Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 1. Avatar Creation and Video Game Enjoyment: Effects of Life-Satisfaction, Game

Transcription

1 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 1 Avatar Creation and Video Game Enjoyment: Effects of Life-Satisfaction, Game Competitiveness and Identification with the Avatar Sabine Trepte and Leonard Reinecke, University of Hamburg and Hamburg Media School Katharina-Maria Behr, University of Zurich Paper presented at the 60th Annual Conference of the International Communication Association (ICA), June 2010, Singapore Correspondence should be addressed to: Sabine Trepte University of Hamburg Department of Psychology Von-Melle-Park Hamburg, Germany Phone: Fax: sabine.trepte@uni-hamburg.de

2 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 2 Abstract Based on the model of complex entertainment experiences (Vorderer, Klimmt & Ritterfeld, 2004), the competitiveness of a computer game (media prerequisite) and the individual life satisfaction (user prerequisite) are hypothesized to influence game enjoyment. Avatar-player similarity was hypothesized to determine identification with the avatar, which in turn was suggested to enhance the enjoyment experience. In a quasi-experimental study, (N = 666) participants were asked to choose the personality features of an avatar for six different game scenarios. The results demonstrate that both the games competitiveness as well as the participants life satisfaction influenced avatar choice and identification. Players created dissimilar avatars in competitive games and similar avatars in non-competitive games. Participants who were well satisfied with their lives created avatars that resemble themselves in terms of personality factors, whereas dissatisfied users created dissimilar avatars. Playeravatar similarity was positively related to identification. This correlation was significantly stronger for non-competitive games. Identification with the avatar was strongly related to game enjoyment. When controlling for the influence of identification on enjoyment, playeravatar-similarity was negatively related to enjoyment, suggesting that identity play can be an independent source of enjoyment in computer games. Keywords: video games, avatar, life-satisfaction, identification, enjoyment

3 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 3 Avatar Creation and Video Game Enjoyment: Effects of Life-Satisfaction, Game Competitiveness and Identification with the Avatar With regard to video and computer games, avatars are considered computer generated visual representations of people (Nowak & Rauh, 2005; Schroeder, 2002). The term avatar is a Sanskrit word and identifies the god Vishnu s manifestations on earth (Castronova, 2002). Player representations as a part of a game can be traced back to ancient games that were played more than 4000 years ago, but the 1985 multi-user domain Habitat was the first to use the term avatar for user representations in virtual worlds (Castronova, 2002). Avatars are not only used in games, but also as users representations in e-commerce applications, social virtual environments, and in geographically separated workplace meetings (Schroeder, 2002). Avatars have become an integral part of video and computer games. For example, in The Sims 2, players can not only change their avatar s appearance, but also personality traits. Especially in MMORPGs (e.g., World of Warcraft, Eve Online), users can choose from a variety of features in order to manipulate the appearance, character, skills, and in some games even the ancestry of their avatars. Therefore, Chan and Vorderer (2006) suggest expanding the rather technical definition of avatars as graphical representations to defining avatars as characters that represent users in a virtual environment and community. Avatars are not just the interface players use in order to control the game, but may also represent the players own or created identity (Asimina & Joinson, 2009; Bessière, Seay, & Kiesler, 2007; Kafai, Fields, & Cook, 2007; Kang & Yang, 2006). While players may choose different avatar features and appearances for example in MMORPGs, they have to use avatars with pre-defined appearance, character, and history in many other games. One prominent example is the scientist and main protagonist Gordon Freeman from Half-Life and Half-Life 2. In these games, players cannot choose or create an individual avatar, but take on the role of a pre-defined character. This type of avatar is very typical for competitive games with a rather linear structure compared to the often less linear

4 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 4 role playing games. Pre-defined avatars are designed rather in order to fulfill the games requirements, e.g. in terms of the avatars skills, than to provide various possibilities for the users to adapt the avatar according to their preferences. Pre-defined avatars can especially be found in action games, e. g. undercover agent Sam Fisher in Splinter Cell, agent 47 in Hitman, and Lara Croft in Tomb Raider. But not all competitive or linear games feature pre-defined avatars only. Sports games played on Nintendo's Wii are an illustrative example. The avatars on Wii consoles are called Miis, and players can either choose from pre-defined ones or create individual Miis according to their preferences. A Mii s outward appearance can be customized e.g. according to body-mass, dress-color, haircut or liver spots. A Mii can be used across different games like Bowling or Tennis, and players can save their Miis on Nintendo's remote control in order to use their personal avatar when meeting friends and playing together on someone else's Wii. Previous research suggests that players prefer similar avatars in terms of gender, outward appearance and even sex (Hsu, Lee, & Wu, 2005; Rymaszewski et al., 2007; Trepte, Reinecke, & Behr, 2009). Hsu, Lee and Wu (2005) conducted a qualitative study with 16 frequent buyers of computer games. Participants assigned 28 different Pac-Man games to three groups in terms of perceived fun. Afterwards they compared games from different groups according to design features. Hsu, Lee and Wu (2005) report that players enjoyed games with characters similar to their own more than games with dissimilar characters. A preference for same-sex avatars can also be found for virtual environments such as Second Life. Research on the gender preference of users of Second Life has shown that only very few females choose to play a male character (4 %) and not many men choose a female character (14 %). Most users (82 %) stay with their real world gender (Rymaszewski et al., 2007). Nevertheless, in some situations, players might also want to play with dissimilar avatars. First exploratory studies revealed that gamers try out gender swapping as a form of identity play (Hussain & Griffiths, 2008). Kafai, Fields and Cook (2007) investigated the

