gaming motivation and its association with problematic gaming (Snodgrass et al., Reference Snodgrass, Dengah, Lacy and Fagan2013; Wang & Cheng, Reference Wang and Cheng2022), which reinforces the need to validate the scale in the language and culture in which it is used. In a similar vein, the role of motives in gaming behavior and their clinical implications can vary across genders (López-Fernández et al., Reference López-Fernandez, Williams and Kuss2019). Even though almost half of the population of gamers in Spain and other countries are females nowadays (AEVI, 2022; Newzoo International B. V., 2020), most MOGQ studies have focused on predominantly male samples, and studies in which females are fairly represented are much needed (López-Fernández et al., Reference López-Fernández, Mezquita, Griffiths, Ortet and Ibáñez2020b).
The main goal of this study is thus to test the psychometric properties of a Spanish version of the MOGQ (factorial structure, internal consistency, and validity) in a gender-balanced sample. Structure validity will be established through a combination of exploratory and confirmatory factor analyses (EFA/CFA). Convergent and divergent validity will be tested through the relationships between the MOGQ and problematic gaming symptoms, self-reported gaming time, impulsivity traits, and loot box consumption. Finally, we will explore gender differences among the previously cited variables.
Method
Participants and Procedure
Participants were recruited from four Spanish universities (the Catholic University of Murcia, the University of Granada, the University of Extremadura, and the University of Basque country). An email was sent to students inviting them to participate and complete an online survey. Participants were informed about the objectives of the study and the confidentiality of their responses. They gave consent to participate voluntarily. To encourage participation, five €15 gift cards were raffled at the end of the study. A total of 1,130 people participated in the study. The inclusion criteria were: (a) Being Spanish-speaking gamers, (b) reporting playing video games for at least 2 hours per week, and (c) being at least 18 years old. Participants were excluded if they did not meet the inclusion criteria, i.e., declaring they play less than two hours per week, being under 18 years old (n = 48), or providing invalid information (n = 104) (e.g., playing more than 7 days per week or more than 24 hours per day). The final sample consisted of 845 college students. The sample included 426 males (50.41%) and 417 females (49.35%), and participants were aged between 18 and 50 (M = 23.51, SD = 5.03) (please, see Table 2 for sample reported characteristics). All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study was approved by the Catholic University of Murcia Ethics Committee (CE031905). This study is part of a larger project, and another study utilizing the same database has already been published (Infanti et al., Reference Infanti, Valls-Serrano, Perales, Vögele and Billieux2023).
Sociodemographic and Game-Related Variables.
Measures
The Motives for Online Gaming Questionnaire (MOGQ) (Demetrovics et al., Reference Demetrovics, Urbán, Nagygyörgy, Farkas, Zilahy, Mervó, Reindl, Ágoston, Kertész and Harmath2011), aims to measure seven gaming motives (social, escape, competition, skill development, coping, fantasy, and recreation). A description of the seven dimensions can be found in Table 3. The questionnaire had 27 items scored on a five-point scale (1 = Almost never/Never to 5 = Almost always/Always). The original version showed good internal consistency and Cronbach’s alpha values range from .79 to .90. To develop the Spanish version, a native Spanish psychologist translated the questionnaire from English to Spanish. Next, an English philologist with Spanish linguistics skills translated the questionnaire back from Spanish to English. Finally, a group of three psychologists (two English and one Spanish), fluent in both languages, reviewed the final version. They concluded that the language and expressions were clear and adjusted to the linguistic context.
Questionnaire Variables of the Motives for Online Gaming Questionnaire and the Impulsivity Scale.
Note. UPPS-P = Urgency (negative), Premeditation (lack of), Perseverance (lack of), Sensation Seeking, Urgency (positive), Impulsive Behavior Scale.
