1. Introduction
Internet gaming has become growingly popular across age, sex, and culture [
1,
2]. Adolescents are particularly vulnerable to psychosocial and psychological problems associated with Internet gaming [
3]. In 2013, the Fifth Edition of Diagnostic and Statistical Manual of Mental Disorders (DSM-5) considered Internet gaming disorder (IGD) a mental health disorder; its nine diagnostic criteria were similar to those of substance dependence [
4]. In 2018, the 11th edition of the International Classification of Diseases and Related Health Problems (ICD-11) further included online and offline gaming disorder as a disease [
5]. It is imperative to understand determinants of IGD.
In particular, maladaptive cognitions are determinants of addictive behaviors such as IGD [
6], Internet addiction [
7], and pathological gambling [
8,
9,
10]. Maladaptive cognitions on gaming include two categories: gaming related cognitive distortions and potentially harmful beliefs about the self in relation to gaming [
6,
11]. Potentially effective behavioral interventions for reduction of IGD (e.g., cognitive-behavioral therapy [
12,
13]) need to modify related cognitions [
14]. In literature, emerging research has attempted to identify various cognitions of Internet gaming that are factors of IGD [
6,
15,
16], as exemplified by a recent systematic review of 36 such studies [
6]. The review proposed a 4-factor model that summarized the domains of potential cognitive determinants of IGD [
6]: (1) overvaluation refers to the beliefs about reward value and tangibility of Internet gaming (e.g., perceptions that gaming rewards are as important as anything else); (2) maladaptive rules as a domain refers to justification of playing Internet games despite negative consequences; (3) gaming for self-esteem refers to over-reliance on Internet gaming to meet self-esteem needs; (4) gaming for social acceptance refers to using Internet gaming as a way to gain social acceptance [
6]. Based on these four domains, the authors developed the Internet Gaming Cognition scale (IGCS), which discriminated between highly engaged adolescent gamers with and without IGD [
17]. Some previous studies reported specific cognitive factors (e.g., preoccupation or cognitive salience) of IGD, but such factors are not specific to Internet gaming [
6,
18] or were limited to a relatively narrow range of cognitions related to Internet gaming [
19]. IGCS has the strength of including factors that are specifically related to Internet gaming. Furthermore, it has considered such cognitive factors comprehensively, as it was derived from a systematic review of cognitive factors that were related to Internet gaming. Thus, the model has hence good potential for improving understanding and interventions regarding adolescent IGD. Cross-cultural validation of IGCS is warranted.
The present validation study hence investigated the psychometric properties of the Chinese version of IGCS among adolescents in mainland China. Such properties included the factor structure, internal consistency, test-retest reliability, and correlations with some multi-dimensional external variables on IGD, gaming time, impulsivity, self-control, and perceived importance of Internet gaming as primary sources of one’s self-esteem and social acceptance. IGD and gaming time were included as external variables, as maladaptive cognitions were correlated with addictive behaviors. We expect those with strong impulsivity and/or weak self-control to be more likely than others to adopt maladaptive rules. Perceptions that Internet gaming being a source of self-esteem and social acceptance were expected to be associated with gaming for esteem and gaming for social acceptance.
4. Discussion
The present study detected relatively high IGD prevalence of about 11% among all participants and about 16% among Internet gamers. There is a need to reduce IGD among Chinese adolescents. It is important to understand various domains of maladaptive cognitions about Internet gaming and their relationships with IGD, as such cognitions are potentially modifiable and understanding on such cognitions may facilitate development of effective behavioral interventions for reduction of IGD. The findings of this study contribute to that end as all the three newly identified cognitive factors of C-RIGCS were significantly associated with IGD. The two constructs of perceived urges for playing Internet games and perceived unwillingness to stop playing without completion of gaming tasks match with the first dimension of maladaptive gaming-related cognitions that refers to distorted or irrational cognitions promoting and maintaining excessive or pathological gaming [
6,
7], while the construct of perceived rewards of Internet gaming fits into the second dimension of maladaptive gaming-related cognitions that refers to potentially harmful beliefs about the self in relation to gaming [
11]. Comparisons of the strengths of associations between different domains of cognitions of Internet gaming and IGD can further provide insights on the relative contributions of various types of cognitions on development of IGD, and guide design of effective intervention contents. In our case, cognitions about perceived urges for playing Internet games showed the strongest positive associations with IGD symptoms and gaming time, compared to the subscales of perceived rewards of Internet gaming and perceived unwillingness to stop playing without completion of gaming tasks. Future interventions may emphasize such cognitions.
