Multiplayer Online Battle Arena (MOBA) is a game genre in which two teams fight to destroy an enemy base, affiliated with Local and Wide Network (LAWN) games. In the virtual economy of MOBA games, game companies rely on trading cosmetic virtual goods (known as 'skins') with gameplayers in real currency (Mertens, 2017). Statista (2021) shows that online gamers worldwide have spent 43,220.1 million pounds on in-game purchases in 2020. As an essential part of game companies' business model, the leading factors of purchase intention of aesthetic virtual goods is an increasingly exciting research question in recent academia. The purchasing intention interest in gaming industries derives from several prominent theoretical models predicting web-based consumer behaviours. These previous studies provided a solid base for developing intrinsic factors of purchase intention in virtual social and game worlds (Guo and Barnes, 2009). Recent studies on determinants of purchase intention in the virtual game world are transforming from analysing single personal factors (Guo and Barnes, 2009; 2012; Hu and Wu, 2012) to the combination of personal and contextual factors (Rodríguez, 2017) and even the mediation effects of contextual factors with personal factors (Jang, 2021). Therefore, to continue exploring the factors influencing consumers' intention to purchase in virtual video games, this project focuses on personal determinants and contextual influences on motivation, specifically for MOBA in-game purchases. This project gains empirical insights by conducting online questionnaire surveys on how playing time, perceived emotional values and marketing strategies affect gameplayer's willingness to purchase virtual commodities in MOBA games.

Fisher's exact test was used for testing the project model. The findings contribute to the literature on MOBA in-game purchase intentions and render some marketing implications for MOBA game companies for having a sight on the personal factors and contextual factors when monetising virtual commodities.

2. Literature Review

With the rapid development of technology in the gaming industry and the increased               knowledge of the esports sector, research analysing determinants of purchase intent in         virtual worlds (VWs) reveals a segmentation of research content in different dimensions. In   terms of the contexts targeted, previous studies explore from the VWs, virtual social and       game worlds (Bleize and Antheunis, 2019), virtual video games, to specific online video        games type (such as standalone games, Massive Multiplayer Online games and LAWN        games) (Rodríguez, 2017). Additionally, the dimension of influencing factors that these          studies were interested in was developing from cognitive motivational (personal/intrinsic)       aspects to an integrated consideration of the influence of contextual factors. Studies              concerning intrinsic factors often investigate the factors mentioned in several prominent        theoretical models of predicting web-based consumer behaviours, such as the unified theory of acceptance and use of technology (Venkatesh et al., 2003), the Flow theory (Koufaris,       2002) and five theories of consumption values (Sheth et al., 1991). Most of them identify four important overarching factors influencing the purchase intention in VWs: perceived            enjoyment, social influences, customisation and ease of use (Bleize and Antheunis, 2019). On the contrary, studies concerning contextual factors often consider price and marketing strategies. This report reviews five previous pieces of literature in this field: the first three   studies are about the influences of personal factors, the fourth study is about the                combination of personal and contextual factors, and the last one contains the mediation     effects of contextual factors with personal factors. Apart from the first one, the remaining   research used an online questionnaire survey to gain empirical insights.

Guo and Barnes (2009) conducted an exploratory study investigating the relationship            between factors listed in Figure 1 and purchase behaviour in game-oriented VWs. They        conducted a semi-structured interview in four focus groups with selected Chinese virtual       world residents (N=24) based on the selected Chinese virtual world. They used NVivo           qualitative analysis software to analyse the collected data. The study finds that the factors    affecting purchase intention in virtual game worlds vary in different stages. The perceived     playfulness, character competency and requirements of the quest context positively              influence the foremost purchase intention. An enjoyable engagement experience can lead to the pursuit of virtual goods to fulfil gameplayer's intrinsic needs, leading to the possible         purchase of virtual goods in the virtual world. This study helps us understand factors that are probably important in virtual item purchase behaviour; however, the conclusion does not       have generalizability due to the limited selection of participants.