The Internet Gaming Disorder test (IGD–20) (Spanish version by Fuster et al. [Reference Fuster, Carbonell, Pontes and Griffiths2016], original English version by Pontes et al. [Reference Pontes, Király, Demetrovics and Griffiths2014]) evaluates the presence of IGD symptoms based on the DSM–5 framework and the component model of addiction by Griffiths (Reference Griffiths2005). The scale consists of 20 items that are answered using a 5-point Likert scale (1 = Strongly disagree to 5 = Strongly agree). The Spanish version of the IGD–20 showed good psychometric properties and Cronbach’s alpha was .87 (Fuster et al., Reference Fuster, Carbonell, Pontes and Griffiths2016). In the present sample, Cronbach’s alpha (computed using Spearman correlations) was .91 for the total score.
The short Spanish version of the Urgency (negative), Premeditation (lack of), Perseverance (lack of), Sensation Seeking, Urgency (positive), Impulsive Behavior Scale (UPPS-P Impulsivity Scale) (Spanish version by Cándido et al. [Reference Cándido, Orduña, Perales, Verdejo-García and Billieux2012], original French version by Billieux et al., [Reference Billieux, Rochat, Ceschi, Carré, Offerlin-Meyer, Defeldre, Khazaal, Besche-Richard and van der Linden2012]) assesses five impulsivity traits (negative urgency, lack of premeditation, lack of perseverance, sensation seeking, and positive urgency). Table 3 provides a description of these dimensions. The scale consists of 20 items scored on a Likert scale (1 = Strongly agree to 4 = Strongly disagree). The original version showed good internal reliability, with Cronbach’s alpha ranging from .61 and .82. For our current sample, a Cronbach’s alpha for each dimension was computed using Spearman correlations and were .79 for lack of perseverance, .76 for lack of premeditation, .81 for sensation seeking, .66 for positive urgency, and .82 for negative urgency. In the current study, we merged positive and negative urgency as recent studies suggest that they belong to a single construct (Billieux et al., Reference Billieux, Heeren, Rochat, Maurage, Bayard, Bet, Besche-Richard, Challet-Bouju, Carré, Devos, Flayelle, Gierski, Grall-Bronnec, Kern, Khazaal, Lançon, Lannoy, Michael, Raffard and Baggio2021). The Cronbach’s alpha of general urgency was .81.
Participants were asked about different gaming habits, including: (a) How many days per week they dedicated to gaming, (b) how many hours per day they dedicated to gaming, and (c) how much money they spent on loot boxes in the last month.
Data Analytic Strategy
In the present paper, we separated the whole dataset into two different samples. One sample was used for the purpose of an exploratory factor analysis (EFA) (n = 272) while the other sample was used to test the structure validity through a confirmatory factor analysis (CFA) (n = 573). The decision to combine EFA and CFA obeyed to several reasons. First, previous studies have favored slightly different factor structures (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020; Kim & Kang, Reference Kim and Kang2021; Wu et al., Reference Wu, Lai, Yu, Lau and Lei2017), so it is unclear which of them should be favored a priori in a strictly theory-driven CFA. And second, the original seven-factor model was derived from samples that primarily consisted of males (Ballabio et al., Reference Ballabio, Griffiths, Urbán, Quartiroli, Demetrovics and Király2017; Dowran et al., Reference Dowran, Yekta and Aghaie2022; Hamzehzadeh et al., Reference Hamzehzadeh, Sangchooli, Farnam, Rafiemanesh, Shadloo, Ghani, Jobehdar, Amin-Esmaeili, Rahimi, Demetrovics, Király and Rahimi-Movaghar2022; Marino et al., Reference Marino, Canale, Vieno, Caselli, Scacchi and Spada2020), which could compromise its generalizability. The combination of EFA and CFA prevents the structure selection from being influenced by theoretical preconceptions, and it has been successfully applied in previous psychometric MOGQ studies (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020; Kim & Kang, Reference Kim and Kang2021;).
