Excessive online gaming use is a social and mental health phenomenon with implications for emerging adults (ages 18–29; Kuss and Griffiths 2011; Stavropoulos et al. 2016a; Kuss et al. 2012). Internet Gaming Disorder (IGD) was introduced as a condition for further study in the most recent (fifth) edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; APA 2013). The proposed disorder has been defined as the continuous and ongoing use of playing online games leading to significant impairment or distress (APA 2013). However, there are a number of online game genres that may impact on IGD susceptibility, with Massively Multiplayer Online (MMO) games being consistently associated with IGD (Lemmens and Hendriks 2016; Stetina et al. 2011; Ng and Wiemer-Hastings 2005). MMO games are online worlds that constantly evolve in real time, where players may interact within the game via their virtual characters (avatars), and potentially addressing shared group and/or personal goals (Stetina et al. 2011). Progressive challenges are intertwined with a well-developed reward schedule, enriched with opportunities for online socialization and character development (Thorens et al. 2012). These inherent MMO factors within the game structure appear to contribute to their absorbing nature (Lemmens and Hendriks 2016; Stetina et al. 2011; Ng and Wiemer-Hastings 2005; Thorens et al. 2012).

Psychosocial factors that have been proposed as IGD risks range from those that are identity-related (e.g., role playing, identity-building, and exploration; Kuss et al. 2012) to those involving self-esteem drives (Bessière et al. 2007; King and Delfabbro 2014; Beard and Wickham 2016; Yee 2006; Beard et al. 2017). These individual factors are important for understanding both IGD phenomenology as well as understanding the critical aspects of human-computer interaction that have emerged in relation to Internet gaming (Anderson et al. 2016). Although prior studies have indicated that immersion in the online environment is related to more problematic play (Wood et al. 2004), exploration into the degree of presence in the game is nascent (Milani et al. 2017).

Presence is defined as a psychological state, or a subjective perception, in which an individual may experience the game context as real, feeling as “being there” instead of “being here” (Huang et al. 2011). Therefore, presence within MMOs may serve as a way to transition gamers into a parallel state wherein their perception of reality might fade into the background of their consciousness (Stavropoulos et al. 2013; Kaczmarek and Drążkowski 2014). Within the gaming context, subjective experiences of presence are thought to progressively strengthen the sense of the game world being real, which could in turn increase online gaming involvement (Stetina et al. 2011; Huang et al. 2011; Kaczmarek and Drążkowski 2014). This concurs with cross-sectional findings indicating presence as a factor in Internet abuse (Kuss et al. 2012; Huang et al. 2011; Kaczmarek and Drążkowski 2014). Within the context of online play, the repeated experiences of presence can result in a feeling of belonging to the game environment (Stavropoulos et al. 2013). This sense of belonging and “being there” may act as an incentive to increase and maintain online activity, thus resulting in excessive use (Stavropoulos et al. 2013; Griffiths 2005).

Masten’s Risk and Resilience Framework (RRF) has previously been adapted to investigate IGD phenomenology (Griffiths 2005; Masten 2001) and used to identify factors that might contribute to the adaptive or maladaptive use of games. This framework has also been applied to other technological addictions, as it has been recognized that excessive Internet use alone does not indicate addiction (i.e., an individual can be highly engaged, but not addicted; Stavropoulos et al. 2016b). The RRF suggests that behaviors constantly fluctuate across the pathological and/or non-pathological range, with their intensity being defined by the longitudinal interplay between individual (e.g., game genre preference; Eichenbaum et al. 2015; Elliott et al. 2012) and contextual (i.e., offline or online) effects (Masten 2001; Douglas et al. 2008). Given the background of immersion as a risk factor for problematic play, the notion of virtual presence in the game is examined in the present study as a potential IGD risk (Faiola et al. 2013).

These issues are particularly salient for emerging adults (i.e., those aged 18–29 years) due to specific developmental goals and needs present during this period of transition from adolescence to full adulthood (Kuss and Griffiths 2011; Stavropoulos et al.  2016a; Kuss et al. 2012). Unsupervised Internet use, along with the lack of matured adult-type roles (prominent through the developmental period of emerging adulthood), appears to pose gamers at higher risk of IGD (Kaczmarek and Drążkowski 2014; Scott and Porter-Armstrong 2013; Hendriks 1990; Stanley 2008). The identification of emerging adulthood as a vulnerable developmental period is consistent with findings in other domains of substance and behavioral addictions (Kaczmarek and Drążkowski 2014; Scott and Porter-Armstrong 2013; Hendriks 1990; Stanley 2008). Accordingly, the present study will seek to build upon the current empirical understanding of IGD within this high-risk population (i.e., emerging adult MMO gamers).

