مطالعه اکتشافی از ارتباط بین اعتیاد به بازی آنلاین و انگیزش لذت برای بازی گسترده چند نفره آنلاین

Massively multiplayer online role-playing games (MMORPGs) are a popular form of entertainment used by millions of gamers worldwide. Potential problems relating to MMORPG play have emerged, particularly in relation to being addicted to playing in such virtual environments. In the present study, factors relating to online gaming addiction and motivations for playing in MMORPGs were examined to establish whether they were associated with addiction. A sample comprised 1167 gamers who were surveyed about their gaming motivations. Latent Class Analysis revealed seven classes of motivations for playing MMORPGs, which comprised: (1) novelty; (2) highly social and discovery-orientated; (3) aggressive, anti-social and non-curious; (4) highly social, competitive; (5) low intensity enjoyment; (6) discovery-orientated; and (7) social classes. Five classes of gaming addiction-related experiences were extracted including: (1) high risk of addiction, (2) time-affected, (3) intermediate risk of addiction, (4) emotional control, and (5) low risk of addiction classes. Gender was a significant predictor of intermediate risk of addiction and emotional control class membership. Membership of the high risk of addiction class was significantly predicted by belonging to a highly social and competitive class, a novelty class, or an aggressive, anti-social, and non-curious class. Implications of these findings for assessment and treatment of MMORPG addiction are discussed.

Over the last decade, computer technology has greatly advanced to enable rapid interaction with other people in a range of online virtual worlds. This advancement has led to an increasing number of people using the Internet in many different ways and has arguably had a great positive impact on the lives of people that use it. Despite the many positive benefits, there has been an increase in research focusing on the use of the Internet and its negative aspects including both generalized Internet addiction and more specific online addictions such as online gaming addiction (e.g., Lopez-Fernandez et al., 2014 and Wang, 2001). Marlatt, Baer, Donovan, and Kivlahan (1988) defined addictive behaviour as: “A repetitive habit pattern that increases the risk of disease and/or associated personal and social problems. Addictive behaviours are often experienced subjectively as ‘loss of control’ – the behaviour contrives to occur despite volitional attempts to abstain or moderate use. These habit patterns are typically characterized by immediate gratification (short term reward), often coupled with delayed deleterious effects (long term costs). Attempts to change an addictive behaviour (via treatment or self initiation) are typically marked with high relapse rates” (p. 224). This is an all-encompassing operational definition as it can refer to both substance and non-substance behaviours (including gaming addiction). One method commonly used to determine whether a particular behaviour is addictive is to compare it against clinical criteria of more established addictions (Griffiths, 2005). This method makes potential addictive behaviours more clinically identifiable and has been supported by researchers that have carried out research into various ‘technological addictions’ such as television addiction (Sussman & Moran, 2013), mobile phone addiction (Carbonell et al., 2012), internet addiction (Kuss, Griffiths, & Binder, 2013), and gaming addiction (King, Haagsma, Delfabbro, Gradisar, & Griffiths, 2013). Much of the conceptualization of excessive gaming as an addiction stems back to the work of Griffiths in the 1990s who adapted versions of the DSM-III-R for pathological gambling (American Psychiatric Association, 1987) to video game addiction (e.g., Griffiths, 1997, Griffiths and Hunt, 1995 and Griffiths and Hunt, 1998). Other scholars adapted the DSM-IV criteria for pathological gambling to Internet addiction (e.g., Young, 1998). Furthermore, it can be argued that all types of addictive behaviour have elements in common. For instance, Griffiths (2005) operationally defined addictive behaviour as any behaviour that features the six core components of addiction, which were first outlined by Brown (1993) and later modified by Griffiths, 1996 and Griffiths, 2005, (i.e., salience, mood modification, tolerance, withdrawal symptoms, conflict and relapse). Under this model, it is argued that any behaviour (such as gaming addiction) that fulfils the six criteria can be operationally defined as an addiction. To illustrate the level of interest in the area of online addictions, a recent systematic review identified 69 studies examining Internet addiction with sample sizes of over 1000 participants (Kuss, Griffiths, Karila, & Billieux, 2014). Moreover, sophisticated ways of conceptualising and measuring video game addiction, or risk of experiencing it, have been adopted and this has meant some authors (e.g., Kuss & Griffiths, 2012) have been arguing that gaming addiction can best be understood along a continuum, rather than as a dichotomous construct. When using cut-offs for video game