We followed Hair et al.’s (Reference Hair, Black, Babin and Anderson2014) guidelines for EFA and CFA analyses. The separation of the dataset was done to have a ratio of ten observations per variable (e.g., a minimum of 270 observations) for the EFA. The two samples did not present significant differences in demographic data (education level, age, gaming hours per day). Before performing the EFA, we assessed sample adequacy using the Kaiser-Meyer-Olkin (KMO) and Bartlett’s sphericity test. To select the number of factors to retain in our model, we used scree plot, eigenvalue, and parallel analysis methods. We then performed an EFA using the maximum likelihood method with Promax rotation (as the gaming motives might be related and are thus not presumed to be orthogonal) and 100 iterations. These options were also used in the Korean and Turkish validation studies of the MOGQ Scale (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020; Kim & Kang, Reference Kim and Kang2021). As the EFA sample size is larger than 250 and smaller than 350, items having a factor loading value higher than .35 were identified as significant (Hair et al., Reference Hair, Black, Babin and Anderson2014). When an item had more than one significant factor loading or no significant factor loadings, the item was discarded. For the CFA, we used a robust variant of maximum likelihood as an estimator (MLR). Several CFAs were computed to test different competing models and were then compared using a series of fit measures. Models tested included the original seven-factor model, a second-order model, a model created using the EFA factor structure, a bifactor model consisting of a general motivation and the seven original factors (uncorrelated), and a bifactor model consisting of a general motivation and the factors identified by EFA (uncorrelated). Bifactor models have been included following evidence from the Chinese validation of the MOGQ (Wu et al., Reference Wu, Lai, Yu, Lau and Lei2017).
Model Chi-Square, comparative fit index (CFI), Tucker Lewis index (TLI), root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), and Akaike information criteria (AIC) were used for the models’ evaluation. Good fit model references are a normed χ² lower than 5.0, a CFI value higher than .95, a TLI value higher than .95, a SRMR value lower than .08, a RMSEA value lower than .08, and an AIC smaller value (Hooper et al., Reference Hooper, Coughlan and Mullen2008). To assess reliability, we computed the Omega and Cronbach’s alpha coefficients using the CFA sample. Invariance across genders was tested in the CFA sample using multiple group CFA (MGCFA) and chi-square difference tests to compare configural and metric models, followed by a comparison between metric and scalar models. The criterion was the non-significance of the chi-square difference test. We then assessed the convergent and divergent validity by observing the partial correlations between the scores of the MOGQ Scale, the IGD–20 Scale, the short UPPS-P, and the estimated gaming hours per day while controlling for gender. Finally, we observed if there were significant differences in MOGQ Scale scores as a function of gender and loot box consumption. These analyses were computed using R (v4.0.3) and the following packages: haven (Wickham et al., Reference Wickham, Miller and Smith2022), tidyverse (Wickham et al., Reference Wickham, Averick, Bryan, Chang, D’Agostino McGowan, François, Grolemund, Hayes, Henry, Hester, Kuhn, Pedersen, Miller, Bache, Müller, Ooms, Robinson, Seidel, Spinu and Yutani2019), corrplot (Wei & Simko, Reference Wei and Simko2021), psych (Revelle, Reference Revelle2022), lavaan (Rosseel, Reference Rosseel2012), and alookr (Ryu, Reference Ryu2022). The dataset and the code are available via the following Open Science Framework (OSF).Footnote 1
Results
Shapiro-Wilk normality tests showed non-normal distribution for the MOGQ total score and its factors.
Exploratory Factor Analyses
Premise analyses confirmed the sampling adequacy of the EFA sample, with good KMO (measure of sampling adequacy, MSA = .9) and a significant Bartlett’s sphericity test (χ² = 5,270.655, df = 351, p < .001). Eigenvalues rule (value greater than 1), scree plot, and parallel analyses suggested the presence of six factors. EFA results are reported in Table 4. Factor 1 (Items 2, 4, 9, 11, 16, 23, 25) is a mix of escape and coping dimensions present in the original scale. Factor 2 (Items 3, 10, 17, 24) corresponds to the competition dimension. Factor 3 (Items 7, 14, 21) corresponds to the recreation dimension. Factor 4 (Items 5, 12, 19, 26) corresponds to the skill dimension. Factor 5 (Items 1, 8, 15) corresponds to the social dimension. Finally, Factor 6 (Items 6, 13, 20) corresponds to the fantasy dimension. Items 18 (“…because it helps me to channel my aggression”) and 22 (“…because playing gives me company”) did not load on any retained factors and were thus deleted. Item 27 (“…because I can be in another world”) cross-loaded onto Factors 1 and 6 and was also deleted from the EFA model.