5 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 5 virtual world Whyville. In their interview study, they found that avatars also generated avatars dissimilar to the users self to allow for identity play and ameliorated self-representation. In the World of Warcraft study conducted by Bessière, Seay and Kiesler (2007) players reported that their characters are dissimilar in terms of extraversion, conscientiousness and neuroticism. Considering the empirical studies cited above, it can be assumed that dissimilar avatars are preferred to compensate deficits in real-life personality (e.g. depressive affect or introversion), or if the game requirements ask for certain avatar features (e.g. extraverted, risk-taking, fearless in action games) to succeed in the game. The present study complements previous research that demonstrates a positive relationship between avatar-player similarity and enjoyment (e.g., Hsu et al., 2005; Trepte et al., 2009) by the notion that similarity does not necessarily lead to game enjoyment. Depending on individual factors such as life-satisfaction, some players might enjoy different avatars whereas others prefer dissimilar avatars. Also, the avatar-player similarity might be determined by game requirements, such as the game s task-structure or competitiveness. Hence, it is proposed here that players might identify with similar as well as dissimilar avatars. Depending on player as well as on game characteristics, both similarity as well as dissimilarity might lead to identification, which, in turn will influence experiences of enjoyment. It is proposed here, that the fit between game and player characteristics is a presupposition of game enjoyment. The right combination of attributes of the user and of the media stimulus makes an entertainment experience enjoyable and fun. Game enjoyment, just like the enjoyment of other entertaining media stimuli, varies according to media and user prerequisites (Vorderer, Klimmt, & Ritterfeld, 2004). In research on media entertainment, hardware details such as screen size, as well as user states such as psychological needs (Brock & Livingston, 2004), moods (Oliver, 2003; Zillmann, 2000), motives (Vorderer, 2003), and also trait variables and personality dimensions (Hartmann & Klimmt, 2006b; Weaver, 2000), have been considered to determine media enjoyment.

6 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 6 Accordingly, media enjoyment depends on the interplay of media characteristics and user attributes. Based on the Model of Complex Entertainment Experiences (Vorderer et al., 2004) it will be supposed in the present investigation that media enjoyment is determined by a fit of media and user prerequisites. Complex Entertainment Experiences In their model of complex entertainment experiences Vorderer, Klimmt and Ritterfeld (2004) suppose that media users search for positive media experiences. The model is used here as a frame of reference to emphasize that both, user and media prerequisites, have to be taken into account for a systematic consideration of media entertainment. Of course, the empirical study presented in this paper will not be able to prove the model as a whole. In this paragraph, the entire model will first be presented by elaborating on its theoretical groundwork. By adding a new component ( primary action ) and by implementing recent work (on the reappraisal of action ) into the model (Vorderer & Hartmann, 2008), the model is modified to increase its fit to interactive entertainment experiences. Subsequently, the three following paragraphs will elaborate in detail on the model components relevant to the empirical study presented in this paper: (1) the media prerequisites, (2) the user prerequisites, (3) the user s primary actions and the appraisal of these actions. Within the model of complex entertainment, Vorderer, Klimmt and Ritterfeld (2004) integrate Zillmann s (1988a,b) assumption of the media user as a hedonistically driven agent, as well as research in neuroscience (Davidson, 2002). Zillmann (1988a,b) has proven in numerous experiments that media users strive for positive experiences and optimize their level of stimulation (Bryant & Zillmann, 1984; Knobloch, 2003; Knobloch & Zillmann, 2002; Oliver, 2003; Zillmann, 2000). His mood management theory posits that recipients learn pleasurable experiences and search for media content fitting their needs for hedonism and optimal stimulation. Zillmann s (1988a,b) approach is situated in emotional psychology. Complementary, neuroscience is of growing importance for research in media psychology,