Figure 1: Research model based on Guo and Barnes (2009)

Based on the prior exploratory studies in 2009, Guo and Barnes (2012) developed four new possible factors: advancement, customisation, perceived enjoyment and perceived social    status in World of Warcraft (WoW). They used a structural equation model to test whether   these factors are important motivators triggering player pursuit of advanced, valuable virtual items. The finding shows that advancement, customisation and perceived enjoyment were  also significantly related to purchase intention. The direct relationship between social           influence and intention will most likely become non-significant for user purchase behaviour  intention unless it is a compulsory setting. The research model accounts for 37.3% of the    variance in WoW players’ purchase intention (R-square = .373). Additionally, this study fails to distinguish between day-to-day virtual purchase behaviour and advanced item purchase behaviour, and the conclusion may not hold in other VWs because the survey was carried   out in the context of WoW.

Figure 2: Research model based on Guo and Barnes (2012)

Thirdly, the study of Ho and Wu (2012) is an empirical investigation of the factors and theory of consumption values that affect the intent to purchase virtual goods in online games. The  study focused on whether game type, satisfaction with the game, character identification,     and theory of consumption values affect the purchase intention of virtual goods. They           collected data (N=523) from role-playing and war-strategy online game users who had          purchased virtual goods via an online questionnaire survey with seven-point Likert scales.    The study shows different results compared with previous research: game type determines  the factors influencing the purchase intent in online games. Role-playing game users are      affected by functional quality, playfulness, and social relationship support; war-strategy         game users are affected by satisfaction with the game, identification with the character, and price utility and playfulness. The adjusted R-square of overall models had a significant 0.03 variation; hence we can possibly believe the robustness of the findings. Regarding its         limitations, they stated that the difference in results might be due to the survey background of specific games and the choice of the convenience sampling method.

Figure 3: Research model based on Ho and Wu (2012)

Rodríguez (2017) analysed the underlying factors that influence the players of LAWN games (Counter Strike: Global Offensive) to purchase aesthetic virtual goods with real money. He   used a linear regression model to analyse the relationship. The study finds that emotional     value, fondness and perceived value of aesthetic virtual objects were significant indicators    that influenced the purchase intention of players. The gaming experience, such as                enjoyment, hours played per week, in-game rank and purpose of play, were also significant indicators but had a low impact (0.05). Demographic variables and game rank had no         significant influence. These influencing factors had solid explanatory power because they   caused a significant variation in the adjusted R-square coefficient when addressing the      gaming experience and emotional variables. However, the results might have been            significantly affected by failing to include some critical variables, and it also cannot address the data issues after conducting a questionnaire.

Lastly, Lang (2021) investigated how gameplay motivation and game-design factor (offering free items as a sales promotion) affect players’ willingness to purchase in-app contents and the mediation effects of free items with gameplay motivation on intention to purchase virtual items. The study found that play frequency, social interaction and item experience positively affected intention to purchase. The free-item experience effectively encourages the in-app    purchase of all gameplayers as a positive mediator. The level of stage has a negative           influence. In this study, over 30 per cent of the purchase intention was explained by the        factors above with direct effects. Regarding the research problems, an online survey only for one game may not be enough to determine the effect variances of key motivators in other    game genres. Also, the results may overemphasise the tendency of players who eagerly      want to obtain virtual items for free because it offered free items as rewards for participants.

Figure 4: Linear Regression model based on Rodríguez (2017)

Figure 5: Conceptual model based on Lang (2021)

3. Hypotheses

Playing game is a way of entertainment for most people, mainly for enjoyment (Jang, 2021). People tend to voluntarily invest more time and money in games when they feel satisfaction, enjoyment and a high sense of autonomy in the game (Jang, 2021). Besides, Balakrishnan  and Griffths (2018) found that online mobile game addiction (playing games more frequently and spending more time) enhanced the intention to purchase online mobile in-game apps.   According to previous studies, this project tests the effect of playing time on purchase          intention in MOBA games.

