The Effects of Avatar-based Customization on Player Identification

Elisa Nieto | Download | HTML Embed
  • Sep 12, 2014
  • Views: 26
  • Page(s): 26
  • Size: 1.41 MB
  • Report



1 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 1 The Effects of Avatar-based Customization on Player Identification Selen Turkay, Harvard Initiative for Learning and Teaching, Harvard University, Cambridge, MA, USA Charles K. Kinzer, Teachers College, Columbia University, New York, NY, USA ABSTRACT Games allow players to perceive themselves in alternate ways in imagined worlds. Player identification is one of the outcomes of gameplay experiences in these worlds and has been shown to affect enjoyment and reduce self-discrepancy. Avatar-based customization has potential to impact player identification by shap- ing the relationship between the player and the character. This mixed method study aims to fill the gap in the identification literature by examining the effects of avatar-based customization on players identification with and empathy towards their characters in a massively multiplayer online game, Lord of the Rings Online (LotRO). Participants (N = 66) played LotRO either in customization or in no-customization groups for about ten hours in four sessions over two weeks in a controlled lab setting. Data were collected through interviews, surveys and observations. Results showed both time and avatar-based customization positively impacted players identification with their avatars. Self-Determination Theory is used to interpret results. Keywords: Avatars, Customization, Empathy, Identification, Mixed Method, MMOs INTRODUCTION identities without fear of disapproval by mem- bers in their real-life social circle (Turkle, 1995). Media researchers have been writing about There are multiple types of virtual worlds the ramifications of assuming technologically (i.e., social, gaming, educational) with various mediated identities since the inception of online affordances. Massively multiplayer online virtual worlds (see Turkle, 1994; 1995). These games (MMOs) have emerged to be one of virtual environments can provide anonymity the most popular gaming genres over the last and the freedom from the conventions of our decade, and have been studied from various everyday identities in areas such as gender, age perspectives (e.g., player demographics, ad- or social status. They also offer opportunities to diction, socialization, player motivations). This take on various personas, create or adopt new popularity is partly because MMOs affordances allow players to temporarily become a game character and adopt the salient characteristics of that character (Looy, Courtois & de Vocht, DOI: 10.4018/ijgcms.2014010101 2010). As detailed in the following section, Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

2 2 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 player identification with the avatar/character does not differentiate avatar customization from is central to how players experience the game. character customization. For instance, identification impacts player engagement with, and enjoyment of, the game (Klimmit, Hefner &Vorderer, 2009). Determin- BACKGROUND ing aspects of games, such as avatar customiza- Many scholars have discussed the important tion, that improve players identification with role of online identities in videogames and their characters would be of interest to game virtual worlds in development and exploration designers as well as educators who choose of real selves (Turkle, 1995; Thomas 2007; Gee games for their students. 2003). In these technologically mediated social Avatar customization is an understudied virtual environments, players establish digital factor when it comes to identification. Yet, it al- identities using a combination of modalities lows making each character different in MMOs including text, audio and visuals. Such identity by providing various combinations of attributes, construction is necessary to communicate with adornments/physical properties, skills, and traits others and with the virtual environment. Ava- (Dickey, 2007). This customization experience tars are the most commonly used expression may help the player to get into the mindset of of identity in virtual worlds (Hamilton, 2009). the character, increasing the likelihood of affect- Visual characteristics of an avatar, name, as ing his/her self-identity. This paper examines well as abilities of player characters, provide players identification with their characters over users with an expression of identity and an several gameplay sessions, varying the partici- opportunity for extended identity formation pants ability to customize their characters, and (Turkle, 1995). poses the following main research question and Self-representation is intentional within two exploratory research questions: the given choice structure of a virtual world. For example, initial character creation choices RQ1: Is there a relationship between engage- may indicate how the player expects the avatar ment with initial avatar customization and to function as a channel for her identity. Con- players identification with and empathy sidering part of identity formation is thinking toward their avatars? about what type person we want to be (Arnett, RQ2: Does players identification with and 2010, p.340), and virtual worlds can function empathy toward their characters change as identity construction environments (Bers, over time? 2001, p. 365). In these environments, users can explore and experiment with the dynamic na- The background section below clarifies ture of identity by interacting with and through the theories and present previous studies that their avatars (Kafai, Fields, & Cook, 2010). guided the formation of the research questions. Similarly, Turkle (1995) describes the creation Following the background, the methods section of an identity in virtual environments (MUDs) introduces the participants and describes the as fluid and multiple. She states that in virtual study. Before proceeding, however, it is worth worlds, people are able to build a self by cycling clarifying the difference between avatar and through many selves (Turkle, 1995, p.178). character. Avatar is defined as the embodi- Players manage their identities through playing ment of the user in virtual environments (e.g., different characters in MMOs (Taylor, 1999). In Ducheneaut, Wen, Yee, & Wadley, 2009). other words, people cycle through their possible Characters in games are fictional identities selves, defined as the cognitive manifestation within the narrative setting of the game. In of enduring goals, aspirations, motives, fears, this paper, avatar and character are used and threats (Markus & Nurius, 1986, p. 954). interchangeably because the research design Such identification facilitates identity develop- Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

3 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 3 ment, especially during adolescence (Erickson, the players motions and intentions has a great 1968). In a similar vein, McDonald and Kim impact on identification (Hamilton, 2009) and (2001) report that young videogame players a perceptual integration with the avatar, namely perceive no distance at all relative to their game the players awareness of her presence both in protagonists, and identify quite closely with her body and in the screen (Dove, 2002). them (p. 254). In addition to avatars abilities to represent Looy, Courtois and de Vocht (2010) call players intentions, previous studies determined for more attention to the concept of identifica- various player behaviors, features of avatars, and tion in game studies. Playing computer games characteristics of virtual worlds, that facilitate is enjoyable, partly because players can enter identification. Among these features are fond- imagined worlds beyond their real-life experi- ness for a character (Cohen, 1999), attractive- ences and perceive themselves in alternate ways. ness of avatars (Kim, Lee, & Kang, 2012), the Consequently, studying players identifica- capabilities of the character (Newman, 2002), tion with their avatars in a virtual environment point-of-view (Lim & Reeves, 2009), and can be crucial for understanding their gameplay physical resemblance of avatars to their users experiences. Cohen (2001) defines identifica- in body shape, race, age, and facial features tion with media characters as an imaginative (Maccoby & Wilson, 1957; Williams, 2011). experience in which a person surrenders con- Players perceived similarity to their characters sciousness of his or her own identity and experi- (or similarity identification) has been called the ences the world through someone elses point of mirror hypothesis (Chandler & Griffith, 2004). view (p. 248). Adapting this definition to video The mirror hypothesis refers to the theory that games, some researchers state that identification viewers tend to relate favorably to on-screen allows for experimentation with ones identity characters who are either like themselves (the by temporarily adopting aspects of the identity mirror), or ones who represent someone the of the target videogame character (e.g., a famous viewer would like to be (the magic mirror). hero, a historical figure, a sportsman) (Klimmt, The magic mirror relates to another type of Heefter, &Vorderer, 2009; Looy et al., 2010). identification: wishful identification. In wishful Identification, therefore, is conceptualized as identification, the observer desires to emulate a temporary shift in players self-perception the character, either in general terms as a role (Klimmt, Hefner, Vorderer, Roth, & Blake, model for future action or identity development, 2010). However, player identification with or in specific terms which extend responses characters is complicated because of the multi- beyond the viewing situation or by imitating a plicity of roles (e.g., subject, audience, director, particular behavior (Hoffner & Buchanan, 2005; user etc.) that a player takes during gameplay Hoffner & Cantor, 1991; Von Feilitzen & Linn, (Flannagan, 1999). For example, when someone 1975). Wishful identification provides a glimpse plays a videogame with an avatar, she is both of what if, and these glimpses are powerful the avatar/player-character and the player at predictors of future behavior (Cohen, 2001). the same time. When the player exerts agency Identification assumes both emotional over the avatar to interact with objects, events and cognitive connections between the player and other players, this is mediated both by and her avatar (Cohen, 2001). In a study on the player characters abilities and players human brain process about identification with abilities, and those have consequences to the game characters/avatars, Ganesh, van Schie, avatar within the designer created game world de Lange, Thompson, and Wigboldus (2011) (Murphy, 2004). In MMOs, players do not found that the extent of players feelings and felt observe autonomous social entities perform- absorption greatly impact identification. This ing on screen, instead they make characters study also showed that the intense emotional perform through character-generated actions, involvement with an avatar resembles the level also called emotes. An avatars representation of of intimacy one experiences when interacting Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