7 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 7 because it might be able to reconcile the divide of cognitive and affective approaches to entertainment effects. Research in neuroscience found simultaneously activated affective and cognitive circuits in the brain during information processing. Thus, Vorderer, Klimmt and Ritterfeld (2004) propose that media enjoyment refers to physiological, affective, and cognitive dimensions. They formulate a broad conceptualization to improve the understanding of entertainment premises. In this understanding media entertainment is accompanied by a pleasant state termed enjoyment. Media enjoyment might have different kinds of manifestations such as suspense, serenity, and laughter as well as sadness or melancholy. In this first conceptualization, the authors posit user prerequisites, media prerequisites, and motives as conditions for media enjoyment. User prerequisites comprise empathy (Zillmann & Bryant, 1985), presence (Lombard & Ditton, 1997), interest (Greenwald & Leavitt, 1985), or affinity to the character (Cohen, 2006). At least one of the media user s prerequisites has to be addressed in order to feel entertained (cf. section user prerequisites ). The design and content of media are identified as media prerequisites that may elicit pleasurable media experience. In terms of user motives the authors consider escapism (Katz & Foulkes, 1962; Moskalenko & Heine, 2003) as well as achievement or competition (J. L. Sherry & Lucas, 2003). The motives have been omitted in more recent work on the model (Vorderer & Hartmann, 2008) and will not be considered in detail here. In recent work on the model, Vorderer and Hartmann (2008) underline the emotional access to media use. They state that the interaction of the user and the media environment leads to a subjective perception of the salient environment and elicits primary responses such as physiological reactions as well as positive and negative emotional responses. In a second step, this primary response is reappraised according to mood-regulation and self-realization goals. Media users might feel entertained after this reappraisal and adjust the perception of their environment. User prerequisites (e.g., personality or life-satisfaction), user motives (e.g., mastery of the game) and media prerequisites (e.g. competitiveness of the game) may

8 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 8 determine the users preliminary reactions such as positive and negative emotions. In traditional entertainment media, such as television, the first reaction might be emotionally or cognitive. Television makes people happy or sad (Oliver, 1993), it makes them think about their own lives and their identity (Trepte, 2006). These preliminary responses also do occur in computer games, but they are very often accompanied by overt behavior. Gamers might feel anger about an enemy (Mullen & Skitka, 2006), think about how they might react, and additionally, take action in the gaming environment. The preliminary responses in computer games are behavioral. They are observable as moves and strategies. One of the very first active steps in many games is the choice of an avatar. This usually takes place right before a game or a game-session begins. To create an avatar and to decide what the avatar should be like, players might be influenced by their individual goals and thoughts as well as the game s requirements and its story (Kafai et al., 2007; Trepte et al., 2009). Different strategies of avatar creation are possible (cf. section Effects of Avatar-Player Similarity on Identification and Enjoyment ). After preliminary responses, the users reappraise how the media use fulfills their goals and needs (e.g., mood regulation or self-realization goals) and how they feel in the present situation (Vorderer & Hartmann, 2008). Reappraisals are necessary and useful to comprehend primary emotional reactions and to connect them with elaborated reference-frames (Fiske & Taylor, 1991). The reappraisal of preliminary reactions might be consolidated as a sense of identification (Cohen, 2001; Cohen, 2006), interest or involvement (Greenwald & Leavitt, 1985). In the contrary, the reappraisal might also lead to a sense of denial or even disgust. Summing up the model of complex entertainment experiences (Vorderer, Klimmt & Ritterfeld, 2004; Vorderer & Hartmann, 2008), it can be stated that the interplay of media prerequisites, individual user prerequisites and motives leads to a primary user response. In the case of computer games, this primary response is accompanied by overt behavior. After their primary response, players reappraise their primary action. By reappraising primary

9 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 9 actions, the user connects the gaming situation with individual goals and tries to evaluate, whether these are supported by the gaming experience. Positive reappraisal may result in feelings of entertainment. In this paper, the interaction of game competitiveness, satisfaction with life, playeravatar similarity, and identification will be taken into account to predict media enjoyment. Whereas game competitiveness represents a media prerequisite and life-satisfaction a user prerequisite, identification can be considered the result of an appraisal process that is directly influenced by the interaction or fit of media and user prerequisites. Both game competitiveness as well as the user s life-satisfaction are likely to influence the choice of avatar features and, as a result, player- avatar similarity and game enjoyment. Within the following three paragraphs, the model components (1) game competitiveness as a media prerequisite, (2) life satisfaction as a user prerequisite and 3) the primary action and its reappraisal are considered. Game Competitiveness as a Determinant of Avatar Creation Technical media prerequisites have fascinated numerous scholars particularly in computer and engineering sciences, but also in communication research and media psychology (Biocca, 1996; Heeter, 1995; Wann & Mon-Williams, 1996). Hardware details such as the screen size (Reeves, Lang, Kim, & Tatar, 1999), a game s design (e.g.`, game dynamics and mechanics`, for a discussion cf. Sellers, 2006) as well as the games content (e.g., narrative or game characters) have been applied as independent variables affecting media enjoyment (e.g., Hsu et al., 2005; Ravaja et al., 2004; Vorderer, Hartmann, & Klimmt, 2003). In media psychology, game prerequisites have primarily been investigated in relation to the user needs they address. In psychological research on video and computer games, competition has received considerable attention as a key prerequisite for the enjoyment of games (Vorderer et al., 2003). Games grant their users a high degree of control over the gaming environment (Grodal, 2000) and the player s actions result in immediate feedback

