Multiple Sequential Mediation in an Extended Uses ...

Multiple Sequential Mediation in an Extended Uses and Gratifications Model of Augmented Reality Game Pokémon Go Ezlika Ghazali Mei Yuen Woon Dilip S. Mutum (corresponding author)

Ezlika Ghazali Department of Marketing, Faculty of Business and Accountancy, University of Malaya, 50603 Kuala Lumpur, Malaysia. email: [email protected] Mei Yuen Woon Graduate School of Business, Faculty of Business and Accountancy, University of Malaya, 50603 Kuala Lumpur, Malaysia. email: [email protected] Dilip S. Mutum (corresponding author) Nottingham University Business School Malaysia, The University of Nottingham Malaysia Campus, Semenyih, 43500 Selangor, Malaysia. email: [email protected] Keywords: Augmented Reality, Mobile Game, Uses and Gratifications, Continuance intention, Nostalgia, Need to Collect, Flow, Community Involvement

This paper has been accepted for publication in Internet Research (https://www.emeraldinsight.com/loi/intr). Please note that the final article may include updates to this current version. Please cite as: Ghazali, Ezlika; Mutum, Dilip S. and Woon, Mei Yuen (2018). Multiple Sequential Mediation in an Extended Uses and Gratifications Model of Augmented Reality Game Pokémon Go. Internet Research, forthcoming.

Title: Multiple Sequential Mediation in an Extended Uses and Gratifications Model of Augmented Reality Game Pokémon Go Abstract: Purpose (mandatory) The paper investigates the mechanism by which Uses and Gratification (U&G) constructs predict continuance intention to play (ContInt) the augmented reality game Pokémon Go (PG), through multiple serial mediation technique, with Enjoyment and Flow as mediators. The model also integrates other motivational factors specific to PG, namely, Network Externality and Nostalgia and investigates the process by which they influence ContInt through players’ inherent need-to-collect animated monsters and online community involvement, respectively. Design/methodology/approach (mandatory) The model was tested using 362 validated responses from an online survey of PG players in Malaysia. PLS-SEM was used to analyse the data. The predictive relevance of the model was tested via PLS-Predict. Findings (mandatory) ContInt is influenced through various mechanisms. Enjoyment is the most important mediator, mediating three U&G predictor constructs (Achievement, Escapism, Challenge, and Social Interaction) and the outcome ContInt. Flow did not have any influence on ContInt unless coupled with Enjoyment as a serial mediator. Network Externality and Nostalgia were found to only influence ContInt through mediators, Online Community Involvement, and Need-to-Collect Pokémon Monsters, respectively. Overall, the results show evidence of four indirect-only mediation paths and one complementary partial mediation path. Originality/value (mandatory) Provides support for an integrated model incorporating psychological, social, and gaming motivational factors. While most other studies focus on direct relationships, we focus on indirect relationships through multiple sequential mediation analysis, following the recent modern mediation analysis guidelines. Contrary to previous findings, Flow was not an important factor in predicting ContInt for gaming and Nostalgia does not link directly to ContInt. Keywords: Augmented Reality Mobile Game, Pokémon-Go, Uses and Gratifications, Nostalgia, Need-to-Collect, Flow, Multiple Sequential Mediation, PLS-SEM, PLS-Predict