addiction, research by Hussain, Griffiths, and Baguley (2012) found that there could be as many as 44.5% of a sample of video game players who are deemed to be at risk of video game addiction, if using a polythetic coding method (i.e. at least four of seven items of a brief Gaming Addiction Scale being endorsed), whereas this estimate could be reduced to as low as 3.6% of all gamers, if using the monothetic coding method (i.e. all seven items being endorsed). Clearly, there appears to be a wide range of players who could be affected by problematic video game play behaviour, but the true prevalence of video game addiction is still uncertain. This may be due to a range of measures being used to tap into the phenomenon but also the tendency of some researchers to primarily see addiction as an either/or construct with gamers being deemed to be either addicted or not. However, it has been argued that video game play, and problems associated with it, needs to be understood as multidimensional with aetiological factors such as structural characteristics and motivation for game play being just as important as differentiating whether someone is addicted to video games or not (Kuss & Griffiths, 2012) One form of virtual world activity that has evolved on the Internet is the playing of Massively Multiplayer Online Role-Playing Games (MMORPGs). These games are now a popular form of entertainment used by millions of gamers worldwide, which provide an intense experience of immersion and can be extremely time-consuming (Kuss & Griffiths, 2012). This has also led to an increase of research into the area of online gaming over the past decade. Some of the areas of investigation have included gamer demographics (e.g. Griffiths et al., 2003, Griffiths et al., 2004, Yee, 2006a and Yee, 2006b), online gaming addiction (e.g., Hussain et al., 2012 and Spekman et al., 2013), within-game group formation (e.g., Chen et al., 2008, Ducheneaut et al., 2006 and Odrowska and Massar, 2014), and within-game social interaction (e.g., Cole and Griffiths, 2007 and Hussain and Griffiths, 2008). Estimates of video game addiction have varied. One meta-analysis of studies (Ferguson, Coulson, & Barnett, 2011) suggested that it could be approximately 3% among gamers. These authors argued that a useful distinction, which overlaps with the continuum concept of video game addiction, is that gaming can be fully engaging and it can also interfere with one’s life, but that a combination of many of these experiences would be needed for full-blown addiction to be present. Another study (Kuss, Griffiths et al., 2013), which focused on internet addiction, also obtained a similar prevalence rate, as 3.2% of the sample of 2257 participants appeared to have likely characteristics of internet addiction. An interesting finding was that a combination of online gaming and openness to experience increased the risk of addiction. A larger study by Kuss, van Rooij et al. (2013) investigated the risk for Internet addiction in a sample 3105 Dutch adolescents by looking at the interplay between personality traits and different Internet applications. The adolescents completed questionnaires including the Compulsive Internet Use Scale (CIUS) and the Quick Big Five Scale. It was found that 3.7% of adolescents were classified as addicted to using the Internet. Playing online games increased the risk of Internet addiction by 2.3%. The amount of online gaming (i.e., the number of hours played) and low scores on extraversion predicted Internet addiction. MMORPGs appear to be highly appealing environments and many gamers are motivated to use them (Griffiths et al., 2003 and Griffiths et al., 2004), and they have also been associated with a higher risk of video game addiction (Ng & Wiemer-Hastings, 2005). Gamer motivation is an area of importance as it provides insight into intentions for playing online from casual through to excessive play. Having knowledge about motivations for online gaming has the potential to provide insights about problematic gaming behaviour. One of the more popular theoretical standpoints of those examining gaming motivations is from a ‘uses and gratifications’ (UaG) perspective (e.g., Sherry et al., 2006, Wu et al., 2010, Yee, 2006a and Yee, 2006b). As Sherry et al. (2006) note, UaG research is based in the structural–functionalist systems approach that attempts to understand the interface between biological entities and the context in which they live. Research following a UaG perspective largely shows that the gaming motivations largely comprise personal and social gratifications. Research by Ryan, Rigby, and Przybylski (2006) involved using a measure of gaming motivations (Yee, 2006a and Yee, 2006b). The authors suggested that strong motivators for online gaming were (i) psychological need for relatedness and (ii) autonomy and competence features. Billieux et al. (2011) investigated the psychological predictors of problematic involvement in MMORPG use. Their sample comprised 54 male gamers who were screened using the UPPS Impulsive Behavior Scale, the Motivations to Play Online Questionnaire (MPOQ) and Internet Addiction Test (Young, 1999). The