Although prior literature has identified immersion as a risk factor for IGD, the longitudinal association of online presence and IGD has not been explicitly examined (Milani et al. 2017). This preliminary study utilized an RRF perspective to assess the effect of presence on IGD symptomology in a high-risk population of young adults who play MMO games. More specifically, presence was explored as it related to IGD both (i) cross-sectionally and (ii) as a predictor of IGD across a 3-month period.

Method

Participants

The sample comprised 125 MMO players aged 18–29 years old (64 online, cross-sectional respondents, Mage = 23.34, SD = 3.39, Males = 49, 77.6%; 61 offline, face-to-face, longitudinal, Mage = 23.02, SD = 3.43, Males = 47, 75.4%). Online and face-to-face data (i.e., time-point 1 [T1]) were combinedFootnote 1 in the cross-sectional analyses and face-to-face longitudinal dataFootnote 2 (i.e., time-points 1, 2, 3 [T1, T2, T3]; see Fig. 1). The data were then analyzed separately to assess presence as a predictor of IGD over a 3-month period (in consensus with the frequently used time bracket of assessing short-term changes in other types of addiction forms; D’Amico et al. 2015). Sociodemographic characteristics of the sample are presented in Table 1. The estimated maximum sampling error with a size of 125 is 8.77% and therefore, within the acceptable limits (Salant et al. 1994).

Fig. 1
figure 1

Participant pool

Full size image
Table 1 Sociodemographic characteristics of the participants in the study
Full size table

Measures

A battery of scales and a series of demographic information questions were applied.Footnote 3

The Internet Gaming Disorder Scale–Short Form 9 (IGDS-SF9) was used to assess IGD severity. The IGDS-SF9 includes nine items reflecting the DSM-5 IGD criteria (e.g., “Do you systematically fail when trying to control or cease your gaming activity?”; APA 2013; Pontes and Griffiths 2015) Items are addressed on a 5-point Likert scale (1 = never to 5 = very often) yielding total scores ranging from 9 to 45 with higher scores indicating higher IGD severity. Internal reliability in the present study was high, with a Cronbach’s α of .92.

The Presence Scale was used to assess online presence (Faiola et al. 2013). This includes presence (Griffiths 2005). The scale comprises 10 self-report items which focus on the participants’ experience in the virtual environment (e.g., “I feel like all my senses are engaged in the environment”). Items are rated on a 5-point Likert scale (1 = very little to 5 = very much). Item scores are added resulting to a range between 10 and 50, with higher scores indicating higher levels of perceived presence. Internal reliability in the present study was high, with a Cronbach’s α of .82.

Procedure

Ethics approval was received from the research team’s university ethics committee. Participants were recruited in the general community using both offline methods (i.e., information flyers) and online methods (i.e., e-mail, social media).Footnote 4 The face-to-face, longitudinal component of the study was conducted over a 3-month period, with the three time-point assessments completed between June 2016 and September 2016. Data collection was identical between the three time-points, and participants’ data were matched using a re-identifiable code.

For online collection, eligible individuals interested in participating and unable or unwilling to attend face-to-face testing sessions were invited to register with the study via a SurveyMonkey link available on MMO websites and forums (i.e., http://www.ausmmo.com.au). The link directed them to the plain language information statement where they had to digitally provide informed consent (by clicking a button).

Results

Linear regression analyses were conducted to assess cross-sectional presence scores predicting IGD severity. Bootstrapping on the minimum recommended level of 1000 was applied (Berkovits et al. 2000). Results indicated that the slope of the regression line was statistically significantFootnote 5 (F(1, 123) = 31.54, p < .001). Regarding the overall effect, 20% of the variance of IGD scores was explained by gaming presence (R2 = 0.204). The standardized coefficient (β) indicated that a standard deviation increase in presence was associated with a 0.45 standard deviation increase in IGD (b = 0.42, SE (b) = 0.7, β = 0.45, t = 5.61, p < .001).