1. Introduction Mobile games have grown in popularity and various types of mobile games have been created, ranging from casual and hybrid to augmented reality (AR) games. AR is a technique that displays virtual content superimposed upon real life objects (Tan et al., 2015), combining the physical world and virtual reality. Pokémon GO (PG), an AR game released by Niantic in 2016, became the most downloaded game app (Gilbert, 2016). PG’s novel use of AR is considered a breakthrough in the gaming world. Despite its success, shortly after its launch in 2016, the growth of PG appeared to lose its momentum in the US. However, in Malaysia, the popularity has grown again with the introduction of a new feature called Mewtwo, as well as the “Legendary Pokémon” (Yau, 2018). The game remains popular and, according to Apptopia, it made US$950 million in 2016 with 752 million downloads. It also remains the game with the most daily active users as compared with other mobile game apps, including Candy Crush (Minotti, 2017). The interest in AR mobile games is bound to increase with the release of several other games, including Harry Potter (Spangler, 2018) and Jurassic World Alive games (Lussier, 2018). Despite the huge potential of AR in the gaming industry, most studies on AR have focused on military, industrial, and medical applications (Krevelen and Poelman, 2010). There is a dearth of literature examining AR in gaming applications. Looking at PG specifically, a few recently published studies have looked at the direct effects of selected drivers in influencing the motivation to play the game (Ghazali et al., 2018; Rauschnabel et al., 2017; Yang and Liu, 2017). More work needs to be done, however, and this paper aims to fill this gap by proposing an integrated and holistic framework that not only breaks down the Uses and Gratifications (U&G)

1 of 35

framework into its four component variables (viz., Achievement, Challenge, Escapism, and Social Interaction), but also extends the framework to include and examine the interplay of social influences to predict Continuance Intention to play PG (ContInt). Flow and Enjoyment constructs, identified in previous gaming studies, were also included in the model. PG also inspired us to examine another variable called Nostalgia to represent the nostalgic influence of Pokémon as a childhood game, to explain the phenomenon behind many people jumping on the PG bandwagon immediately after it became available. Furthermore, this study contributes to the literature in a number of ways. First, it clearly demonstrates that the dimensions of U&G influence ContInt differently and the mechanism by which it influences motivation to continue playing PG also varies. For instance, enjoyment is an important mediator because three U&G dimensions depend on it for predicting ContInt. In contrast, Flow construct only influenced ContInt when it was coupled with Enjoyment (as a serial mediator), though the literature indicates that Flow is one of the most important gaming drivers. Second, PG has received mass media coverage and public awareness of the game has increased tremendously. Besides the news coverage, the energy and excitement of players congregating in various spots around Kuala Lumpur and other big cities in Malaysia was quite infectious. However, it was found that this perception of network externality does not directly influence a player’s ContInt. It was demonstrated that online community involvement was a vital mediator of the relationship between perceived network externality and ContInt. The findings have shown that the ability of network externality to predict ContInt only manifests itself when the players are actively involved in the PG online communities. To the best of our knowledge, no past research has looked into this relationship. Third, two vital constructs that are unique to PG were added, namely, Nostalgia (Hardy, 2016) and players’ intrinsic need-to-collect Pokémon 2 of 35

monsters (NeedCol). It was found that nostalgic feelings towards Pokémon monsters will only have an influence on ContInt if the influence passes through NeedCol. Also, to the best of our knowledge, no past research had examined this relationship, either. Finally, this study presents the results of serial multiple mediation analysis tests on all the indirect paths simultaneously, following recent guidelines of modern mediation analysis. By comparison with most past mediation analysis methods, which followed the Baron and Kenny technique, running analysis of the multiple pathways simultaneously was considered more accurate, as all these factors operate together in influencing ContInt in real-life situations. 2. Literature Review and Hypotheses Development 2.1 Continuance Intentions to Play PG Game developers face an uphill challenge in attracting and retaining players. The sustainability and success of a game has always depended on engaging players who are willing to consistently invest their time, effort, and money into the game. Thus, understanding the determinants of players’ intentions is critical for game providers if game developers are to achieve sustainable success. Previous studies that have looked at gaming have emphasised the importance of continuous usage for information systems because the long-term success of an information system depends on its continued use rather than its adoption (Basak and Calisir, 2015; Chang, 2013). Thus, players’ continuance intention to play is considered more important than their initial intention to play, given the vital role of players’ post-adoption usage in determining the success of game providers (Chang et al., 2014; Gao and Bai, 2014).