H1: Playing time is positively associated with the intention to MOBA in-game purchase of virtual goods.

Early and recent studies focusing on the determinants of purchasing intention in VWs have  one thing in common: all considered the intrinsic motivational factors of purchase behaviour. According to the Hedonic Effect of prestige-seeking consumer behaviour, some people buy goods for their own sensations, emotions and feelings than for the given functional utility      (Vigneron and Johnson, 1999). Hence, this project focuses on the MOBA genre, both           competitive-centred and primarily sells cosmetic items with no impact on gameplay, to          investigate the relationship between these virtual cosmetic items' perceived emotional          values and purchasing intention.

H2: Perceived emotional values are positively associated with intention to MOBA in- game purchases of virtual goods.

To continue explore the contextual factors, this study focuses on the effects of specific marketing strategies including sales promotion and discounts.

H3: Marketing strategies (Sales promotion or discounts) are positively associated with the intention to MOBA in-game purchases of virtual goods.

4. Methodology

This study uses quantitative research methods to test the hypotheses listed in section 3.      Because the qualitative research in social sciences toolkits is most accurate and suitable for investigating the research question in the gaming field of studies (Rodríguez, 2017). An        online questionnaire survey was distributed via WeChat among Chinese college                   gameplayers aged 18 to 25. To make the results more representative, we looked for people who play MOBA games, including Dota 2, League of Legends, Glory of Kings, Dream of       three ancient kingdoms 2, and 7f three ancient kingdoms. We asked them to send them to   their friends in the same way to conduct convenience sampling. We received 124 responses in 10 days (14th–24th April). Three incompetent responses were excluded from making the  data more uniform: two are beyond age scope, and one is not compatible with binary gender. Finally, 121 valid questionnaires were selected, and the average response time was 1 minute 23 seconds.

The survey information included a description of the research study and the survey website's link. The questionnaire was split into three parts to develop the survey instrument: consent    form, measures for the independent variables, and measures for the dependent variable.      The second block of questions revolved around the attitude towards virtual goods and sales promotions. The survey scales of "playing time" were five levels of playing time per week on average, that is "less than 1 hour", "1-5 hours", "6-10 hours", "11-15 hours", and "more than  15 hours". The scales of "perceived emotional values" and "marketing strategies" were five-  point Likert scales that ranged from 1 – "strongly disagree" to 5 – "strongly agree". Finally,    we used "spending" on MOBA games in the past year as a measure of dependent variable   "purchase intent", which scales were "0-50 RMB", "51-100 RMB", "101-150 RMB", "151-200 RMB", "201-250 RMB", "251-300 RMB" and "more than 300 RMB".

Figure 6: Descriptive Statistical Analysis of Samples

This report uses inferential statistical methods to test hypotheses. All variables were ordinal rather than nominal; using the Chi-square test will ignore the ordering of the rows or            columns. Therefore, we recoded the "playing time" as numeric variables scaled from 1 to 5 and recoded "playing time", "perceived emotional values", "marketing strategy", and            "spending" into binary variables. Figure 7 shows the recoded results of the survey. Then we used the Mann-Whitney-Wilcoxon U test to determine the association between ordinal         "playing time" and the binary version of "purchase intention" and conducted Fisher's exact   test to get the more exact probability. The results of Fisher's exact test identify the positive  and negative aspects of the association between variables through p-value and confidence interval and tell us how strong the relationship is through the odds ratio. Therefore, Fisher's exact test was used to determine if there was a significant association between "perceived  emotional values", "market strategies", and "purchase intention in MOBA games".

Figure 7: Recoded results of the survey

Figure 8: Contingency table for playing time and purchase intention

Figure 9: Contingency table for perceived emotional value and purchase intention