Second, a latent growth model (LGM) of IGD behaviors over the three time-points (face-to-face data available only) was calculated using robust maximum likelihood estimation with Mplus software, Version 6.12 (Muthen and Muthen 2010). This approach allowed IGD behaviors at different time-points to be evaluated so that the mean and the variance of the intercept (the initial status of IGD behaviors as measured at T1) and the growth function (change of IGD behaviors across the time-points) were produced. IGD behaviors at the three time-points were fixed at 0, 1, and 2, respectively, to define a linear growth model with equidistant time-points. As part of the parameterization, the intercepts of IGD behaviors at the three time-points were fixed at 0, and their residual variances were freely estimated. The chi-square goodness of fitFootnote 6 (discrepancy index), the root mean square error of approximationFootnote 7 (RMSEA-discrepancy index), the comparative fit indexFootnote 8 (CFI-incremental fit index), the Tucker Lewis indexFootnote 9 (TLI-incremental fit index), the Akaike information criterionFootnote 10 (AIC-information theory goodness of fit), and the Bayesian information criterionFootnote 11 (BIC-information theory goodness of fit) were calculated in order to determine overall model fit. Second, the presence score at T1 was added as a time invariant covariate (predictor) of the IGD intercept and slope. Model fit indices were compared between the two models. Missing values were addressed with full information maximum likelihood (Schafer and Graham 2002) (Table 2).

Table 2 Model(s) fit information
Full size table

A path diagram of the final LGM model is shown in Fig. 2. This includes the latent growth curve with intercept (I), linear parameter (S), and the time invariant covariate of presence (pqtott). The observed variables were the repeated measures of IGD behaviors, obtained at the three time-points: Initial (T1), 1 month after the baseline measurement (T2) and 2 months after the initial assessment (T3).

Fig. 2
figure 2

Mean and standard deviation ratings for IGD severity at 0, 1, and 2 months were 19.49 (7.02), 18.71 (6.84), and 17.93 (5.91), respectively. The growth curve model was conducted twice, with presence as a time invariant covariate, and without presence. Analyses indicated that the model which included presence provided better fit than the model without (AIC and BIC values decreased; see Table 2). Presence at T1 failed to predict the IGD-slope; however, it significantly associated with the IGD intercept (b = 0.369; p < .001). Therefore, presence significantly predicted the initial IGD scores, but it was not predictive of change over a 3-month period

Full size image

Discussion

The present study utilized the RRF, enriching it with key factors from the IGD literature, to explore presence as an IGD risk factor. In line with the relevant findings, MMO gamers who experienced higher presence presented with higher IGD scores (Stavropoulos et al. 2013). The contribution of presence to IGD was notable, as it accounted for over 20% of IGD variance. These empirical findings substantiate presence as an important IGD risk factor.

Although cross-sectional analyses did indicate presence as an IGD risk factor, the growth model did not support the initial levels of presence related to change in IGD scores over the 3-month period. This may either indicate that (a) other factors (e.g., cognitive or personality traits of the gamer) influence the relationship between IGD and presence, or that (b) the study duration needs to be more prolonged for the relationship to be apparent. Consequently, the results of the present study should be considered as a first step examining the relationship between presence and IGD. Future studies may consider a parallel growth model assessing the level of change of presence in relation to the level of change of IGD to examine the relationship between the relative slopes of IGD and presence over time. Due to statistical limitations related to the sample size of the present study (i.e., model convergence), such a model could not be examined here. Nevertheless, the present findings may have important clinical implications for MMO gamers presenting with IGD.

A strong perception of the online world as real may increase the time spent for gaming, and thus magnify potential problems in real life (e.g., education/occupation underperformance/problems; Anderson et al. 2016). Furthermore, the experience of the gamer as being present in-game should be therapeutically considered because IGD associates with known motivational contributors (e.g., escapism and immersion; Kuss et al. 2012; Bessière et al. 2007; Yee 2006). Presence and immersion within virtual worlds will continue to dominate the gaming landscape via the introduction of more immersive technologies (e.g., virtual reality, realistic in-game graphics; Bessière et al. 2007). It might be important that clinical work further explores what impact this may have on an individuals’ desire to engage with MMO games, and to what extent.

The findings and implications should be considered in relation to the limitations of the study. First, the self-selected sample was mostly male; therefore, the results from a non-representative sample cannot be generalized to female gamers or gamers more generally. In addition, the longitudinal design spanned only a few months, which may not be enough time to provide insight into the associations examined. Similarly, potentially confounding factors related to game use were not addressed (e.g., gaming time). Thirdly, the data were self-report and subject to many well-known biases (including memory recall and social desirability). Despite these limitations, the present study had significant strengths, such as combining online and FTF participants, cross-sectional and longitudinal analyses, and emphasized and an under-researched high-risk population (i.e., emerging adult MMO gamers). In such a context, future research could benefit from the exploration of the over-time interplay of other game-related risk factors, such as flow or user-avatar associations. Furthermore, potential protective factors related to the gamer and/or their real context (i.e., peer and family relationships) could also be explored, allowing for a more holistic understanding of IGD.