3 of 35

2.2 Uses and Gratifications According to U&G theory, people actively choose and use particular media based on their needs. These needs are derived from psychological and social situations that produce motives that, in turn, influence media use (Katz and Shapiro, 1985; Weibull, 1985). This theory has been used to study the use of new media technology (Cianfrone et al., 2011), such as personal websites, social networking sites, video games, and online games. Past research has demonstrated that favourable behavioural outcomes of using media are often seen when users’ needs or desires are satisfied (Huang and Hsieh, 2011; Ifinedo, 2016; Xu et al., 2012). Within the gaming context, past U&G studies have identified various social and psychological factors associated with different game genres. For example, Jin (2014) found that players seek entertainment, fantasy, challenge, and escape when playing social network games. Yee (2006) suggested that the motivation to play massive multiplayer online role-playing games can be classified into three categories: achievement, social, and immersion. A study of online games showed that multiple gratifications (e.g., achievement, enjoyment, and social interaction) had a significant effect on a player’s continued motivation to play (Wu et al., 2010). These studies underline the suitability of incorporating U&G as the underlying theory to explain players’ psychological needs in association with their gaming behaviours as it helps researchers understand how and why players obtain gratification during their playing experience. Building on this theory, this study proposes four U&G constructs, namely, Achievement, Challenge, Escapism, and Social Interaction, which are discussed below. Achievement desires for playing online games include the needs to (1) gain power, (2) progress rapidly in the game, (3) gather virtual game objects, (4) gather valuable performance points, and (5) compete with others (Wu et al., 2010). According to Merhi (2016), two main 4 of 35

reasons for playing online games is to satisfy a desire that players cannot achieve in the real world and to show off their ability to other players. Once they have achieved the given goals, they feel satisfied and are more likely to continue playing. Similarly, while looking at online video game usage (interactive hedonic system usage), Lin and Bhattacherjee (2010) found that social image affected attitudes as well. The second U&G construct, Challenge, refers to the level of difficulty in the gaming context, which includes competing with other players or completing the mission given in the game (Liu and Shiue, 2014). However, challenge here refers specifically to challenge-seeking (Jin, 2014), which is considered to be an intrinsic motivation. Some empirical studies have also demonstrated that challenge seeking plays a critical role in motivating the player (Jin, 2014; Lee et al., 2012). The third construct, Escapism, is used by individuals to obtain relief from real-life problems by escaping unpleasant situations, relieving stress, or breaking the tedium of daily life (Merhi, 2016; Li et al., 2015; Demetrovics et al., 2011). According to Ho et al. (2017), people attempt to distract themselves from reality by consuming media, and a few studies have shown that escapism influences pathological video game use (Hilgard et al., 2013). Finally, Social Interaction recognises the importance of communication with others and building relationships with them (Li et al., 2015; Merhi, 2016). Social interaction is a key factor motivating players to continuously engage in games. For example, Chen et al. (2016) found that social interaction significantly influences players’ perceived enjoyment, which in turn significantly influences their intention to play.

5 of 35

2.3 The Mediating Effect of Enjoyment While studying website usage, Van der Heijden (2003) defined enjoyment as the degree to which performing an activity was perceived as pleasurable and fun in its own right. The literature indicates that the various U&G constructs have a relationship with enjoyment. Studies have shown that players usually enjoy the feeling of accomplishment when they move up or advance to a higher level of gaming. For instance, Puente-Díaz (2012) found that achievement significantly influenced enjoyment among competitive athletes. Fulfilling the desire for achievement may provide PG players with personal satisfaction and a feeling of pleasure. Previous studies have also suggested that incorporating reasonable challenges in games can make games fun, which leads to an optimal and enjoyable playing experience (Lee et al., 2012; Teng et al., 2012). Players generally like to be challenged because, once they complete a level or mission, they feel good about the accomplishment. Brown et al. (2008) highlighted that play-testing was essential to ensure that the gaming experience was well balanced and enjoyable for gamers. Furthermore, according to Ho et al. (2017), people attempt to distract themselves from reality by consuming media. When media are sufficiently engaging, the distraction from negative thoughts is a form of diversion that leads to enjoyment. In the gaming context, empirical studies have found that escapism is an important factor affecting enjoyment (Chen et al., 2016; Merhi, 2016). In general, players tend to seek relief from everyday problems by playing games to achieve enjoyment. Through online games, players have the opportunity to play, meet, compete, and chat with other players (Merhi, 2016). Thus, social interaction can be regarded as one of the key factors that determine the enjoyment and motivation of players. For instance, Chen et al.