4 4 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 with a close other (Ganesh et al., 2011). How- playing the game reported more identification. ever, there are differing perspectives on the role They also found a positive relationship between of empathy in the mechanism of identification. identification with game characters and female For instance, Cohen (2001) views empathy as players gaming motivation. being a part of identification, whereas Hefner, The degree of control players have over Klimmit and Vorderer (2007) argue that the their avatars (e.g., characters movement, choice identification process between the player and of avatar appearance) may affect their level of the avatar will be less empathy driven than those identification with their avatars. Direct control between the reader/viewer and protagonist be- over their character can imbue players with a cause empathy is typically an emotion towards sense of agency and may also increase their an external entity. More recently, Looy et al.s positive affect in the game (Hefner, Klimmt & (2010) survey study found that player identifica- Vorderer, 2007). For instance, Schmierbach, tion predicts empathy. However, there is a lack Limperos, & Woolley (2012) showed that when of literature on the possible role of empathy in participants customized their racecars in a rac- the mechanism of player identification with ing game their enjoyment increased. Although videogame characters. players did not customize their avatars in this In addition to its role in identity develop- study, they played the game by controlling ment, researchers have discovered various psy- their racecar. Therefore, we may expect that chological and behavioral effects of player iden- customization of racecars in a race game and tification. Studies have found that identification customization of avatars in a role playing game with player characters reduces self-discrepancy may have similar psychological impact on (e.g., Bessiere, Seay, & Kiesler, 2007) and in- player experiences. In a similar vein, Ganesh fluences players self-efficacy and trust within et al.s (2011) neuroimaging study revealed their virtual communities (Kim, Lee, & Kang, that avatar related self-identification is related 2012). In an online survey study, Ducheneaut, to the experience of agency and control over Wen, Yee and Wadley (2009) found that visitors the observed body. These results imply a rela- of online virtual worlds who perceive a smaller tionship between control over the avatar and psychological difference between their avatar self-identification. and themselves are generally more satisfied Despite the results and implications of with their avatars and spend more time online. the studies noted above, Shaw (2011) points Identification also positively affects play- to a lack of empirical research on the extent to ers intention to purchase game items to increase which control over the avatar strengthens the their competitiveness and improve avatars relationship between the avatar and the player, appearance (Park & Lee, 2011). Through such especially in longer experimental studies. The mechanisms, increased identification may majority of the literature written on such a re- enhance player engagement, motivation, and lationship has been performed through either enjoyment. For instance, Klimmt et al. (2010) a theoretical approach (e.g., Murphy, 2004; conducted two experimental studies with male Hamilton, 2009) or surveys/interviews that players and showed that enacting a character or specifically address only the questions around role in a military themed first-person shooter identification (e.g., Shaw, 2011; Looy, Courtois game affected playersidentity state and increase & de Vocht, 2010; Reijmersdal et al., 2013). their game enjoyment. Identification is not This study aims to contribute to the literature by only important for male players in MMOs or in examining the effect of player control, specifi- fighting games, but is also important for female cally avatar-based customization, on players players. Reijmersdal, Jansz, Peters, and Noorts identification with their avatars. (2013) survey study with female players of a pink game, goSupermodel (WatAgame ApS., 2013), showed that girls who spent more time Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

5 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 5 AVATAR-BASED stories about their avatars, and have a stronger CUSTOMIZATION interest in customizing the appearance of their character (Looy et al., 2010). Many virtual worlds offer users ways to cus- There are different types of customization. tomize their experiences, either through built-in Turkay and Adinolf (2010, p. 1841) sug- options or the ability to create or obtain add-on gested that customization is grouped into three software modules as seen in games such as World broad categories in MMOs: of Warcraft (WoW) (Blizzard, 2004) and Star Wars: The Old Republic (SWTOR) (Bioware, Type I: Customization that affects game me- 2011). Avatar customization is a main form of chanics and dynamics directly and therefore customization in many MMOs, and several has a direct effect on individual player studies have found that avatar customization gameplay. Customizing talent trees in WoW and playing with customized avatars impact is an example of this type of customization; gaming experiences (e.g., Trepte & Reinecke, Type II: Customization that does not affect 2010; Turkay & Adinolf, 2010; Turkay, 2012), game mechanics and dynamics. Avatar learning in games (Okita, Turkay, Kim, Murai, appearance customization is an example 2013), subsequent behaviors after gameplay of this type of customization. Although it (e.g. Dolgov, Graves, Nearents, Schwark, & does not directly affect gameplay, it may Brooks Volkman, 2014), and increase play- affect a players enjoyment of the game; ers emotional attachment to their characters Type III: Customization that does not affect (Waggoner, 2009; Shaw, 2011). A subset of game mechanics and dynamics directly these studies made connections between avatar but may affect player performance, such customization and players identification with as interface customization, may have an their characters. effect on players gameplay experience. For instance, studies of aggression in video- games (Fischer, Kastenmuller, & Greitemeyer, Players use different strategies to create 2010; Holligdale & Greitemeyer, 2013) found and customize avatars depending on their evidence that avatar customization may am- goals, game developers goals and affordances plify the psychological effects of video games of virtual worlds. For example, players may through increased identification with ones customize their avatars appearance (cosmetic character. In their study with 30 children who customization) to reflect their aesthetic views played one of three advergames with differing or dress-up (Fron, Fullerton, Ford Morie, & levels of avatar customization, Bailey, Wise Pearce, 2007; Kafai et al., 2010). Players may and Bolls (2009) found that customization of also customize their characters with a functional game avatars can affect both subjective feelings goal in mind. In Kafai et al.s (2010) study, one of presence and psycho-physiological indica- type of functional customization for Whyville tors of emotion during game-play, which may (1999) users was to disguise within the com- make players experiences more enjoyable. munity (e.g. Gender swapping to hide ones Bailey et al.s (2009) study was one of the first actual gender). In MMOs, players may also to examine avatar customization with physi- choose a certain character class to fit a desired ological measures. However, the games they role within the game such as being able to heal used were not multiplayer, and the participants other players. played the games for a very short time, making It is likely that functional and cosmetic cus- generalization difficult. There is also evidence tomization may result in differing levels of iden- that identification facilitates customization tification with characters. For instance, when behavior. A survey study found that WoW play- players customize their character appearance ers who strongly identify with their characters similar to their real life physical appearance, more often like to role-play, create background their perceived similarity with their characters Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

6 6 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 may increase, whereas if they customize their METHOD characters to be strong and invincible their wish- ful identification may increase more than other Participants and Design identification aspects. However, in this study, cosmetic and functional customization (via Volunteers were invited to participate in this choosing rewards, and character race) were not study, which took place at a medium-sized uni- manipulated separately. Instead, players either versity, using fliers that outlined the study and customized both avatar appearance and avatar promised 50 US dollars to people who would be abilities or did not customize at all. Therefore, selected. Respondents were screened to ensure the rest of this paper refers to the combination their availability for the duration of the study, of cosmetic and functional avatar customization and also to ensure they were not expert game as avatar-based customization, and explores players. Expertise was measured in prescreening the impact of avatar-based customization on with multiple-choice questions about gaming players identification with their characters. habits, number of MMOs they played, and how long they played those MMOs. Seventy-five people were selected after screening. After DURATION OF THE STUDY the study began, one participant experienced motion sickness, and eight had scheduling Previous empirical studies (e.g., Schneider, conflicts. Thus, sixty-six participants (32 males, Lang, Shin, & Bradley, 2004; Hitchens, 34 female) completed the study. Drachen, & Richards, 2012) relied on a rela- Based on the screening questionnaire, the tively short playing time (between 8 minutes majority of the participants were not expert and 45 minutes) to draw conclusion regarding MMO game players. None of the participants players identification with their characters. were current MMO players, none said they had However, MMOs are long-term games, and played an MMO in the last year, and only three gameplay experiences may change over time as had played an MMO for more than a month. players gain expertise, form relationships within Participants familiarity with Lord of the Rings the game community, and develop their goals was also measured by asking about their experi- (Schultheiss, 2007). Thus, a reliable study of ence in, watching related movies, playing card, player experiences in these games requires play board and video games, and reading Tolkiens over a significant period of time, and certainly books. None of the players had played Lord for more than one experimental session. In a of the Rings Online (LotRO) (Warner Bros., similar vein, Klimmt et al. (2010) suggested 2013). The participants average age was 25.63, that game-character identification may emerge which is very close to the average age of MMO from significant playing time and Reijmersdal, players (M = 26.6) as reported in a previous, Jansz, Peters, and Noort (2013) stated that this large scale study (Yee, 2006). possibility should be further investigated. Based The study employed a between-subjects on data about players gameplay time, the aver- design with 33 participants in the customiza- age time of play per character in one week is tion group (CG; 17 females, 16 males) and 10.2 hours (Ducheneaut, Yee, Nickell, & Moore, 33 participants in the no customization group 2006), and a regular player plays about 100-150 (NCG; 17 females, 16 males). Participants were minutes per game session (Mahmassani, Chen, assigned to one of the two groups by gender. In Huang, Williams & Contractor, 2010). With this the CG, participants were given various choices in mind, this studys procedure involved about in the game, such as the opportunity to choose 10 hours of gameplay for each subject, divided their game characters specialties, skills, gender, into four sessions of 2 to 2.5 hours per session and appearance, as well as in-game rewards over two weeks. after they completed quests. In the NCG, the participants were assigned to well-constructed Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