10 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 10 from the game (Klimmt & Hartmann, 2006). While feelings of mastery and effectance are crucial components of the entertainment experience elicited by most games, the degree of competitiveness differs substantially between games. While some games challenge the player by putting high demands on the players problem solving skills (e.g., adventures, puzzle games or simulations) others include human or computer-controlled opponents the user has to compete against (e.g., shooters or action adventures). The positive effect of competitiveness on game enjoyment is illustrated in an experiment by Vorderer, Hartmann and Klimmt (2003). Participants read verbal descriptions of a gaming situation that either featured the necessity to compete against opponents or no such competitive elements. Participants were then asked to rate the enjoyment associated with the respective gaming situation. The data illustrates that participants rated game descriptions featuring competitive elements as significantly more entertaining than gaming situations that do not offer the chance to compete against opponents. In addition to game enjoyment, the competitiveness of a game is also likely to influence the choice of avatar features. Trepte et al. (2009) have argued that in games with clearly predefined game requirements, users can be expected to show a stronger tendency to create avatars with features very close to the game requirements. Trepte et al. (2009) asked players to create their own avatars for seven different game descriptions. They were advised to create avatars they would enjoy playing with. Both male and female participants created avatars with gender attributes that best met the games requirements, even if these attributes conflicted with the players sex role. Accordingly, women created manly avatars, if the game asked for (pre-tested) male characteristics and men created female avatars, if the game asked for (pre-tested) female characteristics. At the same time, both male and female players rated those gaming scenarios as more entertaining that allowed them to choose avatar features that were in accordance with their respective gender role. These results reveal that the avatar s features are chosen in accordance with the game s demands to facilitate mastery of the game.

11 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 11 In a competitive shooter game, an extraverted and dominant avatar appears to be more effective in reaching the game s goals than a creative and sensitive character. Furthermore, avatars in competitive games often require supernatural powers and thus are likely to differ substantially from the player s personality profile. In contrast, non-competitive games (e.g. simulations such as The Sims) are less likely to require highly specialized avatar features. Consequently, players are likely to experience greater freedom in their choice of avatar features and are less restricted by game requirements compared to more competitive games (Klimmt, Hartmann, & Frey, 2007). As competitive games allow for a lesser degree of freedom in avatar creation than non-competitive games, these observations allow for the conclusion that users will prefer similar avatars as long as the gaming environment grants them the freedom to choose. We thus propose the following hypothesis (all variables and their hypothesized interrelations are depicted in Figure 1): H 1: In non-competitive games, players will equip their avatars with personality attributes that bear a higher resemblance to their own personality than in competitive games. Life-Satisfaction as a Determinant of Avatar Creation The media user s appraisal of his/her life might interact with entertainment experiences. Previous research addressed how the use of entertainment media affects life-satisfaction (for an overview cf. Schreier, 2006). Lately, some first steps have been taken to explore the effect of life-satisfaction on entertainment choices. In self-discrepancy theory, Higgins (1987) suggests that we have three types of self-schema: the actual self (who we currently are), the ideal self (how we would like to be) and the ought self (how we think we should be). The ideal self is associated with promotional goals that individuals strive for and the ought self with prevention goals. Discrepancies between the actual self and the ideal or ought selves can motivate change to reduce discrepancies. Applying self-discrepancy theory to video games, Bessière, Seay and Kiesler (2007) showed that players in the MMPORPG World of Warcraft rated their character as having more favorable attributes than they rated themselves

12 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 12 on the same dimensions (Bessière et al., 2007). Particularly players with lower self-esteem and depressive affect rated themselves (actual self) as being significantly less conscientious, less extraverted and more neurotic than they rated their character. Thus, players with low levels of psychological well-being equipped their avatars with more favorable personality attributes than they felt to possess themselves. Players with high self-esteem and no depressive affect rated their actual self and their game character similar in terms of personality structure. The authors suggest that the game world gives players the freedom to create successful and favorable virtual selves regardless of their actual situation (Bessière et al., 2007). Applying the research results reported above to the creation of avatars, it might be conceded that people being dissatisfied with their life strive to create an avatar that has personality features very dissimilar from themselves. The creation of a dissimilar avatar would change the focus from an unpleasant or dissatisfactory real-world personality to a more satisfying representation within the virtual world. The creation of an avatar in terms of personality allows for altering exactly those personality features that players perceive as interfering with their aims and that are bothering them. H 2: Participants scoring high in life-satisfaction will equip their avatars with personality attributes that bear a higher resemblance to their own personality than participants scoring low in life-satisfaction. Effects of Avatar-Player Similarity on Identification and Media Enjoyment Media and user prerequisites have been covered in the two preceding paragraphs, both of which are interrelated and are likely to affect the player s actions (i.e., avatar creation) prior to gameplay. Before creating an avatar, the player evaluates the plot of a game, its goals and features (media prerequisites) and his own state of mind at this point of time (user prerequisites). Based on both evaluations, the primary action is initiated. This primary action