Exploratory Factor Analysis of the MOGQ (EFA sample, n = 272).
Note. EFA = exploratory factor analysis.
** Significant at p < .001 level. * Significant at p < .05 level.
Confirmatory Factor Analyses
Several CFAs were conducted to compare the goodness of fit of the different models. In total, five models were tested. Model 1 is derived from the original MOGQ paper (Demetrovics et al., Reference Demetrovics, Urbán, Nagygyörgy, Farkas, Zilahy, Mervó, Reindl, Ágoston, Kertész and Harmath2011) and consists of seven factors correlated to each other without the presence of a general factor. Model 2 is a second-order model where the items are related to the seven factors which are themselves related to a general factor. Model 3 consists of the six factors derived from the EFA without any general factor. Model 4 is a bifactor model where all the items are related to a general motivation and to seven uncorrelated factors. Finally, Model 5 is a bifactor model where all the items are related to a general motivation and to six uncorrelated factors (as identified by EFA).
Globally, bifactor models (Model 4 and 5) presented the best fit, suggesting the presence of a general factor related to all the items and several uncorrelated factors. Among them, Model 5 (EFA bifactor model, a bifactor model consisting of a general motivation and six uncorrelated factors as identified in EFA) satisfied most of the fit indices (χ²/df = 3.267, CFI = .938, TLI = .925, RMSEA = .068 [.062, .073], SRMR = .065, AIC = 35,981.793) (please, see Table 5). Even though Model 5 had incremental fit values (CFI and TLI) lower than .95, it is important to note that a threshold of .90 was used in the Korean and Turkish validation for these values (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020; Kim & Kang, Reference Kim and Kang2021). We thus conclude that the CFI and TLI values could be acceptable. Invariance analysis across genders (n male = 290; n female = 281) yielded no significant chi-square difference tests. No differences have been found when comparing configural invariance and metric invariance models (Δχ² = 53.212, df = 41, p = .096), and for metric and scalar invariance models (Δχ² = 22.103, df = 17, p = .181).
Model Fit of the Measurement Models for MOGQ Items (n = 573).
Note. CFA = confirmatory factor analysis; RMSEA = root-mean-square error of approximation; SRMR = standardized root mean square residual; AIC = akaike information criteria; CFI = comparative fit index; TLI = Tucker Lewis index; EFA = exploratory factor analysis.
EFA Bifactor Model Reliability
Because the EFA bifactor model presents the best fit, we computed the omega-hierarchical (ωh) for the general factor (general motivation) (Flora, Reference Flora2020). The omega-hierarchical is especially relevant when it comes to measuring the reliability of the general factor of a bifactor model. Indeed, it reflects the proportion of the total-score variance as a result of a general factor, regardless of the multidimensional aspect of the scale (Flora, Reference Flora2020). For the specific factors, we reported the omega (ω) which represents the stability of the specific factors but also the general factor (Reise, Reference Reise2012). The omega-hierarchical for the general factor was .79 (α = .93). Regarding the specific factors, omegas were .79 for social (α = .80), .81 for escape/coping (α = .91), .79 for competition (α = .85), .84 for skill (α = .92), .82 for fantasy (α = .85), and .70 for recreation (α = .82). Thus, the scale presents an acceptable reliability for the general factor and the specific factor scores.