7 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 7 Figure 1. LotRO characters: A CG character (left), a NCG character (right) pre-designed avatars with efficient character Three different game accounts were gener- skills and quest rewards were chosen for them ated for the study. At the time of data collection, that would maximize their character abilities. In LotRO was housed on 19 different servers; the NCG, avatar gender and participant gender each server appeared as different instances were matched (See Figure 1 for female avatar of the same world. Players can play in one of examples). these LotRO worlds, and some servers are more A preliminary analysis showed no signifi- populated than others. Each account allows five cant differences between groups in their age (t characters per server. Using multiple accounts = 0.72, p = .477), Lord of the Rings familiarity made it possible for participants to play in popu- (t = 0.94, p = .348) or MMO experiences (t = lated servers. This maximized the possibility of 1.32, p = .195). social interaction within the game. Participants used gaming-optimized PCs for the study, and wore a headset while playing. MATERIALS AND APPARATUS Identification and empathy were assessed with a 21-item 5-point Likert-type scale, de- Stimulus veloped and tested by Looy et al., 2010. Based Lord of the Rings Online (LotRO) was used for on previous literature, three main dimensions the study. LotRO is a fantasy type MMO based were identified: Wishful Identification (6 items), on the books by J.R.R. Tolkien. In searching e.g., If I could become like my character, I for an appropriate stimulus, three factors were would; Perceived Similarity (6 items), e.g. My taken into consideration: 1) Availability of character is like me in many ways; Embodied avatar-based customization (Turkay & Adinolf, Presence (5 items), e.g., I feel like I am inside 2010). For example, being able to customize my character when playing. Similar to Looy the character appearance and character skills; 2) et al., (2010) study, this study also examined Usability and playability; and 3) The technical players empathy towards their characters with requirements of the game. 4 items, e.g., I am upset when my character dies. Empathy items developed by Looy et Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

8 8 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 al. (2010) were based on Davis (1983) and in an adjoining cubicle, where a second monitor, Cohens (2001) conceptualization of empathy keyboard and mouse were placed. These were that involves emotional reactivity to others. Both connected to and mirrored the participants authors agree upon empathy involving emotion. computer, allowing the researcher to directly Cronbachs for the three dimensions of the observe gameplay, and manipulate the NCGs identification scale per session were measured choices. Participants had no direct line of sight in the current study and found to be satisfactory to the researcher during the session. The setting (ranging from 0.842 to 0.954). remained the same throughout the study. Engagement with avatar customization was assessed with a 5-point Likert scale survey, Procedure which was developed and tested by OBrian & Toms (2010). Cronbachs for the adapted As noted earlier, potential participants were pro- scale was 0.813. In addition, participants rated vided with an online survey after they showed the importance of different avatar customization interest in the study. This survey collected de- parts (e.g., hair style, weight, height) by using mographic data (e.g., gender, age, occupation) a 5-point Likert scale. As explained above, and gaming experiences. Participants who were this study did not manipulate different types of invited to participate in the study were provided customization. In order to investigate how the with an informed consent document upon first outcomes of two main avatar-based customiza- entering the laboratory for the experiment. tion (appearance and skills) may correlate with After each participant read and signed the different sub-parts of the identification scale, informed consent document, they were placed in participants were asked to rate two 5-point front of the gaming computer, and were briefed Likert scale items (i.e., How my character on the studys procedure. They were told that looks is important, what my character can they were going to finish the game tutorial in do is important) after the first and the last the first gameplay session, lasting about 1.5 to play session. 2 hours. Then, the CG participants created their While survey questions asked about iden- LotRO game characters. There was no time tification, empathy, and engagement directly, limit for character creation. Upon completing semi-structured interviews were conducted the character creation process, CG participants to allow participants to tell their gameplay filled out a survey to measure their engagement experience without prompting them on the with character customization. Using a 5-point topic of identification. These interviews were Likert scale survey, participants also rated the conducted after the first and the last session importance and the degree of similarity of the with all participants. A subset of participants customized avatar to themselves in 12 different was interviewed after the second (n = 27) and elements of avatar customization (e.g., height, third (n = 18) sessions as well. This resulted in weight, hair color). The study procedure was a total of 12 people who were interviewed after the same for the NCG, except that they did not every session. General, open-ended questions create their avatars and did not answer subse- were asked (e.g., Tell me about your experience quent avatar customization related questions. this session.) to determine the extent of player NCG participant were assigned to well formed, identification with their characters. but pre-generated characters. Throughout the study, participants in the Setting CG were allowed to make choices to customize their character skills and equipment, and they The gaming computer in a research lab used saw their choices reflected on their character during this study was separated from the main by equipping of new gear. The researcher made area with screens and resided in its own cubicle choices for the NCG on avatar-based functional to avoid distraction. The researcher had a table customizations through mirrored controls, as Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

9 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 9 described above. NCG participants would also of the players [P35] talked about her engage- not be able to see the changes reflected on their ment with avatar customization as, I had no avatars when they equipped new gear. This was expectation and interest at the beginning of accomplished by a function called Cosmetic the task. However, I saw getting drawn into Outfit in LotRO. started to care a lot. Data Analysis Importance and Identification Independent samples t-tests were conducted The most important avatar customization aspect to test differences between CG and NCG per for participants was hairstyle (M = 4.39, SD session. RM-MANOVAs were used to measure = 1.03) followed by hair color (M = 3.85, SD the possible change in players identification = 1.15) and eye color (M = 3.70, SD = 1.47). with, and their empathy toward, their characters When all the participants results were exam- over four sessions. Pearson correlations were ined, avatars height (M = 2.60, SD = 1.55) was conducted to test the direction and strength of rated as the least important avatar customiza- a relationship between engagement with initial tion element. On average, participants highly avatar customization and CG players identifi- cared about how their character looked (M = cation with their characters after each session. 4.15, SD = 1.23). In fact, there was a statisti- Semi-structured interviews were analyzed cally significant positive correlation between thematically. Also, a second set of analysis importance score and engagement with avatar was conducted based on the participants use customization (r = 0.491, p = .002). of first or third person pronouns in discussing Significant correlations were found the player characters in the games they played. between the importance score and Wishful This method was used previously in Hitchens et Identification after each of the four sessions, al. (2012)s study of identification with charac- between importance and Perceived Similarity ters. They asked participants to discuss various after all but the first session, between impor- events and actions in the games they played and tance and Embodied Presence after the last two analyzed the interview transcript based on their sessions (See Table 1 for Pearson correlation pronoun use. For more explanation about the coefficients). No significant relationship was analysis see Hitchens et al. (2012). found between players reported empathy and importance of avatar customization aspects. In order to examine the possible impact RESULTS of different types of customization on iden- tification, another set of correlations was Avatar Customization/ conducted between the three dimensions of Character Creation (Only CG) identification and on items How my charac- The average time spent on character creation ter looks is important (outcome of cosmetic was 12:23.018 (twelve minutes 23 seconds customization) and What my character can do and 18 milliseconds). On average, participants is important(outcome of functional customiza- spent 06:43.574 on race and class selection tion). Results indicate that character appearance and 05:39.442 on avatar appearance and name is more strongly related to identification than selection. character abilities (See Table 2). In general, participants had a positive at- Engagement with Avatar titude toward customizing their avatars, and Customization and Identification reported a moderate level of engagement with customization process (M = 3.81, SD = 0.62), Results show that there is no statistically and felt in control of their customization experi- significant correlation between engagement ence (M = 4.00, SD = 1.03). For example, one with avatar customization and subscales of Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

10 10 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 Table 1. Pearson correlations between importance of avatar customization and subscales of identification Session 1 Session 2 Session 3 Session 4 Perceived Similarity 0.270 0.391* 0.452** 0.645*** Embodied Presence 0.128 0.158 0.426* 0.463** Wishful Identification 0.484** 0.454** 0.441* 0.591*** Empathy 0.070 0.094 0.157 0.230 p < .05; **p < .01; ***p < .001 Table 2. Pearson correlations between outcomes of two main avatar-based customization (cos- metic and functional) and subscales of identification Session 1 Session 4 Cosmetic Functional Cosmetic Functional Perceived Similarity 0.359* 0.289 0.663* -0.007 Embodied Presence 0.311 0.262 0.358* 0.105 Wishful Identification 0.448** 0.241 0.569** 0.078 Empathy 0.331 0.332 0.367* 0.031 p < .05; **p < .01; ***p < .001 identification after the first session. However, Effect of Customization engagement was significantly correlated with on Identification all of the subscales of identification after the fourth session (See Table 3). Independent samples t-tests revealed statisti- These results imply a possible impact of cally significant differences between CG and time on player identification with their charac- NCG (See Table 4) for all three dimensions ters, which is examined below. of the identification scale. After each session, participants statistically significantly differed in Perceived Similarity and Wishful Identifica- tion, in favor of CG. CG reported a significantly higher sense of Embodied Presence than NCG Table 3. Pearson correlations between engagement with avatar customization and subscales of identification Session 1 Session 2 Session 3 Session 4 Perceived Similarity 0.007 0.258 0.350* 0.509** Embodied Presence -0.085 0.145 0.425* 0.536** Wishful Identification 0.267 0.487** 0.480** 0.513** Empathy -0.009 0.112 0.330 0.417* *p < .05; **p < .01; ***p < .001 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