researchers found that problematic use of MMORPGs was predicted by (i) high urgency, and (ii) a motivation to play for immersion. Urgency was defined as the tendency to act rashly when experiencing negative affect states. The findings of the study were potentially useful for understanding predictors and motivations of gamers and the role of immersion as a motivation for playing online. However, the findings were limited by the very small sample size. However, it is worth noting that urgency has been linked to various problem behaviours including drug abuse (Verdejo-García, Bechara, Recknor, & Pérez-García, 2007), pathological gambling (Smith et al., 2007), problematic mobile phone use (Billieux et al., 2007 and Billieux et al., 2008) and problem drinking (Anestis, Selby, & Joiner, 2007). According to Billieux et al. (2011) immersing oneself in a virtual world can lead to negative, real-world consequences (e.g., procrastination, avoiding real-world problems). Yee, 2006a and Yee, 2006b looked at gamer motivations by surveying a sample of 3000 online gamers. An online questionnaire was publicised on various online forums that catered for popular MMORPGs. Yee, 2006a and Yee, 2006b used a 40-item inventory to create a model of player motivations. The results revealed 10 motivation sub-components of Advancement, Mechanics, Competition, Socialising, Relationship, Teamwork, Discovery, Role-Playing, Customisation, and Escapism. These components were grouped into three main motivation components of Achievement, Social, and Immersion. Further analysis to examine the association between the motivation components and problematic gaming showed that the escapism and achievement components were the best predictors of problematic gaming. More recently, Yee, Ducheneaut, and Nelson (2012) attempted to validate the motivations scale. Data were gathered from 2071 American participants and 645 participants from Hong Kong and Taiwan. This allowed the researchers to examine motivations for playing in a non-Western culture. The findings showed that online gaming motivations can be parsimoniously captured using the three-factor model of Achievement, Social and Immersion. Furthermore, the model was validated in Western and non-Western cultures but it gathered data from players of one MMORPG – World of Warcraft. Fuster et al. (2012) explored the psychological motivation for playing World of Warcraft in a sample of 253 male Spanish gamers using an online survey. The survey included a 32-item motivation scale that assessed the gaming motivations of socialisation, achievement, exploration, escapism and dissociation. Factor analysis of the survey responses revealed the presence of four motivations for gaming: socialisation, exploration, achievement, and dissociation. These findings were very similar to other research findings on this topic (e.g., Yee, 2006a, Yee, 2006b and Yee et al., 2012). Furthermore, the results indicated that socialisation was one of the main motivational factors that may potentially link to positive outcomes for gamers’ wellbeing. In a large study of Hungarian online gamers’ preferences and gaming behaviour, Nagygyörgy et al. (2013) used an online survey to recruit 4374 gamers from websites that catered for different types of MMORPGs. A latent profile analysis of gaming preferences revealed eight specific gamer types, of which four types emerged as clear categories, indicating clear preferences for a specific type of game (i.e. role-playing games, first-person shooter games, real-time strategy games, and other games). In general, 79% of gamers belonged to these categories. First-person shooter gamers were almost exclusively male, younger aged, and of a lower socio-economic status. Real-time strategy gamers were older. Females were more likely to play “other” games (e.g., non-violent games, puzzle games) and/or role-playing games. The authors speculated that specific games fulfil specific psychological needs and that gaming preferences are being formed in accordance with these needs. This may have implications for why some gamers play excessively. Although there have been some studies into the motivations to play online games, there is a lack of research into online gaming motivation and its relationship to problematic gaming with MMORPGs. One of the aims of the current exploratory study was to examine the structure of online gaming addiction and to see whether it can best be represented on a continuum. Another aim of the present study was to categorise online gaming motivations and to identify motivating factors in playing various MMORPGs and their association (if any) with problematic gaming and risk of gaming addiction. The study also attempted to address the limitations of previous research by examining both male and female gamers’ motivations as well as examining gamers that played many different types of MMORPGs. This study also attempted to identify the presence of distinct groups of gamers who endorsed specific addiction criteria using latent class analysis (LCA). The identification of motivating factors and addiction indicators may prove beneficial for prevention and treatment of addiction to MMORPGs.