EFA Bifactor Model and External Variables
Using the EFA bifactor model, we explored how impulsivity traits, IGD symptoms, and gaming hours per day correlated with the MOGQ’s dimensions and general motivation. Results are displayed in Table 6. All MOGQ’s subscales showed significant correlations with daily gaming hours (correlations between rs = .14, p < .001, with competition motivation, to rs = .35, p < .001, with general motivation). All gaming motives positively correlated with IGD–20 total score, except recreation motives. In general, correlations between impulsivity traits and motives to play were small and non-significant. Significant correlations ranged from –.07 (p = .041, between lack of premeditation and recreation motivation) to .16 (p < .001, between urgency and fantasy motivation). Finally, loot box consumption was linked to all specific gaming motives, the highest relationship being observed with the competition dimension (rpb = .25, p < .001), and the lowest with fantasy dimension (rpb = .08, p = .021).
EFA Model Spearman’s Partial Correlations with Impulsivity Traits, Gaming Disorder Symptoms as Measured by the IGD–20, Gaming Hours per day, Loot Boxesa
Note. Conditioned on Gender. p-value adjustment method: Benjamini & Hochberg (Reference Benjamini and Hochberg1995). EFA = exploratory factor analysis; MOGQ = Motives for Online Gaming Questionnaire; IGD = Internet Gaming Disorder; UPPS-P = Urgency (negative), Premeditation (lack of), Perseverance (lack of), Sensation Seeking, Urgency (positive), Impulsive Behavior Scale.
a Point-Biserial method instead of Spearman.
Regarding the gender comparison (please, see Table 7), we found males to score significantly higher on general motivation but also in the competition, recreation, and skill dimensions. No significant differences were found for the escape/coping, social, and fantasy dimensions.
Wilcoxon Mann Whitney Test between Female and Male on MOGQ, IGD–20, UPPS-P, and Gaming Hours per Day.
Note. MOGQ = Motives for Online Gaming Questionnaire; IGD = Internet Gaming Disorder Test; UPPS-P = Urgency (negative), Premeditation (lack of), Perseverance (lack of), Sensation Seeking, Urgency (positive), Impulsive Behavior Scale.
Discussion
The main aim of this study was to establish the factor structure of the Spanish version of the MOGQ in a large convenience sample of college students regularly participating in video gaming activities. The combination of EFA and CFA results supports a bifactor model consisting of a general motivation factor and six uncorrelated subfactors (a merged escape/coping motives subfactor, and separated subfactors for competition, recreation, skill, social, and fantasy motives); the final scale showed satisfactory psychometric properties. Omega and Cronbach’s alpha coefficients showed acceptable or good reliability and the correlations with time spent on video games and IGD symptoms support the construct validity of the scale.
The overlap between coping and escape motives is one of the most noteworthy results of the present study and converges with part of the existing literature (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020; Kim & Kang, Reference Kim and Kang2021). Relatedly, some items included in the coping factor in previous studies present questionable levels of robustness and representativity (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020; Kim & Kang, Reference Kim and Kang2021; Wu et al., Reference Wu, Lai, Yu, Lau and Lei2017). For instance, Item 18 ("…because it helps me to channel my aggression"), did not load significantly on any factor and was subsequently removed from the final version in the present study. This item was also removed from the coping factor in the Chinese version (Wu et al., Reference Wu, Lai, Yu, Lau and Lei2017), in which it aligned with skill development and recreational motives. In the Turkish version (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020), the same item negatively loaded onto the recreation factor, and positively onto the coping/escape factor. And finally, in the Korean version (Kim & Kang, Reference Kim and Kang2021) it loaded onto the fantasy factor. We did not find any other unexpected loadings for items within the coping factor, although the Korean version (Kim & Kang, Reference Kim and Kang2021) excluded Items 4 and 11, and the Chinese version (Wu et al., Reference Wu, Lai, Yu, Lau and Lei2017) found Item 4 to cross-load onto the coping and recreation factors.