13 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 13 reflects the interplay of user and media prerequisites. In the following, the avatar creation as an example of the primary action will be considered. After the creation of an avatar, players are likely to reappraise whether their avatar serves their goals and needs such as mastery of the game or sociability (Jin & Park, 2009; Klimmt et al., 2007). They might consider how the avatar fits in with their personal aims and the game requirements (Derrick, Gabriel, & Tippin, 2008). We suggest that the result of this appraisal process (positive vs. negative appraisal) has an influence on identification with the avatar. Identification has been defined by Cohen (2001) as an imaginative process through which an audience member assumes the identity, goals and perspective of a character (p. 261). Identification goes far beyond similarity. It not only describes whether a character and the user are matching in certain respects, but also whether they are merging (Vorderer, Klimmt, & Hefner, 2009). The more the user becomes the media character, the more he or she identifies. Identification is an imaginative process and it defines whether a user and the avatar are connecting emotionally and cognitively (Cohen, 2006). During gameplay, users imagine themselves being the avatar and act out of the avatar s perspective (Vorderer et al., 2009). For players, it should be easier to merge with a character that has a personality structure similar to their own. Thus, identification should be stronger if the avatar and the player are more similar. H 3: Players with a personality profile similar to their avatar s profile, identify more with their avatars than players with a personality profile dissimilar to their own. Following the model of complex entertainment experiences (Vorderer, Klimmt & Ritterfeld, 2004), we suggest that identification is an outcome of the reappraisal of primary user action. Identification may thus produce media enjoyment, if the user feels that his strategy of avatar creation is successful in terms of his or her personal goals. On the contrary, an ineffective strategy could cause boredom or disappointment. Identification seems particularly interconnected with the enjoyment of computer games, because the user himself

14 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 14 influences how much he or she becomes a certain character (Vorderer et al., 2009). The gamer has the chance to create avatars and can thus actively influence the avatar-player relationship. Identification has been shown to predict enjoyment of computer games. Hefner, Klimmt and Vorderer (2007) conducted an experiment with 30 participants who either played the game Battlefield 2 for six minutes or watched a video with a recording of the game. The players experienced significantly more identification with the game characters than the participants in the (video-exposure) control group. The authors reported a significant correlation between identification and enjoyment of r =.58 (p<.001). Thus, the following hypothesis seems justified: H 4: Identification with the avatar is positively related to video game enjoyment. As suggested above, satisfied people might create similar avatars (H2) and avatarplayer similarity might elicit higher levels of identification (H3). As identification is expected to increase media enjoyment (H4), it can be concluded that similarity leads to media enjoyment. In fact, this positive relationship between similarity and enjoyment was found in previous work (Hsu et al., 2005; Rymaszewski et al., 2007). It will be suggested here that the positive relationship between player-avatar-similarity and enjoyment is mediated by the positive correlation between player-avatar similarity and identification. In other words: Avatar-player similarity influences identification, which, in turn enhances game enjoyment. H 5: Identification mediates the relationship of avatar-player similarity and game enjoyment. As discussed above, users are likely to identify with similar avatars, if their lifesatisfaction and game prerequisites ask for a similar avatar but they may also identify with a dissimilar avatar, if game requirements and their life-situation demand for a dissimilar avatar. In their experiment, Hefner, Klimmt and Vorderer (2007) asked players in how far a game description fitted to their self-image. The results demonstrated that a higher fit correlated with higher identification with the avatar. This suggests taking a closer look at the interaction of

15 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 15 identification and a user s evaluation of the game s content. It seems plausible to assume that players are more likely to identify with a game character, if they generally support the game s narration or wish to be like the character. It is suggested here, that a game s competitiveness moderates the relation between avatar-player similarity and identification: In competitive games, players try to meet the game s requirements as good as possible and thus identify best with an avatar that has a personality far beyond reality, thus very dissimilar to their own personality. In contrary, in non-competitive games, identification processes such as the experienced merging of avatar and player are best achieved by similar avatars. H 6: The negative effect of the difference between player and avatar on identification with the avatar will be stronger for non-competitive games than for competitive games. Prior research demonstrates the relevance to explore gender differences in the uses and effects of interactive entertaining media (Hartmann & Klimmt, 2006a; Lucas & Sherry, 2004; Trepte et al., 2009). The hypothesized interrelations between media and user prerequisites under investigation in this study (cf. Figure 1) might differ between male and female users of video games for a number of reasons. Male and female players show different preferences for game genres. Especially violent and highly competitive games are more popular among male compared to female players (Hartmann & Klimmt, 2006a). Women show a lower need for competition (Hartmann & Klimmt, 2006a; Lucas & Sherry, 2004) and competitive behavior is stronger related to male sex-stereotypes than to female sex-stereotypes (Lueptow, 2005). Results from Trepte et al. (2009) demonstrate that both men and women prefer games that require avatars that are in accordance with their own sex role. It thus appears reasonable to assume that the correlational patterns among avatar-player similarity, identification, and enjoyment addressed in the hypotheses above might differ for men and women. To address these potential gender differences, the following research question was posed: RQ: Do the correlational patterns addressed in hypotheses 1 to 6 differ among male and female participants?