3. Results The binary coded 21-item GAS was subjected to a LCA. Before doing so, the most common and least common endorsements of items as being relevant to the participants are displayed in Table 1. Table 1. Frequency of Gaming Addiction Scale Endorsements (in descending order). Item No. (%) 2. Did you spend a large amount of free time on MMORPGs? 691 (59.2) 8. Have you played MMORPGs to release stress? 505 (43.3) 4. Did you play longer than intended? 477 (40.9) 9. Have you played MMORPGs to feel better? 375 (32.1) 3. Have you felt addicted to a MMORPG? 311 (26.6) 5. Did you spend increasing amounts of time on MMORPGs? 306 (26.2) 1. Did you think about playing a MMORPG all day long? 266 (22.8) 6. Were you unable to stop once you started playing a MMORPG? 216 (18.5) 19. Has your time on MMORPGs caused sleep deprivation? 191 (16.4) 7. Did you play MMORPGs to forget about real life? 190 (16.3) 20. Have you neglected other important activities? 173 (14.8) 10. Were you unable to reduce your game time? 158 (13.5) 11. Have others unsuccessfully tried to reduce your MMORPG use? 131 (11.2) 13. Have you felt bad when you were unable to play? 121 (10.4) 18. Have you lied about time spent on MMORPGs? 120 (10.3) 17. Have you neglected others because you were playing MMORPGs? 118 (10.1) 21. Did you feel bad after playing for a long time? 117 (10.0) 12. Have you failed when trying to reduce game time? 90 (7.7) 14. Have you become angry when unable to play? 83 (7.1) 15. Have you become stressed when unable to play? 81 (6.9) 16. Did you have fights with others over your time spent on MMORPGs? 75 (6.4) Table options LCA was then used to combine the response patterns that each person gave to all 21 items. There was a wide range of possible response patterns with 221 (i.e. 2,097,152) possible permutations. A total of 560 response patterns were obtained for the GAS with this sample and the most common response patterns were ‘infrequent/absent’ for all 21 items (n = 184 respondents), followed by ‘frequent/present’ for Item 2 only (n = 84 respondents) and ‘frequent/present’ for Item 8 only (n = 34 respondents). The fit statistics obtained for the 1-class solution through to the 6-class solution can be found in Table 2. As can be seen, there were two fit statistics that supported the probability that 5 latent classes should be extracted. This was because the BIC reached its lowest point at the 5-class solution and the non-significant Lo-Mendell-Rubin Adjusted Likelihood Ratio test value with the 6-class solution pointed to the solution with the one fewer class. Table 2. Fit statistics for the Gaming Addiction Scale. Class Log likelihood No. of free parameters LR χ2 (d.f.) p AIC BIC SSABIC LRT (p) Entropy 1 −10770.325 21 3539.285 (2,096,900) 1.0000 21582.649 21688.955 21622.252 – – 2 −9061.888 43 1928.946 (2,096,846) 1.0000 18209.777 18427.451 18290.868 3395.021 (0.0000) .91 3 −8736.977 65 1945.901 (2,096,855) 1.0000 17603.955 17932.997 17726.535 645.667 (0.0128) .86 4 −8568.442 87 1710.870 (2,096,842) 1.0000 17310.884 17751.295 17474.953 334.915 (0.0002) .86 5 −8467.697 109 1760.422 (2,096,838) 1.0000 17153.393 17705.172 17385.951 200.202 (0.0100) .84 6 −8395.257 131 1549.599 (2,096,809) 1.0000 17052.514 17715.661 17299.560 143.953 (0.4024) .83 Key. LRχ2 = likelihood ratio chi-square, AIC = Akaike information criterion, BIC = Bayesian information criterion, SSABIC = sample size adjusted Bayesian information criterion, LRT = Lo-Mendell-Rubin’s adjusted likelihood ratio test. Table options After concluding that five latent classes could be extracted with the LCA, the appropriate labelling of each class was decided upon by examining the posterior probabilities profile plot (see Fig. 1). Class 5 – the largest class (44.8% of the sample) – was one that was deemed to be at the lowest risk of online game addiction as gamers in that class had probabilities of endorsing items at levels between 0% and 13.7%; only one item (Item 2) had a higher likelihood of endorsement (36.3%) but this was still at a lower probability than those in the other four latent classes. Full-size image (76 K) Fig. 1. Posterior probabilities profile plot: Latent Class Analysis of the Gaming Addiction Scale. Figure options Class 1 was the smallest class (7.2% of