6 of 35

(2016) found that social interaction significantly influences players’ perceived enjoyment. Research has also identified enjoyment as a strong determinant of users’ behavioural intention in a variety of online environments (Chen et al., 2016; Merhi, 2016; Wu and Liu, 2007). Furthermore, Shiau and Luo (2013) indicated that user involvement, satisfaction, and perceived enjoyment predict continuance intention for blog usage. Enjoyment also plays a crucial role in users’ engagement with mobile commerce activities (Wu and Liu, 2007). Merhi (2016) and Merikivi et al. (2017) also found that enjoyment positively influences the intention to play online games. The enjoyable feelings while playing games will lead to positive attitudes and expectations of the game and provide more enduring reasons or motives to keep playing the games (Boyle et al., 2012). Therefore, it is highly likely that enjoyment mediates the relationship between the U&G constructs and players’ ContInt to play PG, a likelihood that may be hypothesised as follows: H1a-d: Enjoyment mediates the relationship between each of the U&G constructs (Achievement, Challenge, Escapism, and Social Interaction) and Continuance Intentions to Play PG. 2.4 The Mediating Effect of Flow Flow is one of several factors that explain users’ behaviour in relation to various new media technologies such as online games (Merhi, 2016; Su et al., 2016; Lee, 2009), social networking services (Gao and Bai, 2014; Zhou, 2015), and mobile purchasing (Gao et al., 2015). It refers to the holistic feeling that people experience when they act with total involvement and also represents an optimal experience; in this state, people have clear goals, exercise control, lose their self-consciousness and experience a distortion of time (Csikszentmihalyi, 1990). The 7 of 35

literature has indicated that the various U&G constructs have a relationship with the flow state. For example, Wu et al. (2010) showed that acquiring achievements in online games would require players to pay attention, and this effort would contribute to the flow state. Moreover, according to Merikivi et al. (2017), positive challenges are the key to maintaining a user's interest and engagement in a game. If a game is too easy, players lose interest; if a game is too difficult, players grow frustrated. In addition, a study by Su et al. (2016) affirmed that challenges enabled players to experience a state of flow. Previous research has also shown that escaping from reality is a critical factor in inducing flow experiences while playing games. For instance, Liu and Chang (2016) found that escapism significantly influences the flow experience of online game players. When players are immersed in the game they escape unpleasant realities, which leads to a flow experience. Li et al. (2015) showed that social support in a particular social network game (SNG) helps players fulfil their social needs and motivates them to continue using the SNG. Moreover, Su et al. (2016) and Lee (2009) indicated that social interaction positively influences flow experience and further influences players’ behavioural intentions. It also appears that flow has an impact on a player’s intention to continue playing games. For example, Hoffman and Nadelson (2010) stated that when players try to achieve goals in a video game, the process fosters a flow experience and further enhances players’ motivational engagement. Furthermore, when players generate flow experience in the process of playing a game, a sense of full immersion is created, which enhances their gaming experience. Players who experience flow while playing games tend to continue their usage to obtain this optimal experience again in the future (Zhou, 2015). Thus, based on the above reasoning, this study proposes the following hypothesis: 8 of 35