11 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 11 Table 4. Statistics for identification subsections for each session Levenes Independent Samples (t-Test) CG NCG F p t p 2 M SD M SD Perceived Similarity 1 3.52 .065 -3.20 .002 .14 2.56 1.08 1.81 0.83 Perceived Similarity 2 1.46 .231 -3.54 .001 .17 2.53 0.98 1.76 0.77 Perceived Similarity 3 3.68 .060 -3.39 .001 .15 2.65 0.98 1.92 0.76 Perceived Similarity 4 10.39 .002 -4.07 .000 .21 2.79 1.09 1.85 0.75 Wishful Identification 1 2.57 .114 -2.15 .035 .07 2.34 0.93 1.89 0.79 Wishful Identification 2 1.92 .170 -3.23 .002 .14 2.37 0.91 1.71 0.73 Wishful Identification 3 2.65 .109 -3.03 .004 .13 2.52 0.98 1.85 0.81 Wishful Identification 4 2.76 .101 -3.21 .002 .14 2.61 1.14 1.80 0.90 Embodied Presence 1 0.02 .892 -1.94 .056 .06 2.51 1.06 2.02 0.98 Embodied Presence 2 0.19 .663 -2.05 .044 .06 2.53 1.02 2.04 0.92 Embodied Presence 3 3.01 .088 -2.60 .011 .10 2.84 1.07 2.21 0.89 Embodied Presence 4 0.19 .661 -4.07 .000 .16 3.08 1.11 2.19 1.00 after the second, third and the fourth gameplay interaction between Group and Sessions, Pil- sessions. lais Trace = .123, F (9, 55) = 0.86, p = .569. Assuming the normality of the data, RM- The examination of the means shows why the MANOVA was conducted to test for a possible interactions may not be significant; the groups difference between the CG participants and differed significantly on dependent variables at NCG participants in the amount of change in the end of the first session. their ratings on the three factors of the Identi- Tests of the Between-Subjects effects table fication scale. Table 5 summarizes the results indicate that there is a significant main effect of of the multivariate tests. Group on Perceived Similarity, Wishful Iden- Since the Boxs M value of 176.23 is as- tification, and Embodied Presence (See Table sociated with a p < .001, Pillais Trace was 6). Tests of Within-Subjects Contrasts showed used for the multivariate tests. Statistically a statistically significant linear relationship significant multivariate effects were found between sessions and Embodied Presence, F for the main effects of Group, Pillais Trace (1, 63) = 10.81, p < .002, 2 partial = .15. = .20, F (3, 61) = 4.96, p = .004, 2 partial = .20 In order to estimate mean differences and for Sessions, Pillais Trace = .26, F (9, 55) between sessions for each dependent variable, = 2.13, p = .042, 2 partial = .26, but not for the repeated tests were conducted. Results show Table 5. Multivariate tests Effect Pillais F df Error df p Partial 2 Trace Between Intercept .895 173.73 3 61 .000 .90 Subject Group .196 4.96 3 61 .004 .20 Within Sessions .258 2.13 9 55 .042 .26 Subject Sessions x Group .123 .86 9 55 .569 .12 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

12 12 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 Table 6. Tests of between subjects effects of group on identification Source Measure SS df MS F p Partial 2 Group Perceived Similarity 10.55 1 10.55 15.11 .000 .19 Wishful Identification 6.73 1 6.73 9.59 .003 .13 Embodied Presence 6.40 1 6.40 8.13 .006 .11 Error Perceived Similarity 43.96 63 0.70 Wishful Identification 44.23 63 0.70 Embodied Presence 49.62 63 0.79 Table 7. Test of within-subjects contrasts for perceived similarity SS F p Partial 2 Session 1 vs. Session 2 0.10 0.28 .596 .00 Session 2 vs. Session 3 1.19 5.84 .019 .09 Session 3 vs. Session 4 0.17 0.50 .482 .01 that there is a statistically significant differ- and NCG for each game session on players ence between Sessions 2 and 3 for all of the Empathy ratings. Results show that CG players dependent variables (See Table 7, 8 and 9). An had more empathy towards their avatars than examination of Figure 2, 3 and 4 also helps us did NCG (See Table 10). to understand the change between sessions. A mixed ANOVA was conducted to as- sess whether there were group differences and Empathy differences between the average ratings of the four sessions. The following assumptions were A set of independent samples t-tests was con- tested: (a) independence of observations, (b) ducted to investigate differences between CG normality, and (c) sphericity. Independence of Table 8. Test of within-subjects contrasts for wishful identification SS F p Partial 2 Session 1 vs. Session 2 0.37 1.12 .294 .02 Session 2 vs. Session 3 1.23 5.79 .019 .08 Session 3 vs. Session 4 0.05 0.22 .644 .00 Table 9. Test of within-subjects contrasts for embodied presence SS F p Partial 2 Session 1 vs. Session 2 0.02 0.04 .847 .00 Session 2 vs. Session 3 3.49 8.44 .005 .12 Session 3 vs. Session 4 1.12 2.81 .099 .04 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

13 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 13 Figure 2. Estimated marginal means of perceived similarity over four sessions observations is met, but normality was not met empathy towards their characters. For CG par- for three sessions. The Shapiro-Wilk test showed ticipants, importance of avatar customization that we reject the hypothesis that the data come was significantly correlated with identification from a normal distribution. The tests were con- with their characters. Most notably, there was a ducted by assuming normality. Mauchlys Test significant linear relationship between partici- of Sphericity indicated that the assumptions of pants experienced Embodied Presence (how sphericity had been violated, X2(5) = 31.23, p < much players felt like they were the characters .001. Thus, the Greenhouse-Geisser correction embodied in the game) and sessions. There was was used to correct degrees of freedom. Results no interaction between groups and time mean- indicated a statistically significant main effect ing that, though CG started ahead, the change of sessions, F (2.31, 147.95) = 11.42, p < .001, over time was similar for both groups in their 2 partial = .15, and of groups, F (1, 64) = 8.91, p identification with their characters. CG par- = .004, 2 partial = .12. However, the interaction ticipants engagement with character creation between session and group was not statistically in the first session had a stronger relationship significant, F (2.31, 147.95) = 0.70, p = .512. with their identification with their characters A test of within-subjects contrasts showed a as the sessions proceeded. significant linear relationship between sessions The quantitative results indicate that since and the participants rating of their empathy the rate of increase is about the same for both toward their characters, F (1, 64) = 19.93, p < groups, there must be other factors, independent .001, 2 partial = .24 (See Figure 5). of the treatment, that impact identification. Be- In summary, quantitative analyses showed low are qualitative findings that aim to provide that both avatar-based customization and time perspective on this possibility. impacted participants identification with and Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

14 14 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 Figure 3. Estimated marginal means of wishful identification over four sessions Qualitative Findings on Compared to NCG, CGs conversation Identification and Empathy included more instances of perceived similarity with their characters. This similarity was not Both groups talked about their characters in the only about appearance, but also about charac- interviews when they were asked to recount their ters behavioral characteristics. For example, experiences. A comparison over time showed [P08] said, I think she is kind of like me that twice as many participants mentioned their sometimes I get lost and I go off the wrong characters during the last interview (n = 39) direction sometimes ... [play] too bold. than during the first interview (n = 20). Over Another participant, [P59], was accepting all four interviews, 90% of the CG mentioned their the quests in the game and when asked why she characters, compared to 50% of NCG players. did that, she replied My character is kind of CG Participants used the pronoun I (or my like me. She cannot say no. This association character) more than NCG did while discussing was facilitated by character creation in the first their feelings about or events related to their session and built up over time, and the majority game characters (72% vs. 35% of the time) of responses about perceived similarity came (e.g. I fought with a group of monsters). NCG from their interviews in the last two sessions. mostly used s/he or the character to refer For example, [P7] reflected on her character in to their avatars. However, both CG and NCG the last interview as I chose this character alternated between referring to their character So whether hes the kind of people I admire or in the first and third person while telling their the kind of people I think I am, there is some experiences. This shift may indicate an ongo- similar things that I have or I want to have ing process of forming a relationship with their connected with me. This quote also exemplifies characters. Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