The superiority of six-factor models relative to their seven-factor counterparts –with coping and escape motives loading on a single factor– supports the theoretical model proposed by Demetrovics et al. (Reference Demetrovics, Urbán, Nagygyörgy, Farkas, Zilahy, Mervó, Reindl, Ágoston, Kertész and Harmath2011) for the original version of the MOGQ and closely replicates the findings from the Turkish version (Evren et al., Reference Evren, Evren, Dalbudak, Topçu and Kutlu2020). Although a six-factor structure was also reported for the Korean version of the MOGQ this was due to the disappearance of the coping factor (Kim & Kang, Reference Kim and Kang2021). In contrast, the Italian, Hungarian, and Persian validations leaned towards the seven-factor structure (Ballabio et al., Reference Ballabio, Griffiths, Urbán, Quartiroli, Demetrovics and Király2017; Demetrovics et al., Reference Demetrovics, Urbán, Nagygyörgy, Farkas, Zilahy, Mervó, Reindl, Ágoston, Kertész and Harmath2011; Dowran et al., Reference Dowran, Yekta and Aghaie2022; Hamzehzadeh et al., Reference Hamzehzadeh, Sangchooli, Farnam, Rafiemanesh, Shadloo, Ghani, Jobehdar, Amin-Esmaeili, Rahimi, Demetrovics, Király and Rahimi-Movaghar2022). Although general East-West cultural differences have been proposed as an explanation for this divergence (Wu et al., Reference Wu, Lai, Yu, Lau and Lei2017), our results are closer to those of the Chinese validation than to those from European and Middle East countries, and thus fail to support this explanation.
Complementary evidence points to the possible role of the distribution of individual differences in the samples used for different studies. Motives are tightly related to gaming preferences, and these to other individual characteristics and sociodemographic variables (Gómez-Gonzalvo et al., Reference Gómez-Gonzalvo, Molina and Devís-Devís2020, Kim, Nam, et al., Reference Kim, Nam and Keum2022; Rehbein et al., Reference Rehbein, Staudt, Hanslmaier and Kliem2016; Veltri et al., Reference Veltri, Krasnova, Baumann and Kalayamthanam2014). More specifically, some of these studies report differences in gaming preferences between males and females (Gómez-Gonzalvo et al., Reference Gómez-Gonzalvo, Molina and Devís-Devís2020; Phan et al., Reference Phan, Jardina, Hoyle and Chaparro2012; Veltri et al., Reference Veltri, Krasnova, Baumann and Kalayamthanam2014), with male video gamers tending to be more competitive than women and showing a preference for games that require navigation and attentional skills (Buser et al., Reference Buser, Cappelen, Gneezy, Hoffman and Tungodden2021; Tungodden & Willén, Reference Tungodden and Willén2022). Although we are aware that these results must be interpreted with caution (as video game studies is a male-dominated area with a high presence of gender stereotypes and women are often invisibilized (Lange et al., Reference Lange, Wühr and Schwarz2021, Wohn et al., Reference Wohn, Ratan, Cherchiglia and Fang2020), male participants in our study scored significantly higher than female ones in the competition, recreation, and skill motives.
With regard to the possible role of gender, an important coincidence between the Chinese validation and ours, as well as a methodological advantage relative to other studies, is the almost gender-balanced sample composition in both (54.6% males to 45.4% females and 50.41% males to 49.35% females, respectively). This makes our results more representative of the factor structure of motives in the general population of gamers. Although, in the present study, we found measurement invariance across genders, additional factor invariance analyses, as well as gender-informed analyses for other individual characteristics and demographics, are still needed.