the sample) – labelled the ‘high risk’ class – was viewed as the respondent group most at risk of online game addiction as, relatively speaking, respondents had a higher probability than the other four classes of endorsing all but two of the 21 GAS items. An ‘intermediate risk’ of online game addiction was also identified comprising 12.2% of the sample. These were called ‘intermediate risk’ because people in this class tended to mirror the ‘high risk’ class on several items, particularly Items 1–7, but the likelihood of endorsing such items for those in this class was markedly lower than those in the ‘high risk’ class. This was the case for all but two of the GAS items. The second-largest class, comprising 20.1% of the sample, was labelled as the ‘emotional control’ class as they had a high propensity to endorse items about using video games to relieve stress (90.3% likelihood) and to feel better (70.4% likelihood). Another class – the ‘time-affected’ class – was 15.7% of the sample and was epitomised by having a high likelihood of endorsing the item “Did you think about playing all day long?” and they were also second most likely of the five classes to endorse Items 5 and 21, that both focused on spending lengthier periods of time playing MMORPGs. Items relating to enjoyment-related motivations from the MPOGQ were analysed in another LCA to examine whether there was a number of consistent patterns in responding that indicated the presence of likely classes of online game enjoyment. A total of 751 response patterns were identified in this sample out of 16,384 potential response patterns (i.e. 214). The most common response patterns included low-moderate levels of enjoyment for all 14 items (n = 40 participants), followed by high enjoyment with Item 1 only (n = 16 participants), high enjoyment with Items 1–7, 10–11, and 13 (n = 13 participants). The frequency of endorsement for each of the items is illustrated in Table 3 and the fit statistics for the possible class solutions for all of these response patterns are outlined in Table 4. As can be seen, the BIC level reached its lowest point with the 7-class solution – it should be noted that the BIC is generally regarded as the best information criterion of all the available information criteria for assessing model fit ( Nylund et al., 2007); with the 7-class solution, accuracy of classification was generally high at 82.3%. The likelihood ratio chi-square statistic for the 7-class solution was also non-significant, which indicated acceptable model fit, although this statistic is an absolute index and all of the other class solutions produced non-significant fit too. Table 3. Frequency of Gaming Motivation Endorsements (in descending order). Item No. (%) 11. Being part of a friendly, casual guild 757 (64.9) 1. How much do you enjoy working with others in a group? 718 (61.5) 7. Chatting with other players 667 (57.2) 6. Getting to know other players 664 (56.9) 3. How much do you enjoy finding quests, NPCs or locations that most people do not know about? 621 (53.2) 2. How much do you enjoy exploring the world just for the sake of exploring it? 610 (52.3) 5. Helping other players 608 (52.1) 8. Competing with other players 451 (38.6) 10. Exploring every map or zone in the world 446 (38.2) 4. How much do you enjoy collecting distinctive objects or clothing that have no functional value in the game? 397 (34.0) 9. Dominating/killing other players 390 (33.4) 13. Trying out new roles and personalities with your characters 363 (31.1) 12. Being part of a serious, raid/loot-oriented guild 324 (27.8) 14. Doing things that annoy other players 153 (13.1) Table options Table 4. Fit statistics for Latent Class Analysis on video gaming items relating to Enjoyment motivations. Class Log Likelihood No. of free parameters LR χ2 (d.f.) p AIC BIC SSABIC LRT (p) Entropy 1 −10463.549 14 5644.066 (16,331) 1.0000 20955.099 21025.970 20981.501 – – 2 −9735.703 29 4221.952 (16,310) 1.0000 19529.405 19676.209 19584.095 1442.080 (0.0000) 0.857 3 −9453.845 44 3674.724 (16,295) 1.0000 18995.691 19218.427 19078.668 558.443 (0.0000) 0.827 4 −9274.092 59 3318.810 (16,281) 1.0000 18666.185 18964.854 18777.450 356.144 (0.0000) 0.824 5 −9137.484 74 3135.712 (16,272) 1.0000 18422.968 18797.570 18562.521 