16 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT Pls. insert Figure 1 about here Method Pretest of Stimulus Material: Competitive vs. Non-Competitive Games In a pretest, 28 participants (29% men, 71% women; mean age = yrs., SD = 5.15) rated the competitiveness of six descriptions of computer games. The six game descriptions were largely taken verbatim from those of existing computer games (Grand Theft Auto: San Andreas, Crysis, Urban Chaos, The Sims, My Animal Hospital, and Second Life). They were obtained from the official product websites. All cues referring to the original games (e.g. titles or character names) were excluded from the descriptions (see Appendix). Competitiveness was assessed with two items: This game is about winning (scale ranging from 1 = does not apply at all to 5 does fully apply ) and This game contains competition with other players or competitors (scale ranging from 1 very little to 5 very much ). The two items showed high internal consistency over all game descriptions (Cronbach s α =.808) and were summed up to form a single index of competitiveness. Mean competitiveness ratings were computed for every game description and one-sample t tests were computed to test for significant deviations from the scale midpoint (3.0). Descriptions rated significantly higher than 3.0 were categorized as competitive games. This was the case for game one (GTA: San Andreas, M = 4.36, SD = 0.62), game two (Crysis, M = 4.43, SD = 0.88) and game three (Urban Chaos, M = 4.14, SD = 0.71), all t(27)s > 8.57, all ps <.001. Analogously, descriptions rated significantly lower than 3.0 were considered non-competitive games. This was the case for game four (The Sims, M = 1.73, SD = 0.73), game five (My Animal Hospital, M = 2.25, SD = 0.83) and game six (Second Life, M = 1.63, SD = 0.94), all t(27)s < , all ps <.001.

17 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 17 Main Study Participants A total of 666 persons participated in the main study. Participants were recruited in introductory psychology classes at a large University in Germany (n = 233) and on the website of a popular German gaming magazine ( (n = 433). Student participants received course credit for participation. The sample comprises 457 men (68.6 percent) and 209 women (31.4 percent). Their ages ranged from years (M = 23.2 yrs.; SD = 5.37 yrs). None of the participants of the stimulus pretest took part in the main study. Measures Personality factors. The 10-item version of the Big Five Inventory (BFI-10`, Rammstedt & John, 2007) was used to measure the personality profiles of the participants and the avatars they created for each of the six game descriptions. The BFI-10 comprises five subscales measuring the Big Five personality factors (Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness), with two items per subscale. Participants rated each item (e.g. I see myself as someone who is outgoing, sociable ) on a 5-point scale ranging from 1 disagree strongly to 5 agree strongly. For the rating of the six avatars, the prefix of each item ( I see myself as someone who ) was changed to This Avatar (e.g. This Avatar is outgoing, sociable ). Despite the small number of items, the BFI-10 shows sufficient validity for research purposes (Rammstedt & John, 2007). In a principle components factor analysis, the five-factor structure found by Rammstedt & John (2007) could be replicated for the personality ratings of the participants of this study. The mean factor loadings ranged from.790 for Agreeableness to.884 for Extraversion, whereas the Cronbach s alphas of the two-item subscales ranged from α =.435 for Agreeableness to α =.819 for Extraversion. Player-avatar difference. To compute a global indicator for the difference in the personality profiles of the players and their avatars, the participants five BFI-10 sub scores

18 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 18 (Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness) were subtracted from the respective sub scores of avatar personality. The absolute values of these differences were averaged to form a single indicator of player-avatar difference for each of the six avatars. Identification with avatar. Two items ( I could fully identify with this avatar and I could fully immerse myself in this avatar ) were used to measure identification with each of the six avatars created by the participants. Participants rated the two items on a 5-point scale ranging from 1 strongly disagree to 5 strongly agree. The items showed high internal consistency over all game descriptions (Cronbach s α =.79). Game enjoyment. The entertainment expectation regarding the six game descriptions were assessed with two items ( I could have a lot of fun with this game and I could have a lot of fun with this avatar ) rated on a 5-point scale ranging from 1 strongly disagree to 5 strongly agree. Internal consistency of the items was high (Cronbach s α =.89). Satisfaction with life. Participants general life satisfaction was assessed with the Satisfaction with Life Scale (SWLS`, Diener, Emmons, Larsen, & Griffin, 1985). The SWLS consists of five statements (e.g. In most ways my life is close to my ideal. ) regarding an individual s general satisfaction with life. Participants were instructed to rate each item on a 7-point scale ranging from 1 strongly disagree to 7 strongly agree. The items of the SWL- S showed a satisfactory internal consistency (Cronbach s α =.86). Procedure Participants recruited in university classes and on the gaming websites received the same online questionnaire. In contrast to the web-sample, student participants answered the survey in a university computer lab. Upon arrival on the project website, participants read an introduction text informing them about the scientific purpose of the research project and granting them anonymity in case of participation. On the following pages of the survey, participants received the six pretested computer game descriptions in random order. After a