The overlap between coping and escape motives is also supported by their consistent and shared associations with gaming-related problems (Griffiths et al., Reference Griffiths, van Rooij, Kardefelt-Winther, Starcevic, Király, Pallesen, Müller, Dreier, Carras, Prause, King, Aboujaoude, Kuss, Pontes, Lopez Fernandez, Nagygyorgy, Achab, Billieux, Quandt and Demetrovics2016; Kuss et al., Reference Kuss, Louws and Wiers2012) and other problematic behaviors (Estévez et al., Reference Estévez, Jauregui, Lopez-Gonzalez, Macia, López, Zamora, Onaindia, Granero, Mestre-Bach, Steward, Fernández-Aranda, Gómez-Peña, Moragas, Mena-Moreno, Lozano-Madrid, del Pino-Gutiérrez, Codina, Testa, Vintró-Alcaraz and Jiménez-Murcia2021; Froushani & Akrami, Reference Froushani and Akrami2017; Jauregui et al., Reference Jauregui, Onaindia and Estévez2017; Lee-Winn et al., Reference Lee-Winn, Mendelson and Johnson2018). This convergence of internal and external validity results has led some authors to suggest that coping and escape behaviors are underpinned by a common mechanism (see Barrada et al., Reference Barrada, Navas, Ruiz de Lara, Billieux, Devos and Perales2019). According to this view, problematic gaming would work as an overt emotion regulation strategy and would be maintained by negative reinforcement (Kardefelt-Winther, Reference Kardefelt-Winther2017). Still, an open question remains regarding the potential distinction between the healthy and unhealthy use of video gaming to deal with stress and other aversive states. On the one hand, as noted earlier, coping/escape motives are consistently associated with the severity of a range of potentially problematic activities (see also Chang & Lin, Reference Chang and Lin2019; de Hesselle et al., Reference De Hesselle, Rozgonjuk, Sindermann, Pontes and Montag2021; Kircaburun et al., Reference Kircaburun, Demetrovics, Griffiths, Király, Kun and Tosuntaş2020; Laconi et al., Reference Laconi, Pirès and Chabrol2017; López-Fernández et al., Reference López-Fernández, Mezquita, Griffiths, Ortet and Ibáñez2020a; Melodia et al., Reference Melodia, Canale and Griffiths2022; Rafiemanesh et al., Reference Rafiemanesh, Farnam, Sangchooli, Rahimi, Hamzehzadeh, Ghani, Jobehdar, Amin-Esmaeili, Shadloo, Demetrovics, Király and Rahimi-Movaghar2022). On the other hand, many individuals report using video gaming as a buffer against life stress, with video gaming positively contributing to their well-being as a consequence of this (Bourgonjon et al., Reference Bourgonjon, Vandermeersche, De Wever, Soetaert and Valcke2016; Ceranoglu, Reference Ceranoglu2010). In the words of Gee (Reference Gee2007) “there are escapes that lead nowhere, like hard drugs, and escapes like […] gaming that can lead to the imagination of new worlds, new possibilities to deal with those perils and pitfalls, new possibilities for better lives for everyone” (p. 1741). In this direction, Giardina et al. (Reference Giardina, Starcevic, King, Schimmenti, Di Blasi and Billieux2023) proposed to distinguish between escapism (a temporal and adaptive process with positive effects for the individuals) and escape (a maladaptive avoidant coping strategy used to disconnect from reality).
In any case, further research (and maybe a less biased approach; see Granic et al., Reference Granic, Lobel and Engels2014) is needed to understand the connection between coping/escape and video gaming problems. Tentatively, some factors beyond motives could explain why negatively reinforced video gaming acquires a disproportionate salience and value in some individuals, relative to other activities, in a way that ends up being harmful (see Perales et al., Reference Perales, King, Navas, Schimmenti, Sescousse, Starcevic, van Holst and Billieux2020). Among these factors, there is a pre-existing impoverished or not sufficiently rewarding environment, lack of skills to satisfy the same motives using other activities or strategies, emotion regulation problems, or impulsivity. The importance of the latter is indeed supported by the observed significant association of urgency with the specific motives most strongly related to IGD–20 scores. However, urgency-motives correlations remain weak (see also Zsila et al., Reference Zsila, Orosz, Bőthe, Tóth-Király, Király, Griffiths and Demetrovics2018), which supports the divergent validity of the motives scale and suggests that impulsivity could play a mediating or moderating role in problematic video game use, as it seems to occur in other problematic behaviors (Adams et al., Reference Adams, Kaiser, Lynam, Charnigo and Milich2012; Marzilli et al., Reference Marzilli, Cerniglia, Ballarotto and Cimino2020).