H2a-d: Flow mediates the relationship between each of the U&G constructs (Achievement, Challenge, Escapism, and Social Interaction) and Continuance Intentions to Play PG. As emphasised above, flow has been identified as one of the factors that explain players’ behaviour and it usually occurs when players reach a critical level of engagement with a game. According to Liu and Li (2011), when players are in a flow state, they experience an optimal level of enjoyment. Similarly, Guo et al. (2016) reported that students who experience flow during an online course are more likely to perceive the course as fun. Further, Lee and Tsai (2010) suggested that total involvement, enjoyment, control, concentration, and intrinsic interest are important characteristics of the flow experience. Previous studies have also consistently found that enjoyment is relevant to continued use (Lee and Tsai, 2010; Merikivi et al., 2017). Looking at the model, it is highly likely that the mediators (viz., Flow and Enjoyment) are in a hierarchical causal relationship, with Flow affecting Enjoyment as well. Hence, the following mediating hypothesis is proposed: H3a-d: The relationship between each of the U&G constructs (Achievement, Challenge, Escapism, and Social Interaction) and Continuance Intentions to Play PG is sequentially and positively mediated by Flow and Enjoyment. 2.5 Social Influences and Continuance Intentions 2.5.1 Network Externality Katz and Shapiro (1985) defined Network Externality (NetExt) as the utility that a user derives from the consumption of a product, and it increases as the number of product users increases. The interaction and sharing between more users creates a greater sense of pleasure, which further leads to an improved experience and a higher level of satisfaction. Zhou and Lu (2011) examined 9 of 35

the effect of network externalities on mobile instant messaging and found that it significantly affected perceived usefulness and satisfaction and subsequently determined user loyalty. Further, Zhou (2015) reported that the referent network size has a significant influence on users’ ContInt for mobile social network sites. It is highly likely that NetExt would also have a similar impact on mobile gaming as well. PG became a huge social phenomenon with a large number of players. During the height of PG’s popularity, some public parks and cafés were literally swamped with hundreds of players hanging around virtual Pokestops (Kain, 2016). Next, we examine community involvement. 2.5.2 Online Community Involvement Hsu et al. (2012) defined an online community as a large, loosely knit, and geographically distributed group of individuals engaged in a shared practice of problem solving, knowledge exchange, or social interactions that mainly occur through computer-mediated communications. Online community involvement (CommInv), in the context of this study, refers to the behaviours of sharing and obtaining information from other users in the virtual world (Zhao et al., 2012). Most studies looking at online CommInv have examined factors that influence or affect this involvement (Kavanaugh et al., 2005). However, the current study seeks to find out whether stronger CommInv would lead to a stronger intention to continue playing. PG players are often members of various overlapping social communities, including groups on Facebook, where members share their experiences, news, and tips. In this study, CommInv refers to activities in which players interact in the PG social community (such as the PG Facebook group), for example, by sharing information and providing suggestions or opinions. Participation in virtual communities can lead to customer loyalty (Shang et al., 2006) and brand loyalty (Kim et al., 2004). Literature also suggests that online CommInv (Zhang et al., 2015) and engagement with a 10 of 35

virtual brand community (Brodie et al., 2013) positively impacts behaviours such as loyalty. Thus, it is highly likely that CommInv would have a mediating role in the relationship between network externalities and the intention to continue playing the game. This leads to the next hypothesis: H4: Online Community Involvement mediates the relationship between Network Externality and Continuance Intention to Play PG. 2.6 Gaming Motivational Factors and Continuance Intentions 2.6.1 Nostalgia Holbrook and Schindler (1991) defined nostalgia as a preference (general liking, positive attitude, or favourable affect) for objects (people, places, or things) that were common (popular, fashionable, or widely circulated) in the past (in early adulthood, in adolescence, in childhood, or even before birth). Derbaix and Derbaix (2010) stated that nostalgia derives from a desire to return to a time in the past, the wish to go back to the “good old days” or the belief that things were better in the past. On the other hand, Muehling et al. (2014) described nostalgia as an affective state or emotion that is somehow evoked by past events. Since the past is inaccessible, nostalgia increases consumers’ desire to affiliate with an idealised past and, in this way, experience emotional gratification (Belk et al., 2003). Koetz and Tankersley (2016) showed that nostalgia constitutes a predominantly positive emotion that affects the relationship between individuals and a brand. Moreover, Marchegiani and Phau (2011) found that various cognitive and attitudinal reactions significantly alter when consumers experience a higher intensity of nostalgia. Increasingly various brands, including watches (e.g., Omega), movies (the Star Wars franchise), fashion, phones (e.g., Nokia), and even cars (e.g., Volkswagen), are trying to appeal 11 of 35

