Choosing Buddy Icons that look like me or represent my personality: Using Buddy Icons for social presence

Computers in Human Behavior 29 (2013) 1456–1464

Contents lists available at SciVerse ScienceDirect

Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh

Choosing Buddy Icons that look like me or represent my personality: Using Buddy Icons for social presence Kristine L. Nowak ⇑ Communication Department, University of Connecticut, 850 Bolton Road, Storrs, CT 06269-1085, USA

a r t i c l e

i n f o

Article history: Available online 5 March 2013 Keywords: Self presentation Avatars Realism Social presence Anthropomorphism

a b s t r a c t People choose aspects of the self to present that they believe will lead others to have positive impressions of them. The diffusion of telecommunication technologies has led to some of this self-presentation being done using Instant Messaging and other social media applications and devices. When people use Instant Messaging, they select graphical representations to represent them called Buddy Icons. This project asks users to describe the Buddy Icons they are currently using and what they intend to self-present in selecting them to test the extent to which self presentation theory can explain the choices people are making. Overall, participants reported that they felt their Buddy Icons accurately reflected physical characteristics, psychological aspects of the self, or both, which is consistent with previous research that people are relatively honest in their self-presentation online. Those who selected more human like (anthropomorphic) Buddy Icons reported them as more representative of the physical, as opposed to the psychological, self. Finally, users who felt their Buddy Icon accurately represented them reported a stronger sense of identification and felt their Buddy Icon could increase social presence. Implications of these results for online self presentation and impression management are discussed. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction People can selectively use online communication to self present information to an unprecedented number of people (Birnbaum, 2008; Van Der Werf, 2007). Though online communication systems allow people to present specific attributes while concealing, altering, or inventing others, this does not necessarily mean online self presentation is less honest or accurate than offline self presentation (Back, Schmuckle, & Egloff, 2008; Mitja Back et al., 2010; Ellison, Heino, & Gibbs, 2006; Yee, Bailenson, & Ducheneaut, 2009). During the online self-presentation process with some systems, people can select computer generated visual images to represent them (Vazire & Gosling, 2004). This feature is available in a variety of online settings including synchronous text-based chat applications called Instant Messaging (IM), where they are called ‘‘buddy icons’’. All of the most frequently used Instant Messaging applications including as AIM, Yahoo Messenger, Google Chat, and MSN Instant Messenger allow users to display or even create a Buddy Icon of their choice. These Buddy Icons are a widely used (Tamborini & Westerman, 2006) form of self presentation. This project asks current IM users to evaluate the Buddy Icon they are currently using. It examines how realistic and anthropomorphic people report their Buddy Icons to be and the extent to

⇑ Tel.: +1 860 486 4080; fax: +1 800 486 5422. E-mail address: [email protected] 0747-5632/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.chb.2013.01.027

which they feel the Buddy Icon accurately represents them either physically or psychologically and how much they identify with it. It uses structural equation modeling to examine whether Impression Management Theory can predict why people select their Buddy Icons and evaluates whether those who report their Icons to be anthropomorphic and realistic feel more or less physical or psychological homophily, and identification as well as the extent to which users selected the Buddy Icon because they believe it would increase social presence, or salience and connections with others.

2. Self presentation and Impression Management strategies Self-presentation is a process through which people present information about the self intended to influence the impressions others have of them and includes everything people use to express and present the self to others (Arkin & Shepperd, 1990). Impression Management Theory predicts that individuals manage self presentation based on their goals and understanding of what is expected in a situation (Goffman, 1959). Impression Management Theory can be applied to online communication and self-presentation and Impression Management do not happen exclusively online or offline and whether online or offline, people self present in ways that they believe will give others impressions of them that will help fulfill their interaction or relationship goals (Aronson, Wilson, & Akert, 2010; Myers, 2008).

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

2.1. Self-presentation strategies online People adapt their self presentation strategies to utilize the features and social norms of different systems of the medium or system, the goals of the interaction (Back et al., 2008; Birnbaum, 2008; Tamborini & Westerman, 2006), as well as targeted audience for the message (Kassin, Fein, & Markus, 2008). Some argue that self-presentation online (using blogs and web pages) is really a ‘staged performance’ in which an identity or aspects of identity can be selectively targeted for presentation (Chandler, 1998; Trammell & Keshelashvili, 2005). They point out the lack of a visible spontaneous reaction in many computer mediated interactions that allow for more planning time. Users can carefully consider what to say or how to represent themselves, giving them more control over what others see and know about them (Birnbaum, 2008; Bortree, 2005; Tidwell & Walther, 2002; Walther, 1996). Individuals express themselves offline while integrating the many facets of communication technologies to their impression management strategies. It is important to understand that online communication is not isolated and face-to-face and online communication frequently occur in tandem with the same people and sometimes even at the same time (Birnbaum, 2008; Mitrano, 2006; Turkle, 1995). Online systems can be used to enhance and maintain friendships and relationships that began offline (Dwyer, 2007) and most offline relationships eventually move offline (Parks & Floyd, 1996). Thus, the information people self-present online is not necessarily different from what they present offline, but the way the information is presented is clearly different. Communication media require people to adapt and modify the methods and strategies of self-presentation (Boyd, 2004; Chandler, 1998; Vazire & Gosling, 2004). Offline, people use language, clothing choice and behaviors (Schlenker, 1980), and online people use screen names, text and Buddy Icons (Ellison et al., 2006) for this part of the process.