270.662 (0.0723) 0.811 6 −9018.927 89 2932.791 (16,261) 1.0000 18215.854 18666.389 18383.695 234.896 (0.0495) 0.820 7 −8963.490 104 2780.645 (16,242) 1.0000 18134.979 18661.447 18331.108 109.838 (0.2450) 0.823 8 −8919.888 119 2783.817 (16,235) 1.0000 18077.777 18680.178 18302.193 86.387 (0.2913) 0.824 Key. LR χ2 = likelihood ratio chi-square, AIC = Akaike information criterion, BIC = Bayesian information criterion, SSABIC = sample size adjusted Bayesian information criterion, LRT = Lo-Mendell-Rubin’s adjusted likelihood ratio test. Table options As can be seen in Fig. 2, Class 1 (13.4% of the sample) were highly likely (97.1%) to endorse Item 2 (“Exploring the world for the sake of it”) and 92.8% likely to say that they enjoyed getting to know other players (Item 6) and 95.6% likely to enjoy chatting with other players. As a result, this class was termed the ‘novelty’ class as they were continually looking for new information, either about the MMORPG world or about their fellow players. Those in Class 2 (15.7% of those surveyed) were viewed as members of a ‘highly social and discovery-orientated’ class. Of all the seven classes, Class 2 was the top ranked in terms of endorsing items such as enjoying the collection of distinctive objects (62.5% likelihood), helping other players (86.6% likelihood), and being part of a friendly and casual guild (88.9%). They were also 100% likely to endorse Item 2, thus indicating their discovery-orientated enjoyment from playing an MMORPG. Full-size image (91 K) Fig. 2. Posterior probabilities profile plot: Latent Class Analysis of the enjoyment motivations for playing an MMORPG. Figure options Class 3 (9.9% of the sample) was mainly characterised by aggressive, anti-social and having non-curious tendencies. They were 93.8% likely to say that they enjoyed dominating and killing other players in the virtual world. Furthermore, gamers in Class 3 were the most likely to say that they enjoyed irritating other players, and had the lowest probabilities of all seven classes in endorsing items relating to seeking out novel situations or people (namely Items 2 and 10). Respondents in Class 4 (9.2% of the sample) were differentiated by their sociability and competitiveness. They were the most likely of all seven classes to enjoy getting to know other players (99.6% probability) and in competing with other players (97.4%). As a consequence, this group was termed the ‘highly social and competitive’ class. Class 5 (13.1% of the sample) was termed as the ‘low intensity’ enjoyment class as they were those with the lowest probability of agreeing with several of the enjoyment-related items, namely Items 3, 4, 6, 8, 9, 11, 12, and 13. For the additional analyses examining how the motivations for playing were associated with experiences of online game addiction, Class 5 appears to be a useful comparison class with the other enjoyment classes, given the limited range of enjoyment that Class 5 seemed to derive from playing an MMORPG. Class 6 was the second largest class (17% of the sample) and was primarily characterised by exploration as the source of enjoyment when playing. This class was termed the ‘discovery-orientated’ class, which was owing to their high levels of likely endorsement of Items 2 (84.5%) and 3 (80.4%). The next most likely item for respondents in this class to be endorsing was Item 10 (“exploring every map or zone in the world”), which had a 65.6% chance of being endorsed by those in this class. The largest class – Class 7 – comprised 21.6% of those surveyed and was termed as the ‘social’ class. This was because the highest likelihood of endorsements for this group of respondents were all for socially-focused items, namely Items 1, 6 and 7. The multinomial logistic regression analysis involved predicting online gaming addiction latent class membership based on the motivations for online game enjoyment latent class membership and demographic background. For game addiction class membership, ‘low risk’ was the reference category; likewise, game enjoyment class membership was compared with the reference class of ‘low intensity’ enjoyment. Table 5 shows that two of the predictor variables – online game enjoyment class membership and gender – were significant. Table 5. Likelihood