to this nostalgia (Brown et al., 2003). Belk et al. (1991) also showed that nostalgia was one of the motivators of collecting. Cross (2016, p.1) referred to this phenomenon as ”nostalgic collecting,” associating it with fast-changing consumer goods, “especially those encountered in childhood and youth.” 2.6.2 Need-to-Collect Pokémon Monsters Collecting is a common human behaviour and is an important part of many people’s lives. According to Belk (1995), collecting is the process of actively, selectively, and passionately acquiring and possessing things removed from ordinary use and perceived as part of a set of nonidentical objects or experiences. Many individuals have a natural desire to collect items for various reasons, such as enjoyment, leisure, prestige, and set completion (Carey, 2008; Zolfagharian and Cortes, 2011). Clearly, the motivations behind collecting are complex and multifaceted, and it is notable that many of the motives offered as central to collecting revolve around the self (McIntosh and Schmeichel, 2004). PG’s slogan, “Gotta catch ‘em all,” highlights its core identity as a collection game; in this game, players assume the role of a trainer by catching and collecting Pokémon. The present study suggests that NeedCol may inspire players to hunt down all Pokémon and influence players to continue playing the game. In addition, a player's memory of the original Pokémon game might inspire the player to play PG in order to restore their fond memories of the game. PG evokes a childhood memory for many millennials who grew up with the franchise in the 1990s (Friedman, 2016; Hardy, 2016). Thus, nostalgic attachment was also expected to be one of the important factors influencing players to start playing and continue playing the game and it is highly likely that NeedCol plays a mediating role, hypothesised as follows:

12 of 35

H5: The Need-to-Collect Pokémon Monsters mediates the relationship between Nostalgia and players’ Continuance Intention to Play PG. 3. Methodology 3.1 Data Collection and Sampling A non-probability purposive sampling method was used to collect the data via an online survey (Su et al., 2016). Given the nature of online mobile games, access to the target respondents via an online survey was deemed appropriate (Lee, 2009; Merikivi et al., 2017). The dual-language questionnaire (English and Malay) was translated into Malay using a back-translation method. In the questionnaire, the Malay version was placed after the English version to improve comprehension and reduce common method bias. The goal of this study was to explain the key target construct—continuance intention to play. As such, the sample consisted of current players of PG. Respondents were recruited via popular PG online communities on Facebook (over 60,000 members) and Google Plus in Malaysia over a two-month period. A filter question was used to ensure that respondents met the criterion that they were still playing PG. The online survey yielded 362 usable responses after removing outliers via a preliminary test. Seven respondents had responses that were too consistent and have been removed, as they were considered suspicious (Hair, Hult, Ringle, and Sarstedt, 2017; Podsakoff et al., 2003). All the constructs were measured by means of reflective multiple indicators following Diamantopoulos et al. (2012), using a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). The process of creating the instrument for each construct began with examining

13 of 35

theoretical and empirical literature. Table 2 summarises the sources used to operationalise the constructs, which have been validated in the current literature. The instrument consisted of three sections. The first captured gaming experience, such as duration of play, play frequency, time spent playing, and in–app purchase details. The second recorded the respondents’ motivations and behavioural intentions for the game. The last section collected the demographic information of the respondents. Prior to the survey, the questionnaire was pre-tested on 20 respondents and reviewed by a Professor in the field of Information Systems to verify the appropriateness and completeness of the instruments and to confirm the content validity. Based on feedback from the respondents and the aforementioned expert reviewer, minor modifications were made to wordings of some questions to reduce ambiguity. For example, the term “shopped” was changed to “purchased” and the term “powerful” was explained by using the example of achieving a higher CP level. 3.2 Analysis Methods To estimate any structural equation model, empirical data could be analysed via the covariancebased (CB-SEM) or variance-based Partial Least Squares Structural Equation Modeling (PLSSEM) method. PLS-SEM was chosen for this study because of several important reasons. First, the benefits of testing mediation via PLS-SEM is that bias is reduced, as the bootstrapping procedure makes no assumptions about the nature of the distribution of the constructs or the sampling distribution of the statistics, and all the mediated interactions are examined concurrently rather than individually (Hair et al., 2014). In addition, testing multiple mediations simultaneously allows for better appreciation of the complete effect, by comparison with the separate testing of each mediator (e.g., with a regression model), of which the indirect 14 of 35