With regard to the rest of the motives, fantasy was distinct enough from coping/escape to confirm a separated motivational dimension, but the two motive types are aligned in terms of their relationship with urgency and gaming-related problems. The most relevant difference between the two is the specific significant (but weak) association between fantasy motives and sensation seeking. This general parallelism could indicate that fantasy and coping/escape motives can have a similar negative function. At the other end, recreation motives are least strongly related with video gaming problems, replicating previous results (Bäcklund et al., Reference Bäcklund, Elbe, Gavelin, Sörman and Ljungberg2022; Kircaburun et al., Reference Kircaburun, Demetrovics, Griffiths, Király, Kun and Tosuntaş2020; Moudiab & Spada, Reference Moudiab and Spada2019). Again, this converges with evidence from other behavioral domains in which fun, thrill, or entertainment motives are unrelated to problems derived from those activities, or even protect against them (see Barrada et al., Reference Barrada, Navas, Ruiz de Lara, Billieux, Devos and Perales2019).
Finally, and concerning gaming habits, we found loot box consumption to be significantly (but weakly) related to all MOGQ dimensions, with the two highest correlations observed for competition (rpb = .25, p < .001) and skill (rpb = .22, p < .001), and the two lowest for fantasy (rpb = .08, p = .021) and coping/escape (rpb = .10, p = .03). In other words, there is little overlap between predictors of video gaming problems and predictors of loot box use. Loot box use seems to be motivated by the desire to perform better or to be more competitive in the game, which makes sense in view that loot box contents in some popular games are often not only cosmetic but also impactful on gaming performance (e.g., player packs in FIFA Ultimate Game; Lemmens, Reference Lemmens2022). Consequently, problematic loot box use might be due to other mechanisms than those linking problematic gaming with negative reinforcement motives (see, for example, Li et al., Reference Li, Mills and Nower2019, for a depiction of the specific contribution of loot box consumption to the comorbidity between gambling and gaming problems).
The present study is not without limitations. First, participants were recruited from university colleges, which limit the representativeness of the sample and the generalizability of results. Second, the cross-sectional design of the study does not allow establishing the directionality of links between the variables of interest. Third, data were collected using self-report questionnaires, meaning they could be affected by a desirability bias or question misinterpretation. Fourth, the participants in this study are heterogeneous and belong to different segments within the gaming community in terms of gaming intensity. Therefore, generalizing the results to specific player profiles (e.g., high-intensity gamers) is not warranted. Finally, most of the correlations found to be significant were so because of the large sample used, but they are mostly small-sized and range in a narrow interval. In view of that, their practical and theoretical significance should be interpreted cautiously, especially in the case of tentative explanations for comparisons between correlations (i.e., differential correlation patterns of motives with different outcomes).
The main strengths of the study include the large sample, which allows a robust set of associations, while the combined use of EFA and CFA ensures that factor structure selection is not biased by theoretical preconceptions. Theory-driven models are contrasted against data-driven counterparts, where the latter are cross-validated across subsamples. The superiority of six-factor solutions over seven-factor ones can thus be considered robust for the population of reference; the MOGQ is a reliable and valid instrument to assess motives to play online games. In theoretical terms, our study also confirms that gaming motives are related to gaming disorder symptoms and highlights the comparative importance of coping/escape and fantasy motives. Furthermore, this study confirms that gender and urgency could be relevant to understand how motives impact video gaming and its outcomes.
Authorship credit
Both A. Infanti and C. Valls-Serrano contributed equally to this article and should be considered co-first authors. The authors confirm contribution to the paper as follows: Conceptualization: Carlos Valls-Serrano; formal analysis: Alexandre Infanti; investigation: Carlos Valls-Serrano; methodology: Alexandre Infanti; writing-original draft: Carlos Valls-Serrano, Alexandre Infanti and José C. Perales; writing-review & editing: Joël Billieux and José C. Perales; supervision: Joël Billieux and José C. Perales.
Data sharing
Data supporting the findings of this study are available in the following OSF link https://osf.io/jk94v/?view_only=118f5cee309a4d9aa48fdf1dde1392e4.
Funding statement
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflict of interest
None.