15 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 15 Figure 4. Estimated marginal means of embodied presence over four sessions her choice of the character being a boundary I chose my guy because I wanted to be right between her ideal (other) and real self. there in the middle of the fight I didnt want Wishful identification is about players to be the person in the back healing people or desire to be like their LotRO characters and was doing long-distance shots I think that had posi- also exemplified by several participants. The tive effect I can sustain a lot more damage. I most common quotes in this category from CG enjoyed that part I felt like I was my character participants were about how they wanted to be Being able to just give in and swing the axe so represented by aesthetically pleasing avatars. that was rewarding (P53, Interview 4, CG, M) Some were more interested in their charac- ters skills or functionality. For example, [P53]s Although not as common as in CG, some quote exemplifies how he chose his character NCG players bonded with their avatars. For based on what he wanted to do in the game: Table 10. Means and standard deviations of the participants empathy scores Levenes Independent Samples t-Test CG NCG F p t p 2 M SD M SD Empathy 1 0.44 .510 -2.36 .022 .08 2.78 1.00 2.19 1.01 Empathy 2 0.54 .467 -2.21 .030 .07 2.92 1.18 2.34 0.94 Empathy 3 1.03 .315 -2.78 .007 .11 3.23 0.90 2.58 1.01 Empathy 4 1.24 .269 -3.17 .002 .14 3.41 1.03 2.58 1.10 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

16 16 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 Figure 5. Estimated marginal means of empathy over four sessions example, [P18] felt quite immersed in the game, Achievements and experiences in the game creating a strong connection with her alter ego, help players feel good or be proud of them- Aydal, over each session. In the last session, selves. For example, [P40] talked about how the game became more challenging and her she liked non-player characters compliments character got killed. This experience made a on her achievements ... I like that when they big impact on her; she became very anxious complemented my character... but Im pretty and her hands shook. This connection to her sure it is standardized its like oh you did avatar seemed to be extreme. well very good fighter. Below [P18] explains how her characters Social context was also an important factor increased competence might have contributed for players identification with their characters. to such connection: Examples of two different perspectives are from [P59] and [P70]. In the third session, Hmmm... I dont know I feel like it is the small [P59] formed a fellowship (a group) with stuff like when you trying to you are growing other players, which is a common practice in in the beginning levels and when you get better MMOs. She stated that I actually think when you reach another level you can do more that I am the character. But, before, I was not things and you start to feel more than what you so connected with my character. Yeah, doing are and be happy for yourself that kind of stuff the quest together [with other players] I was contribute to how I built the connection it is very self-conscious Quite different from all the small stuff made me like the character this experience was [P70]s stance when she more (P18, Interview 4, NCG, F) was asked and declined to join a fellowship consisting of higher level of players. She said if they think that my character is a novice Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

17 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 17 then Im also a novice player. I dont want to view. Third person view allows players see the be seen that way by others. I dont want to back of their avatar on the screen and have the reveal my identity [that I am a novice player]. avatar present on the screen at all times. The I just pretend that Im such a good player who first person view allows players to see the game is really skillful so they all want to be friends world from their avatars eyes without seeing the with me... Although she may identify with her avatar on the screen. Participants who tried first character, she wants to make a good impression person view when playing reported an elevated in the game. feeling of embodied presence. The games 3D Participants talked about both functional graphics and detailed real world modeling, and cosmetic character customization. Char- incorporating animations and environmental acter creation and avatar customization at the details, facilitated such feelings as well. beginning of the first session was mentioned For example, [P70] talks about her enjoy- mostly as cosmetic customization. Some partici- ment related to detailed graphics and how this pants talked about both. Here is a representative impacts her feeling of embodiment: quote from [P64] highlighting the process of character customization, It was fun to create I like the realistic characters and it is very de- my own character to put multiple characteris- tailed. I did not even imagine that I could laugh tics on it and it was also fun, as the game goes in the game. The gestures are so specific and along, to keep customizing it by adding more detailed makes me feel I get immersed in the weapons and all like changing colors and all. character. (P70, Interview 4, CG, F) Participants in NCG reported less attach- ment and less empathy toward their characters. In summary, a majority of the participants For example, [P3] talked about how she got used formed some form of a relationship with their to her character dying in a battle The character characters, though this occurred more in the CG died once today I think Im getting used to than NCG. For some participants, the charac- watching my character dying, because I know ters were a representation of themselves; for shes going to come back. Similarly, [P9] talked others, the characters were nothing but mere about how, as time passed, she cared less about toys or vehicles. In all cases, this relation and the character but more about the story that the identification with characters was dynamic and character was embedded in. changed over time. Sometimes, players sense A similar comment came from [P23] em- of wishful identification with the character got phasizing his apathy: quite strong, like in the case of [P18]. In addition to character creation and avatar customization, It [survey] was asking whether I would feel sorry socialization, sense of accomplishment and the when the character dies. No. I know that its realistic game world increased participants a game characterIt is enjoyable to play that identification with their characters. Perceived character as it is but there is no relationship similarity was attributed to both physical ap- between me and the character. I could not find pearance of the characters and characters player anything in him to identify with myself. (P23, assigned personality. Embodied presence was Interview 4, NCG, M) facilitated by the game design and graphics as well as the games narrative. When it comes to embodied presence, participants talked about a few different experi- ences that made them feel like their character DISCUSSION was an extension of themselves. One of these Results showed that CG players identified with experiences related to the games first person their characters significantly more than NCG view. LotRO provides players with two different did. The psychological aspect of making choices view options: first person view and third person Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

18 18 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 might explain this result: because players chose (e.g. physical appearance, social status), and their avatars aspects they felt more associated Wishful Identification with their avatars as with the character by taking ownership of Within Subject variables did not significantly it. Another explanation might be related to change over time. However, treatment had a the distance between real self and ideal self. moderate to large main effect on the differences Customization might have allowed players to in these variables. create characters closer to their ideal self, which Qualitative findings showed that many of might have increased their identification with the CG players chose their characters so that their characters. Within the CG, players spent they possessed some aspect of themselves, such considerable time customizing their character in as a skill (e.g., playing an instrument) or physi- the first session and their reported engagement cal characteristic (e.g., hair, eyes, built). Such was strongly related to their identification in the matching along with the psychological aspect last session. This result implies that as players of the act of choosing might have increased CG spent more time in the game with their custom- players identification with their characters by ized avatars, their identification increases. creating a channel to relate with their characters. Correlational analysis between identifica- Trepte and Reineckes (2010) study supports tion and two main outcomes of avatar-based this finding of the strong relationship between customization (i.e., cosmetic and functional) player-avatar similarity and identification. indicates that cosmetic avatar customization Character creation allowed players to create was more strongly related to identification than their own goals for the game. In turn, they functional customization for the players in this started the game with a goal to accomplish study. Considering players low expertise with for their characters. This initial goal setting is MMOs, this result is expected. It may take more motivating for people as they engage in goal time for players to understand the goals of the related tasks (Latham & Locke, 1979). character as result of their choices and identify Throughout the study, CG reported signifi- with them, than to identify with the characters cantly higher Embodied Presence than NCG through aesthetic manipulations. did. Players Embodied Presence changed sig- Although NCG players, who were given a nificantly over time. Treatment and sessions did character, did not identify with those characters not interact. This means that CG still felt more as much as CG players did, some of the NCG present in the game than NCG, but the rate players showed identification with the char- of change in Embodied Presence was about the acters they were given. According to previous same for both groups. The significant change studies, a match in gender and ethnicity influ- in Embodied Presence of CG players might be ences identification (e.g., Lim, 2006). This may a result of increased player agency and social help us understand NCG players identification interaction through avatars (Taylor, 2002) as with their characters. As mentioned previously, they progressed in the game. As players built NCG players were assigned human avatars a closer relationship with their avatars through whose gender, hair color, skin color, and eye continuous customization, their sense of being color matched with the players. However, the embodied by the avatars might have increased. impact of character creation/avatar customiza- Another reason for the increase might be that tion was significantly more than mere gender as players got more familiar with the game match and appearance resemblance. and with their characters, it became easier to CG players felt their characters were more immerse themselves in the game through their similar to themselves and wished to be like their characters, and feel increased presence. characters more than their counterparts did. Players identification with and empathy Players Perceived Similarity (gauging how toward their characters increased over the much players perceived themselves and their four sessions as they built a profile for their characters as having common characteristics, avatars. This was evidenced both by qualitative Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