ratio tests for multinomial logistic regression for video game motivation class membership and demographic background to sample. Effect -2 log likelihood Chi-square d.f. Sig. Intercept 850.085 .000 0 – Motivation class 952.389 102.304 24 .000 Gender 866.623 16.537 4 .002 Relationship status 853.436 3.351 4 .501 Age 865.107 15.022 12 .240 Table options In examining Table 6, it can be seen that there is a comparison between the likely class memberships for each of the game addiction latent classes when compared with a reference class. The following statistically significant trends were observed when associating this probability with respondents’ likely online game enjoyment class memberships and demographic variables. For being part of the ‘high risk’ class versus the ‘low risk’ class, participants were 14.4 times more likely to be in the ‘highly social and competitive’ class, 9.08 times more likely to be in the ‘novelty’ class, and 4.78 times more likely to be in the ‘aggressive, anti-social and non-curious’ class rather than the ‘low intensity enjoyment’ class. With membership of the ‘time-affected’ class versus the ‘low risk’ class, respondents were 5.25 times more likely to be in the ‘highly social and competitive’ class, 3.08 times more likely to be in the ‘novelty’ class and 2.16 times more likely to be in the ‘aggressive, anti-social and non-curious’ class. In addition, age category membership was also key for membership of the ‘time-affected’ class, with respondents being 2.49 times more likely to be in the youngest age group or 2.31 times more likely to be in the second youngest age group rather than being in the oldest age group. In belonging to the ‘intermediate risk’ class versus the ‘low risk’ class, members were 9.15 times more likely to be in the ‘highly social and competitive’ class and 5.92 times more likely to be in the ‘novelty’ class. The likelihood of having other class memberships also ranged from being 2.39 times more likely of being in the ‘social’ class to 3.63 times more likely of being in the ‘social and discovery-orientated’ class. Furthermore, membership of the intermediate risk class was also dominated by male participants as males were 1.82 times more likely to belong to this class when compared with females. Table 6. Multinomial logistic regression with motivation and demographic variables predicting video game addiction latent class membership. Associations (OR, 95% CIa) with: Class 1 high risk Class 2 time-affected Class 3 intermediate risk Class 4 emotional control Motivation class 1. Novelty 9.08 (2.94–28.05) 3.08 (1.57–6.04) 5.92 (2.48–14.15) 4.67 (2.25–9.71) 2. Social and discovery-orientated .61 (.11–3.43) 1.54 (.76–3.15) 3.63 (1.55–8.50) 5.09 (2.58–10.03) 3. Aggressive, anti-social, non-curious 4.78 (1.43–15.99) 2.16 (1.06–4.39) 3.09 (1.17–8.15) 2.26 (.98–5.22) 4. Highly social and competitive 14.40 (4.32–47.99) 5.25 (2.45–11.25) 9.15 (3.44–24.36) 5.61 (2.34–13.44) 5. Social 2.96 (.95–9.25) 1.72 (.92–3.22) 2.39 (1.02–5.58) 3.33 (1.71–6.48) 6. Discovery-orientated 3.16 (.99–10.10) 1.25 (.62–2.49) 2.96 (1.28–6.86) 2.57 (1.28–5.14) 7. Low intensity enjoymentb – – – – Gender Female 1.59 (.83–3.05) .81 (.49–1.34) 1.82 (1.15–2.86) 1.78 (1.21–2.60) Maleb – – – – Relationship status Single 1.39 (.78–2.47) 1.18 (.79–1.76) .84 (.55–1.28) 1.12 (.78–1.61) In a relationshipb – – – – Age 17 years or younger 2.47 (.95–6.42) 2.49 (1.26–4.94) 1.32 (.69–2.54) 1.42 (.82–2.46) 18–25 years 1.71 (.70–4.15) 2.31 (1.25–4.29) 1.18 (.69–2.02) 1.18 (.75–1.87) 26–30 years 1.27 (.40–4.03) 1.62 (.73–3.60) .95 (.45–1.96) 1.55 (.88–2.73) 31 years or olderb – – – – Intercept −4.01 −2.46 −2.68 −2.46 a Confidence intervals not including unity indicate statistical significance. b Comparison level. Table options The likelihood of being part of the ‘emotional control’ class versus the ‘low risk’ class was affected by a range of enjoyment class memberships, including being 5.61 times more likely to belong to the ‘highly social and competitive’ class, 5.09 times more likely to be part of the ‘social and discovery-orientated’ class, and 4.67 times more likely to be a member of the ‘novelty’ class of online game enjoyment motivations. Males were also 1.78 times more likely to belong to this class than females.