As a prerequisite of designing interventions to modify cognitions, it is warranted to develop measurement tools for assessment of domains of cognitions related to Internet gaming in a comprehensive manner. Development of the original IGCS was a good starting point, as it was carefully derived from a systematic review and has been applied to design a potentially effective pilot intervention study that reduced IGD symptoms. The scale is, however, new and has not been subjected to cross-cultural validation. The CFA of the present study did not confirm the 4-factor structure of the original IGCS, which has not been subjected to CFA, although some favorable psychometric properties (e.g., internal reliability and external validity) were reported [
17]. The lack of goodness-of-fit in the present study might be due to cultural and contextual differences. For instance, the level of tangibility and nature of negative consequences of Internet gaming may differ across countries; investigation of the differences is beyond the scope of this study. Nevertheless, some inferior psychometric properties of the original IGCS were also observed. For instance, the correlations among the four subscales of the original IGCS were very high (0.88 to 0.98), implying substantially overlapping constructs. For instance, social acceptance may contribute to self-esteem, while gaining self-esteem/social acceptance via Internet gaming may lead to perceptions on maladaptive rules. Relatedly, although the 24 items of the original IGCS were grouped into four domains according to similar contents/meanings and the principles of standard cognitive conceptualization [
32], some items correlated strongly with multiple subscales of the original IGCS, as seen from the double loading of nine items in the EFA in this study.
We hence re-analyzed the 24 items in two relatively large split sub-samples and identified a new 3-factor structure by EFA (perceived rewards of Internet gaming, perceived urges for playing Internet games, and perceived unwillingness to stop playing without completion of gaming tasks) to form C-RIGCS, which is an improvement over the original 4-factor structure of IGCS. First, it was confirmed by CFA. Second, nine double-loaded items were removed to form the C-RIGCS; the correlations among the three factors became much lower than those among the original four factors; the three new factors thus showed clearer conceptual distinctions. Third, the modified C-RIGCS showed satisfactory psychometric properties, including internal consistencies (i.e., 0.70 or above for Cronbach’s alpha), test-retest reliability (i.e., 0.70 or above for ICC coefficients), and content validity (i.e., strong correlations between the items and their corresponding subscales but weaker correlations between the items and other subscales). No ceiling effect was noticeable. The perceived urges for playing Internet games subscale showed a slight floor effect, possibly due to the relatively high proportion of non-gamers (33.1% did not play Internet games in the past year) in the sample, whose potentially low gaming urges might have contributed to the observed floor effect.
Furthermore, the 3-factor model is conceptually different from the original 4-factor model. The new perceived rewards of Internet gaming factor included items of various subscales of the original IGCS. Although the different types of potential rewards (e.g., self-esteem, social acceptance, and tangible rewards) described by the original IGCS are conceptually different, Chinese adolescents may just perceive various types of rewards of Internet gaming as a single entity. The high Cronbach value of this subscale supports this conjecture. Perceived urges for playing Internet games and perceived unwillingness to stop playing before completion of gaming tasks were the other two factors identified in this study. It seems that urges to play may be associated with preoccupation with Internet gaming, loss of interest, and inability to control Internet gaming in other activities (DSM-5 criteria); perceived unwillingness to stop playing without completion may be associated with significant loss due to Internet gaming (DSM-5 criteria). Thus, the 3-factor structure may have some conceptual implications.
Besides, the three new subscales were all significantly associated with the six external variables. As expected, the new perceived rewards of Internet gaming factor showed stronger associations than the other two factors with the perceptions of Internet gaming being the primary sources of self-esteem and social acceptance. It is plausible that beliefs on rewards of Internet gaming was associated with reliance on Internet gaming as primary sources of gaining self-esteem and social acceptance. Likewise, the perceived urges for playing Internet games subscale showed stronger associations with both impulsivity and self-control than the other two factors; urges for Internet gaming may be signs of impaired self-control and impulsivity. Some degrees of convergent validity have hence been observed in these cases.
The findings may lead to some new research directions. Future studies should validate both the original version of IGCS and C-RIGCS in English and other languages to understand better the comprehensive structure of cognitions related to Internet gaming and their relationships with IGD. The inter-relationships among the three factors of C-RIGCS should also be studied, although it is beyond the scope of this study. For instance, the relationship between gaming for reward beliefs and IGD may be mediated by cognitions on urges/stopping. Future studies should also clarify differences in gaming-specific cognitions between Internet gamers and non-gamers. As the cognitions of adults and adolescents about Internet gaming are likely to differ, future studies should test validity and required modification of the C-RIGCS among adults.
The present study has several limitations. First, responses were self-reported and may involve reporting bias although the survey was anonymous. Second, the schools were not randomly selected and only seven schools in two major Chinese cities were involved; selection bias may occur. Third, about 1/3 of the participants did not play Internet games in the past 12 months. They also answered the questions as they might have played games previously, and they can also have beliefs about Internet gaming. It is a strength that our findings showed that the validation results among the gamer subsample were highly consistent with those derived from the overall study population. Lastly, although the original scale was based on a systematic review and a conceptual model, a number of items have been removed; the new analysis was hence based on EFA rather than the original model that did not show satisfactory goodness of fit.