19 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 19 and quantitative results. Although the rate of by controlling their characters, these characters change was not significantly different for CG may become the other for them. Future stud- and NCG, customization corresponded to the ies should investigate ways to capture objective initial variance among the players identification and subjective data during gameplay as well as and empathy. These differences between CG via surveys after the gameplay is completed. and NCG were maintained over four sessions. Self Determination Theory (SDT, Deci & Player motivation or retention depends on Ryan, 1985) posits that autonomy, competence their emotional connection with the game and and relatedness are necessary for peoples well- how meaningful that connection is to the players. being, and was used previously to explain mo- For CG players, having opportunities to acquire tivational aspects of MMOs (Przybylski, Rigby unique, visually appealing items to customize & Ryan, 2010). As a meta-motivation theory, their experiences might have facilitated this SDT can also be used as a lens to understand emotional connection (Koehne, Bietz, & Red- findings of the current study. Autonomy satis- miles, 2013). The fact that NCG players could faction, mostly utilized as the feeling of having not see their choices being reflected on their control over an activity, is crucial in encouraging characters might have inhibited the construction people to come back to do the same activity. The of a relationship between the player and the choice-making involved in customization may character. For NCG, increased responsibility give people a sense of autonomy. In support of (e.g., goals given to them) through the game such an implication, Teng (2010) showed that narrative (Schneider et al., 2004) and interac- customization increases gamer loyalty. tion with their avatars might have resulted in In MMOs, players are introduced to more increased identification and empathy toward choices in the form of customization as they level their characters. up and gain expertise, but the most concentrated Klimmit et al. (2009) propose that identi- and impactful choice-making happens during fication is never an absolute process in which character creation. In line with this, results the players entire identity is replaced by the showed that participants spent considerable identity of the character, but rather a partial amount of time making decisions about their alteration in self-perception. This partial al- avatar appearance and character skills. The teration may allow players to both experience positive relationship between CG participants empathy towards their characters (the character engagement with character creation in the first as the other) and experience a merge with their session, and their increased identification with character (the character as the self). In a way, their characters as the sessions proceeded, there is an emotional task switching during implies the character creation process has the gameplay. long-term effects on players experience, such Empathy can also be a product of identifica- as identification. It is safe to put forward that tion. One of the primary effects of identification CG participants higher identification with their is empathy towards the subject of identification characters, compared to NCG participants, is (Nathanson, 2003; Looy et al., 2010). Perhaps, due to the sense of agency and autonomy they players feel empathy towards their characters felt as result of making various choices while after their identification process, or after the customizing their characters in the first ses- gameplay is over. The lack of differences in sion. The improved sense of agency facilitated the results between empathy and identification identification and empathy. can also be a consequence of the limitation of The across the board increase in players current measures used to investigate player identification with their characters over time identification and empathy. This study asked may be due to an increased sense of compe- participants to report their identification and tence as they accomplish goals in the game. empathy after their gameplay was over. When For example, [P18], who was proud of her players are not actively involved in the game characters accomplishments, identified with Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

20 20 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 her character closelythis temporarily reduced dynamic rather than static facilitates students her self-discrepancy (Higgins, 1987) and getting growth mindset (Kolb & Kolb, 2009). Virtual defeated shook her relationship with her char- worlds give users imagined worlds and tools to acter. Previous work with young adults found test various identities through active processes that strong characters were attractive, implying of design. An avatar is the main tool users have the importance of feeling successful, for wishful for identity exploration. Strengthening the rela- identification (Janz, 2005). Players need for tionship between the player and the avatar can relatedness might have encouraged them to facilitate identification through which students create similar characters to themselves, thus can form their identities (Weinreich & Saunder- fostering players perceived similarity. Another son 2003). This study showed that avatar-based form of relatedness is relating to other player customization facilitates players identification characters in the game. Playing with others and with their characters by increasing their sense having common missions might have fostered of autonomy and agency. Such processes can players embodied presence, thus increasing build students self-efficacy, resulting in higher overall identification with their characters. achievement motivation (Ames, 1992). Con- Results also showed significant differences sidering that focused decision making during between Sessions 2 and 3 for the subscales of character creation in the beginning resulted in identification, whereas there was no difference CG players higher identification with their between first two sessions and the last two ses- characters, educational game designers should sions. The reason for this might be that during consider giving players periodic chances to the first two sessions participants were still re-customize their characters appearance and learning the game and the controls. Hamilton skills in addition to main avatar customization (2009) suggests that through the interaction with at the beginning of game-play. This may allow and via the avatar (through embodied interac- students to re-consider their characters goals, tion), players become coupled to the avatar. In as well as increase their sense of agency. such coupling, mastering the game controls is Second, avatar-based customization may crucial, and it was after the second session that catalyze students learning. Mantovani and players got comfortable with game controls. In Castelnuovo (2003) state that identification their SDT-based motivational model of video might be an important factor in learning in games, Przybylski, Rigby and Ryan (2010) virtual environments, increasing emotional emphasized the importance of mastery of game impact and relevance, and Ganesh et al. (2011) controls for player engagement and motivation. showed that players remember avatar-related The findings of this study support this asser- events more than non-avatar related events. The tion. Mastery of controls might be the gateway results of the current study suggest that when to players identification with their characters. players customize their characters they will identify with their characters further and focus Implications for more on events related to their characters. When Educational Games they dont customize, they focus on the games narrative. We suggest that when teachers make While not tested in an educational game, the decisions among different types of educational results of this study imply that allowing learners games to use in their classrooms, they should to customize their avatars in educational virtual take into account both content and customiza- environments may have beneficial effects in tion. If students learn best from the narrative, the areas of motivation and other areas that perhaps a game with less or no customization correlate with learning. will help. If students learn best from the game This study shows that avatar-based cus- system, or social interaction, customization may tomization may increase students motivation. lend a helping hand. For instance, students may In educational settings, viewing identity as customize the appearance of their historical Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

21 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 21 protagonist in a history themed video game. This LIMITATIONS should help them to recall more of the actions and the speeches of the historical character in This study has multiple limitations. Its results the videogame via the mechanism of player may not be fully applicable to other types of identification (Cohen, 2001). virtual environments due to their structural Analysis of pronoun use (I v.s. she/he differences, for example in their narrative and or my character) showed that CG participants avatar customization processes. While MMOs used I more often to refer their characters when are narrative driven gaming worlds, social vir- they were telling their game experiences. This tual worlds such as Second Life usually do not perspective taking should increase students have a built-in narrative. Narrative is known to recall. For instance, in a study of first person impact player identification, (Schneider, Lang, vs. third person perspective taking, Lozano, Shin, & Bradley, 2004), and future research Hard, and Tversky (2007), showed participants should address the possible mediating role of a video clip of a person assembling a piece of narrative on the impact of customization on furniture and told them to verbally describe the players identification with avatars. They should process in either the first-person perspective ascertain under which conditions users identify or third-person perspective. Later, participants more with their avatars in social virtual worlds. who described the process in the first-person In a similar vein, implications for educational perspective made significantly fewer errors. We virtual environments should be viewed cau- may expect similar effects in video games where tiously because LotRO is not an educational students take characters perspective rather than MMO and, while we believe educational-game see their character as a separate entity. implications can be (cautiously) made, this study Last, avatar-based customization may in- did not test for learning. crease behavioral outcomes of serious games. Not all virtual worlds provide their users For instance, if we aim to encourage students with similar types of customization options; to exercise more or eat healthy through interac- some provide detailed customization (e.g., City tions within a game or virtual world, we should of Heroes[This should probably be in italics allow them to choose and customize their and have a citation.]) others dont. Studies have characters. Fox and Bailenson (2009) found shown that customization tools available to users that participants exercised more when they when designing avatars (Turkay, 2012) and the used a virtual representation of self rather than type of virtual environment may affect players a virtual representation of other. We may expect online identity in absolute means (Koehne et that increased identification via customization al., 2013). For example, in Vasalou and Joinson would reinforce the effect of avatars on driving (2009)s study, participants were asked to cre- behavioral outcomes via gameplay. However, ate 2D avatars to represent themselves in an future research should investigate more clearly online forum (Yahoo! Answers). In that virtual the impact of avatar-based customization on environment, users may create avatars similar student learning, motivation and behavior, to their real-selves because they may want to be and examine whether customization increases recognized as an answer giver. Realistic avatar player identification in various types of games appearance is also not possible in many fantasy (e.g., non-role playing games, games without MMOs since the avatars are not always humans, narrative, educational games vs. commercial although most are humanoids. LotRO allows games). players to create humanoid characters, but the choices are still limited for each customized part. This study did not manipulate different types of customization (see Turkay & Adinolf, 2010). When players create their characters, they dont only make choices about their avatars Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