19 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 19 game description was presented, participants were asked to think of an avatar they would like to play with for the respective game. Participants were instructed to create an avatar by equipping it with personality traits assessed with the 10-item version of the Big Five Inventory (BFI-10`, Rammstedt & John, 2007). Furthermore, participants rate identification with the self-created avatar and the entertainment value of every game description. In the final section of the survey participants rated themselves on the Satisfaction with Life Scale (Diener et al., 1985) and the BFI-10 and reported age, sex, the average number of times they play video or computer games per week and the average length of a usual gaming session. In order to assure high external validity, game descriptions were based on existing games. Cues leading to an identification of the games were removed, because recognizing the games behind the description may have influenced the participants choices. As described above, the participants read game descriptions and indicated their preferred avatar's personality using a questionnaire. They did not create avatars in real games. This approach was chosen to improve internal validity. Although various computer and video games allow users to create their own avatars, this is not always the case. Moreover, avatar creation processes and tools vary across games. By using vignettes, we kept features of avatar creation invariant across different games and eliminated variables that might have caused confounding effects, such as visual game characteristics or different sounds. Results For data analysis, the original data file was restructured. To allow for the analysis of all avatars in a single statistical model, every avatar designed by the participants was taken as a separate unit of analysis. Thus, the player-avatar difference, identification with the avatar and media enjoyment scores of all six avatars designed by each participant represent separate cases in the data file, resulting in 666 (number of participants) x 6 (number of avatars per participant) = 3996 cases. All four variables considered in hypotheses two to six showed

20 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 20 significant zero order correlations. Satisfaction with life was negatively related to the playeravatar-difference (r = -.16, p <.01), to identification with the avatar (r = -.06, p <.01), and to media enjoyment (r = -.06, p <.01). The player-avatar-difference showed significant negative correlations with identification with the avatar (r = -.24, p <.01) and with media enjoyment (r = -.06, p <.01). Finally, identification with the avatar was positively correlated with media enjoyment (r =.62, p <.01). Hypothesis one predicted that in non-competitive games, players will equip their avatars with personality attributes that bear a higher resemblance to their own personality than in competitive games. An independent samples t test was computed to compare the playeravatar difference of avatars created for competitive vs. non-competitive games. Games were categorized as competitive (GTA: San Andreas, Crysis, Urban Chaos) or non-competitive (The Sims, My Animal Hospital, Second Life) according to the results of the pretest reported above. The results of the t test revealed that avatars created for competitive games showed a significantly higher player-avatar difference (M = 1.15, SD =.47) than avatars created for non-competitive games (M =.88, SD =.45), t(3994) = , p <.001, r =.28. Accordingly, participants showed a preference for dissimilar avatars in competitive and for similar avatars in non-competitive games. Consequently, the first hypothesis was supported by the data. Hypotheses two to six were tested with a structural equation model computed with AMOS 7.0 (Arbuckle, 2006). The hypothesized model is depicted in figure 1. All variables, except player-avatar difference, were modelled as latent variables. The five items of the Satisfaction with Life Scale (Diener et al., 1985) were used to estimate the latent construct of satisfaction with life. The latent constructs of identification with the avatar and media enjoyment were estimated from the four items listed in the measures section. Player-avatar difference was measured with a single indicator (cf. measures section) and thus was entered into the model as an observed variable. The measurement models led to a satisfactory description of the latent constructs.

21 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 21 The relationships between the studied variables outlined in hypotheses two to six were all tested in the model. Three indicators were used to assess the fit of the predicted model, the χ² test, the root mean square error of approximation (RMSEA) and the comparative fit index (CFI). The model fit the data well, χ²(32) = , p <.001, RMSEA =.058, CFI =.977). Although the significant χ² value indicates suboptimal model fit, this is most likely an effect of the large sample size of the present study (pls. note that 3996 cases were in the file = 666 (number of participants) x 6 (number of avatars per participant)). As Byrne (2001) notes, the χ² test tends to underestimate model fit for larger samples significantly. In contrast to the χ² statistic, the RMSEA (>.8`, Browne & Cudeck, 1992) and CFI (>.9`, Bentler, 1990) values of the tested model indicated adequate fit (MacCallum, Browne, & Sugawara, 1996). As predicted in hypothesis two, the model revealed a significant negative relationship between satisfaction with life and the difference between player and avatar personality (β = -.17, p <.001). Accordingly, players with higher satisfaction with life created avatars that show a higher resemblance to themselves than participants with lower satisfaction with life and the second hypothesis is supported by the data. Hypotheses three predicted that the observed difference between the players personality profile and their avatars is negatively related to identification with the avatar. It was assumed that players with similar avatars find it easier to become their avatars. This assumption was supported by the data. The player-avatar difference was significantly negatively related to identification with the avatar (β = -.26, p <.001). Accordingly, the participants tended to identify more strongly with avatars that showed a higher resemblance to themselves than with avatars that showed a stronger difference to the player. Furthermore, as predicted in hypothesis four, the model indicated a strong relationship between identification with the avatar and game enjoyment (β =.77, p <.001). Accordingly participants experienced more media enjoyment in case of games that allowed for a stronger identification with the avatar which supports hypothesis four.