effects may be inflated due to the correlations associated with each mediator. Furthermore, mediation testing via PLS-SEM could be achieved with smaller sample sizes while still achieving higher levels of statistical power (Hair, Hult, Ringle, Sarstedt, et al., 2017) than with previous testing approaches like the Sobel test. Statistical power refers to how well the technique is able to determine significance among relationships with low, medium, and high effect sizes. Given that this study aimed to examine: (i) several indirect effects simultaneously (ii) in one complex model that (iii) combined and explored few theories while (iv) adding new concepts (e.g., Need-to-Collect and Nostalgia); PLS-SEM was deemed the most suitable. Moreover, PLSSEM has the ability to produce small parameter estimation bias when the nature of the sample population is unclear (Sarstedt et al., 2016), such as in this study and in most social science research in general. Finally, the main research goal was to explain and predict continuance intention to play PG via multiple indirect effects. PLS-SEM was selected because it was the preferred alternative to CB-SEM in research focussing on prediction (Shiau and Chau, 2016). 3.3 Model Estimation and Results Evaluation SmartPLS3 software (Ringle et al., 2015) was used in this study and the results were reported based on guidelines provided by Hair et al. (2017) and Ramayah et al. (2018). The analytical procedures were conducted in two stages: in the first stage, the measurement model was tested to assess the reliability and validity of the measurement instruments; in the second stage, the structural model was evaluated to examine the hypothesised relationships (Anderson and Gerbing, 1988).

15 of 35

4. Results Table 1 summarises the profiles of the respondents, showing that 61.3% of the respondents were male, 38.7% female, and the majority are between the ages of 18 and 35. More than half of the respondents had played the game for 3 months or more. [Place Table 1 about here] 4.1 Measurement Model The assessment of the reliability of the items reveals that 47 of 50 reflective items have an outer loading of >0.7 (Table 2). Only three indicators (i.e., SI5, FL3, and NE2) exhibit lower loadings of 0.638, 0.656, and 0.693, respectively. Following Hair et al. (2017), deletion was unnecessary as the composite reliability (CR) and average variance extracted (AVE) exceeded the threshold and those indicators contributed to the content validity of the measurement. Thus, the reliability of reflective items was considered satisfactory. Likewise, the reflective measurement model achieves a CR score of 0.804 and higher, supporting the internal consistency of the measurements in measuring each construct. In addition, all the values of AVE are greater than the critical threshold of 0.5, which provided evidence of the convergent validity of the measures. In terms of the discriminant validity assessment, two approaches were used. Firstly, the loadings of the items were examined and no cross loading with higher values were found with opposing constructs. Secondly, the most conservative discriminant validity test available to date was applied (Franke and Sarstedt, 2018): the heterotrait-monotrait (HTMT) ratio of correlations (Henseler et al., 2015).

16 of 35

There are two ways to examine discriminant validity via the HTMT technique: (1) using HTMT fixed cut-off or (2) applying HTMT-based inference test (Franke and Sarstedt, 2018). The first procedure is to ascertain whether the HTMT value of a construct approaches 1.0. The nearer it is to 1.0 (or the more it exceeds 1.0) would be interpreted as violation of discriminant validity. Henseler et al. (2015) recommended HTMT cut-off values of 0.90 or 0.85, while Voorhees et al. (2016) found that an HTMT cut-off value of 0.75 was more useful. The second procedure tested the null hypothesis (H0: HTMT≥1) against the alternative hypothesis (H1: HTMT