22 22 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 appearance, but also about their characters Arnett, J. J. (2010). Adolescence and emerging adult- abilities (i.e., what they will do in the game). It hood: A cultural approach (4th ed.). Upper Saddle River, NJ: Pearson-Prentice Hall. is possible that these different type of customiza- tions result in different behavioral, psychologi- Bailey, R., Wise, K., & Bolls, P. (2009). How cal and motivational outcomes. Future studies avatar customizability affects childrens arousal should differentiate the impact of different types and subjective presence during junk food-sponsored online video games. Cyberpsychology & Behav- of customization on players identification with ior, 12(3), 277283. doi:10.1089/cpb.2008.0292 and empathy toward their characters. PMID:19445632 This study was conducted with novice MMO players. The results may not be fully Bers, M. (2001). Identity construction environments: Developing personal and moral values through the applicable to players of varying expertise. It design of a virtual city. Journal of the Learning Sci- is reasonable to think that experts may bring ences, 10(4), 365415. doi:10.1207/S15327809JL- dramatically different expectations to their S1004new_1 gameplay, their customization practices, and Bessiere, K., Seay, A. F., & Kiesler, S. (2007). The how these may impact their relationships with ideal elf: Identity exploration in World of Warcraft. their avatars. Cyberpsychology & Behavior, 10(4), 530537. Related to expertise, a longer study may doi:10.1089/cpb.2007.9994 PMID:17711361 find different results. Although this study in- Chandler, D., & Griffiths, M. (2004). Who is the fair- volved significant amounts of gameplay, it is est of them all? Gendered readings of Big Brother 2 likely that as players gain more expertise, the (UK). In E. Mathijs, & J. Jones (Eds.), Big Brother impact of certain type of customization may International: Format, Critics and Publics (pp. become more salient and others may become 4061). London, UK: Wallflower Press. weaker for increasing player identification with Cohen, J. (1999). Favorite characters of teen- and empathy toward their characters. age viewers of Israeli serials. Journal of Broad- casting & Electronic Media, 43(3), 327345. doi:10.1080/08838159909364495 CONCLUSION Cohen, J. (2001). Defining identification: A theoreti- cal look at the identification of audiences with media In conclusion, identification was a strong con- characters. Mass Communication & Society, 4(3), tributor to players positive or negative game 245264. doi:10.1207/S15327825MCS0403_01 experiences. Avatar-based customization played Davis, M. H. (1983). Measuring individual dif- an important role in players identification ferences in empathy: Evidence for a multidimen- with their characters by increasing their sense sional approach. Journal of Personality and Social of autonomy. Future studies are needed to dif- Psychology, 44(1), 113126. doi:10.1037/0022- ferentiate the effects of customizing character 3514.44.1.113 skills from customization of avatar appearance Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation as these relate to identification, as well as to and self-determination in human behavior. New York, examine how differences in given customiza- NY: Plenum Press. doi:10.1007/978-1-4899-2271-7 tion choices constrain or enhance identification Dickey, M. D. (2007). Game design and learning: A and identity exploration possibilities in multiple conjectual analysis of how massively mutliple online virtual environments. role-playing games (MMORPGs) foster intrinsic motivation. Educational Technology Research and Development, 55(3), 253273. doi:10.1007/s11423- 006-9004-7 REFERENCES Dolgov, I., Graves, W. J., Nearents, M. R., Schwark, J. Ames, C. (1992). Classrooms: Goals, structures, D., & Brooks Volkman, C. (2014). Effects of coopera- and student motivation. Journal of Educational tive gaming and avatar customization on subsequent Psychology, 84(3), 261271. doi:10.1037/0022- spontaneous helping behavior. Computers in Human 0663.84.3.261 Behavior, 33, 4955. doi:10.1016/j.chb.2013.12.028 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

23 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 23 Dove, T. (2002). The Space between: Telepresence, Hefner, D., Klimmt, C., & Vorderer, P. (2007). Iden- re-animation and the re-casting of the invisible. In tification with the player character as determinant of M. Reiser, & A. Zapp (Eds.), New screen media: video game enjoyment. In L. Ma, M. Rauterberg, & cinema/art/narrative. London: British Film Institute. R. Nakatsu (Eds.), Entertainment computingInter- national Conference of Entertainment Computing Ducheneaut, N., Wen, M. D., Yee, N., & Wadley, G. (pp. 3948). Berlin: Springer. (2009). Body and mind: A study of avatar person- alization in three virtual worlds. ACM Conference Higgins, E. T. (1987). Self-Discrepancy: A Theory on Human Factors in Computing Systems, 2009. Relating Self and Affect. Psychological Review, doi:10.1145/1518701.1518877 94(3), 319341. doi:10.1037/0033-295X.94.3.319 PMID:3615707 Ducheneaut, N., Yee, N., Nickell, E., & Moore, R. J. (2006). Alone together?: Exploring the social Hitchens, M., Drachen, A., & Richards, D. (2012). dynamics of massively multiplayer online games. An investigation of player to player character Proceedings of the SIGCHI Conference on Human identification via personal pronouns. Proceedings Factors in Computing CHI 06 (pp. 407-416). ACM, of The 8th Australasian Conference on Interactive New York, NY. doi:10.1145/1124772.1124834 Entertainment: Playing the System. Auckland, NZ, New Zealand. doi:10.1145/2336727.2336738 Erikson, E. (1968). Identity youth and crisis. New York: Norton. Hoffner, C., & Buchanan, M. (2005). Young adults wishful identification with television characters: The Fischer, P., Kastenmller, A., & Greitemeyer, T. role of perceived similarity and character attributes. (2010). Media violence and the self: The impact Media Psychology, 7(4), 325351. doi:10.1207/ of personalized gaming characters in aggressive S1532785XMEP0704_2 video games in aggressive behavior. Journal of Experimental Social Psychology, 46(1), 192195. Hoffner, C., & Cantor, J. (1991). Perceiving and doi:10.1016/j.jesp.2009.06.010 responding to mass media characters. In J. Bryant, & D. Zillmann (Eds.), Responding to the screen: Flanagan, M. (1999). Mobile identities, digital stars, Reception and reaction processes (pp. 63101). and post-cinematic selves. Wide Angle., 21(1), 7793. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. doi:10.1353/wan.1999.0002 Hollingdale, J., & Greitemeyer, T. (2013). The chang- Fox, J., & Bailenson, J. N. (2009). Virtual self- ing face of aggression: The effect of personalized modeling: The effects of vicarious reinforcement and avatars in a violent video game on levels of aggres- identification on exercise behaviors. Media Psychol- sive behavior. Journal of Applied Social Psychology, ogy, 12(1), 125. doi:10.1080/15213260802669474 43(9), 18621868. doi:10.1111/jasp.12148 Fron, J., Fullerton, T., Ford Morie, J., & Pearce, Jansz, J. (2005). The emotional appeal of violent video C. (2007). Playing dress-up: Costumes, role-play games for adolescent males. Communication Theory, and imagination. Philosophy of Computer Games 15(3), 219241. doi:10.1111/j.1468-2885.2005. Conference, Reggio Emilia, Italy tb00334.x Ganesh, S., van Schie, H. T., de Lange, F. P., Kafai, Y. B., Fields, D. A., & Cook, M. S. Thompson, E., & Wigboldus, D. H. (2011). How the (2010). Your second selves: Player-designed human brain goes virtual: Distinct cortical regions avatars. Games and Culture, 5(1), 2342. of the person-processing network are involved in doi:10.1177/1555412009351260 self-identification with virtual agents. Cereb Cor- tex., 22(7), 15771585. doi:10.1093/cercor/bhr227 Kim, C., Lee, S., & Kang, M. (2012). I became PMID:21917741 and attractive person in the virtual world: Users identification with virtual communities and avatars. Gee, J. P. (2003). What video games have to teach Computers in Human Behavior, 28(1), 16631669. us about learning and literacy? New York, NY: doi:10.1016/j.chb.2012.04.004 Palgrave Macmillan. Klimmt, C., Hefner, D., & Vorderer, P. (2009). The GoSupermodel [PC Game] (2013). watAgame. video game experience as true identification: Copenhagen, Denmark. A theory of enjoyable alterations of players self- Hamilton, J. G. (2009). Identifying with an Avatar: perception. Communication Theory, 19(4), 351373. A multidisciplinary perspective. Proceedings of doi:10.1111/j.1468-2885.2009.01347.x Cumulus Conference09. Melbourne, Australia. Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