22 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 22 As predicted in hypothesis five, the model showed a significant positive correlation between player-avatar difference and media enjoyment (β =.11, p <.001) when controlling for identification with the avatar. Accordingly, the originally negative direct relationship between the observed player-avatar difference and media enjoyment (r = -.06, p <.01) turned into a positive relationship when controlling for the mediating effect of identification with the avatar. Hypothesis five was thus supported by the model. To test the moderating effect of type of the game s competitiveness on the relation between player-avatar-difference and identification that was predicted in hypothesis six, a two-group model differentiating between competitive and non-competitive games was computed (pls. see figure 2). Again, games were categorized as competitive or noncompetitive based on the results of the pretest reported above. Accordingly, the cases of all avatars created for games one to three were used to estimate the model for the group of competitive games, whereas the remaining cases of avatars created for games four to six were used to estimate the model for the group of non-competitive games. All variables were modelled as described for the initial model testing hypotheses two to six. The resulting model fitted the data well, χ²(64) = , p <.001, RMSEA =.041, CFI =.977. This final model is presented in Figure 2. Player-avatar differences showed a significant negative relation to identification with the avatar for both types of games. However, this path was considerably stronger for non-competitive games (β = -.38, p <.001) than for competitive games (β = -.14, p <.001) as was predicted in hypothesis six. To test this difference for significance, the path between player-avatar-difference and identification was tested for invariance across both groups. This constrained model showed a significant increase of χ² compared to the freely estimated initial model (Δχ²(1) = 51.63, p <.001). Accordingly, the type of game (competitive vs. non-competitive) significantly moderates the relation between player-avatardifference and identification. Thus, the negative effect of the difference between player and

23 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 23 avatar on identification with the avatar is significantly stronger for non-competitive games than for competitive games, which supports hypothesis six Pls. insert Figure 2 about here To explore potential gender differences addressed in RQ1, the causal paths of the final statistical model (cf. Figure 2) were tested for invariance among male and female participants. Only one path-coefficient differed significantly among male and female players: The playeravatar difference had a significantly stronger influence on identification in competitive games for female (β = -.36, p <.001) than for male players (β = -.08, p <.01), Δχ²(1) = 22.03, p <.001. All other causal paths of the model were invariant between male and female participants. Discussion This study explored the impact of satisfaction with life and of computer game type on avatar creation as well as the relationship between player-avatar differences, identification with the avatar, and entertainment. We argued that users optimize the combination of their prerequisites and the attributes of the media stimuli they choose with the aim to maximize media enjoyment. The results supported our hypotheses, indicating that players preferred avatars resembling their personality especially in non-competitive games and the more they were satisfied with their own lives. In contrast, players tended to choose rather dissimilar avatars for competitive games and if they were less satisfied with their own lives. Higher similarity between player and avatar personality led to increased identification with the avatar. This relationship was stronger for non-competitive games compared to competitive games. In both competitive as well as in non-competitive games, identification was positively related to video game enjoyment.

24 Running head: AVATAR CREATION AND VIDEO GAME ENJOYMENT 24 The player's identification with the avatar seems to be crucial for entertainment experience, but is not necessarily tied to similarity. Players might identify with dissimilar avatars, depending on the type of game and on their satisfaction with life. In competitive games, avatars designed in accordance with the game requirements do not necessarily match the player's personality attributes, but they are promising in terms of mastery experiences and control. Mastering challenges and controlling the game successfully, in turn, are strong predictors of video game play and enjoyment (Grodal, 2000; Klimmt et al., 2007; J. Sherry, Lucas, Greenberg, & Lachlan, 2006). Avatars that resemble the participant's personality or other attributes provide more references to the player's current situation than dissimilar avatars. In cases of low satisfaction with life, this might remind the player of discrepancies between his or her actual and the ideal self (Higgins, 1987; Tangney, Niedenthal, Vowell Covert, & Hill Barlow, 1998). As suggested in the model of complex entertainment (Vorderer et al., 2004), our research underlines that user prerequisites, motives, and media prerequisites should be considered when trying to explain media enjoyment. In terms of user prerequisites, the users' competitiveness (Hartmann & Klimmt, 2006a) is also a variable that might influence the relationship between player-avatar similarity and entertainment. For players with high trait competitiveness, playing with an avatar designed to match the game requirements and thus allowing for mastery of the game should be more important than player-avatar similarity. In contrast, players with low trait competitiveness might prefer similarity to the avatar over matching the game requirements. In terms of game prerequisites, the structures of video and computer games as determinants of identification processes and entertainment deserve further attention. In our study, we investigated who people would like to be in games that are prototypically competitive or non-competitive in nature. But in many current video and computer games, it is up to the players to decide whether to play a game in a very competitive or non-competitive way, like in MMORPGs (Williams et al., 2006) or in action-