24 24 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 Klimmt, C., Hefner, D., Vorderer, P., Roth, C., & Mantovani, F., & Castelnuovo, G. (2003). Sense of Blake, C. (2010). Identification with video game presence in virtual training: enhancing skills acqui- characters as automatic shift of self-perceptions. sition and transfer of knowledge through learning Media Psychology, 13(4), 323338. doi:10.1080/1 experience in virtual environments. In G. Riva, F. 5213269.2010.524911 Davide, & W. A. IJsselsteijn (Eds.), Being there: Concepts, effects and measurement of user presence Koehne, B., Bietz, M. J., & Redmiles, D. (2013). in synthetic environments (pp. 167181). Amsterdam, Identity Design in Virtual Worlds. Paper presented The Netherlands: Ios Press. at the 4th International Symposium, IS-EUD 2013, Copenhagen, Denmark. Markus, H., & Nurius, P. (1986). Possible selves. The American Psychologist, 41(9), 954969. Kolb, A., & Kolb, D. (2009). The learning way: doi:10.1037/0003-066X.41.9.954 Meta-cognitive aspects of experiential learn- ing. Simulation & Gaming, 40(3), 297327. McDonald, D., & Kim, H. (2001). When I die, I feel doi:10.1177/1046878108325713 small: Electronic game characters and the social self. Journal of Broadcasting & Electronic Media, 45(2), Latham, G. P., & Locke, E. A. (1979). Goal setting: 241258. doi:10.1207/s15506878jobem4502_3 A motivational technique that works. Organiza- tional Dynamics, 8(2), 6880. doi:10.1016/0090- Murphy, S. D. (2004). Live in your world, 2616(79)90032-9 play in ours: The spaces of video game iden- tity. Journal of Visual Culture, 3(2), 223238. Lim, S. (2006). The effect of avatar choice and visual doi:10.1177/1470412904044801 POV on game play experience (Unpublished doctoral dissertation). Stanford University, Stanford, CA. Newman, J. (2002). The myth of the ergodic video- game. The International Journal of Computer Game Lim, S., & Reeves, B. (2009). Being in the Research., 2(1), 18. game: Effects of avatar choice and point of view on psychophysiological responses dur- OBrien, H. L., & Toms, E. G. (2010). The devel- ing play. Media Psychology, 12(4), 348370. opment and evaluation of a survey to measure user doi:10.1080/15213260903287242 engagement. Journal of the American Society for Information Science and Technology, 61(1), 5069. Looy, J. V., Courtois, C., & Vocht, M. D. (2010). doi:10.1002/asi.21229 Player identification in online games: Validation of a scale for measuring identification in mmorpgs. Paper Okita, S. Y., Turkay, S., Kim, M., & Murai, Y. (2013). presented at the Proceedings of the 3rd International When observation beats doing: Learning by teaching Conference on Fun and Games, Leuven, Belgium. with virtual peers and the effects of technological design choices on learning. Computers & Education, Lord of the Rings Online [PC Game] (2013) Warner 63, 176196. doi:10.1016/j.compedu.2012.12.005 Bros. Entertainment Inc. Park, B.-W., & Lee, K. C. (2011). Exploring the Lozano, S. C., Hard, B. M., & Tversky, B. (2007). value of purchasing online game items. Computers in Putting action in perspective. Cognition, 103(3), Human Behavior, 27(6), 21782185. doi:10.1016/j. 480490. doi:10.1016/j.cognition.2006.04.010 chb.2011.06.013 PMID:16765339 Przybylski, A. K., Rigby, C. S., & Ryan, R. M. (2010). Maccoby, E. E., & Wilson, W. C. (1957). Identifica- A motivational model of video game engagement. tion and observational learning from films. Journal Review of General Psychology, 14(2), 154166. of Abnormal and Social Psychology, 55(1), 7687. doi:10.1037/a0019440 doi:10.1037/h0043015 PMID:13462664 Reijmersdal, E. A. V., Jansz, J., Peters, O., & van Mahmassani, H. S., Chen, R. B., Huang, Y., Williams, Noort, G. (2013). Why girls go pink: Game char- D., & Contractor, N. (2010). Time to play? Activ- acter identification and game-players motivations. ity engagement in multiplayer online role-playing Computers in Human Behavior, 29(6), 26402649. games. Transportation Research Record, 2157(2), doi:10.1016/j.chb.2013.06.046 129137. doi:10.3141/2157-16 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

25 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 25 Schmierbach, M., Limperos, A. M., & Woolley, Turkay, S. (2012). User experiences with avatar J. K. (2012). Feeling the Need for (Personalized) customization in Second Life and Lord of the Rings Speed: How Natural Controls and Customization Online. Proceedings on Teachers College Educa- Contribute to Enjoyment of a Racing Game Through tional Technology Conference. Enhanced Immersion. Cyberpsychology, Behavior, and Social Networking, 15(7), 364369. doi:10.1089/ Turkay, S., & Adinolf, S. (2010). Free to be me: A cyber.2012.0025 PMID:22687145 survey study on customization with World of War- craft and City of Heroes/Villains players. Procedia: Schneider, E. F., Lang, A., Shin, M., & Bradley, S. Social and Behavioral Sciences, 2(2), 18401845. D. (2004). Death with a story: How story impacts doi:10.1016/j.sbspro.2010.03.995 emotional, motivational, and physiological responses to first-person shooter video games. Human Com- Turkle, S. (1994). Constructions and reconstruc- munication Research, 30(3), 361375. tions of self in virtual reality. Playing in MUDs. Mind, Culture, and Activity, 1(3), 158167. Schultheiss, D. (2007). Long-term motivations to doi:10.1080/10749039409524667 play MMOGs: A longitudinal study on motivations, experience and behavior. In A. Baba (Ed.), DiGRA Turkle, S. (1995). Life on the screen: Identity in 2007-Situated Play (Proceedings of Digital Games the age of Internet. New York: Simon & Schuster. Research Association International Conference Vasalou, A., & Joinson, A. (2009). Me, myself and I: 2007) (pp. 344348). Tokyo, Japan. The role of interactional context on self-presentation Shaw, A. (2011). He could be a bunny rabbit for through avatars. Computers in Human Behavior, all I care: Exploring identification in digital games. 25(2), 510520. doi:10.1016/j.chb.2008.11.007 Proceedings of DiGRA 2011: Think Design Play: The von Feilitzen, C., & Linne, O. (1975). The effects fifth international conference of the Digital Research of television on children and adolescents identifying Association (DIGRA). with television characters. Journal of Communica- Star Wars. (2011). The Old Republic [PC Game]. tion, 4(4), 3155. Bioware. Waggoner, Z. (2009). My avatar, my self: identity Taylor, T. L. (2002). Living digitally: Embodiment in video role-playing games. Jefferson, N.C.: Mc- in virtual worlds. In R. Schroeder (Ed.), The Social Farland. Life of Avatars: Presence and Interaction in Shared Weinreich, P., & Saunderson, W. (Eds.). (2003). Virtual Environments. London: Springer-Verlag. Analyzing identity: Cross-cultural, societal and doi:10.1007/978-1-4471-0277-9_3 clinical contexts. London: Routledge. Teng, C.-I. (2010). Customization, immersion sat- Whyville [PC Game] (1999). Numedeon. CA. isfaction, and online gamer loyalty. Computers in Human Behavior, 26(6), 15471554. doi:10.1016/j. Williams, K. D. (2011). The effects of homophily, chb.2010.05.029 identification, and violent video games on play- ers. Mass Communication & Society, 14(1), 324. Thomas, A. (2007). Blurring and breaking through doi:10.1080/15205430903359701 the boundaries of narrative, literacy, and identity in adolescent fan fiction. A New Literacies Sampler, World of Warcraft [PC Game] (2004). Blizzard 29, 137. entertaintment Inc. Trepte, S., & Reinecke, L. (2010). Avatar creation and video game enjoyment: Effects of life-satisfaction, game competitiveness, and identification with the avatar. Journal of Media Psychology: Theo- ries, Methods, and Applications, 22(4), 171184. doi:10.1027/1864-1105/a000022 Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

26 26 International Journal of Gaming and Computer-Mediated Simulations, 6(1), 1-26, January-March 2014 Selen Turkay holds a doctorate in Instructional Technology and Media from Teachers College Columbia University. Currently, she is a postdoc research fellow at Harvard Initiative for Learning and Teaching, Harvard University. Her general research interests include design of personalized, interactive, and collab- orative learning environments in particular gaming and virtual worlds. Specifically, she studies the effects of design choices on learning agency and outcomes, as well as learner experiences including engagement and motivation. Her research approach is a synthesis of mixed methods, qualitative to quantitative (focus groups, case studies, diary studies, content analysis, surveys, eye-tracking, true experiments). Charles K. Kinzer is a Professor of Education and Technology at Teachers College (TC) Columbia Uni- versity, Director of the TC Game Research Lab, and principal investigator, at the Teachers College site, of the Games for Learning Institute. His work has been published widely and presented at the Serious Games Summit, Games Learning and Society, Meaningful Play, Games for Change, the American Educational Research Association, GDC, LRA, and other professional and corporate meetings, and has been funded by the USDOE/IES, NSF, the Robert Wood Johnson Foundation, Microsoft Games, and others. He teaches courses and works extensively with technology in schools and other settings, in areas from software devel- opment and simulations, to virtual environments and games, to reconceptualize educational opportunities for teaching and learning through technology and design across literacy and STEM content areas. Copyright 2014, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.

Load More