1457

The perception of increased control over impression management online may allow for more selectivity and self-censorship and a wider variety of strategies for exchanging information about the self (Birnbaum, 2008; Trammell & Keshelashvili, 2005; Vazire & Gosling, 2004) but does not make the self presented online less authentic. While some users do lie to others, distort, or misrepresent information, the fears that people will take advantage of features of computer mediated communication to deceive others more online than offline seem unwarranted. People report a desire to present honest representations of themselves even when online, likely because intimacy and relationship maintenance relies on acceptance of the authentic self and most people seek to establish and maintain favorable but believable impressions while presenting information to help them reach their goals (Back et al., 2008; Bargh, McKenna, & Fitzsimons, 2002; Ellison et al., 2006). The affordances of computer systems may give communicators greater control over some of the information others have about them, but it gives less control over other information. Online information can be created with a particular audience in mind but the user cannot control what others see, who views their information, or when. A user cannot control who has access to information once it is shared because it can be shared and re-posted by friends to others without the permission or knowledge of the original sender of the information (Back et al., 2008; Birnbaum, 2008) so family and coworkers can see pictures and posts from friends, and friends of friends, or even strangers. This poses problems with peoples’ ability to control their images and makes targeting messages and sides of a self to particular audiences harder (Boyd, 2004; Chandler, 1998; Vazire & Gosling, 2004). The fact that information is available across multiple groups of people including different groups of friends, as well as family, make it unlikely that deception would go unnoticed or unchallenged, which could lead to more open and honest self-presentation than in offline environments (Ellison et al., 2006). Users

Fig. 1. Predicted model.

1458

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

may modify their self-presentation to be more generic, and less targeted to make it appropriate for multiple audiences including family, friends, coworkers, acquaintances and even strangers at undetermined times in the future. The selection of avatars and Buddy Icons to represent the self seem to follow a similar process to other decisions people make about what to present to others (Ducheneaut, Wen, Yee, & Wadley, 2009), but it also includes some elements of information processing and person perception. When people select an image to represent the self, they first evaluate it as part of a message not associated with the self and then go through a process predicted by information processing theory to evaluate it and then determine whether or not to use it for a particular interaction (Nowak, Hamilton, & Hammond, 2009). Fig. 1 shows the predicted variables influencing Buddy Icon selection. It begins with the process model based on previous research looking at how people perceive avatars being used by others and asking how likely a participant is to select that avatar to represent the self. This previous research based on information processing theory has shown the influence of computer use and perceptions of avatars influence how people perceive others in online interactions (Nowak & Rauh, 2008; Nowak et al., 2009). This project tests whether the same process is followed in selecting Buddy Icons for self-presentation and whether people use these criteria to select the Buddy Icons they actually selected for interactions. The predicted model not only extends the previous research to see if the same process predicting person perception is followed with decisions about self-presentation, but it also separates psychological from appearance homophily and predicts that these variables will increase identification with the Buddy Icon and that people who highly identify with their icon selected it to increase social presence as discussed in the next section.

3. The role of Buddy Icons in identification and social presence As with any other form of self presentation the types of Buddy Icons users select, and the reasons for those selections, is complicated. However, it is likely that people select Buddy Icons that will help others get to know them, present part of the self, and make them more salient and increase social presence. The highly customizable nature of avatars, or graphical representations of the self online, provides not only a virtual image through which the user interacts with others, but also allows users to create a visual identity that represents elements of the self to others (Lee, 2004a, 2004b; Pena, Hancock, & Merola, 2009; Yee et al., 2009), and Buddy Icons are a type of avatar. Buddy Icons, or avatars, are a widely used form of self presentation and provide a way to examine self presentation of identity and the adaptation of the process when it moves online (Pena et al., 2009). In some ways, Buddy Icons are analogous to the physical body offline in that they serve as the visual form of identification and representation of the person. They provide clues to the user’s personality and physical characteristics and can serve as a proxy for the self, influencing how people are perceived within those interactions (Nowak & Rauh, 2008; Nowak et al., 2009), as well as their sense of self (Lee, 2003, 2004a, 2004b; Pena et al., 2009). In other ways Buddy Icons are very different from the offline body. While offline self presentation is limited to things like selecting clothes or hairstyles, online self presentation includes the ability to choose all elements of the visual image presented to others. Online communicators can be freed from the associations of their traditional bodily appearance (Paulos & Canny, 1997), because they are consciously selected and easily changed. This can open up new ways to experience and represent the self (Biocca & Nowak, 2002). Even so, users report a preference for avatars that are realistic, look like them and accurately represent something about them

(Vasalou, Joinson, & Pitt, 2007), though individual differences influence this selection, as discussed in the next section. 3.1. Individual differences influence perceptions of Buddy Icons The process of self presentation is influenced by individual differences, the interaction partner, the user’s motivations, and the context of the interaction (Aronson et al., 2010; Bortree, 2005; Kassin et al., 2008; Myers, 2008; Vasalou & Joinson, 2007). Previous research has shown that males report slightly higher levels of IM use than females and that those with more IM experience and computer use have different perceptions of technology (see also Compeau, Higgins, & Huff, 1999; Eastin & LaRose, 2000). Anthropomorphism is the extent to which an image is perceived to resemble human characteristics and has human morphology (Bailenson, Yee, Merget, & Shroeder, 2006; Hamilton & Nowak, 2010; Nowak et al., 2009). Objects and accessories, which lack visual human-like features, are perceived as having low visual anthropomorphism. Since the user’s perception of a Buddy Icon has been shown to influence the likelihood he or she will select it, and IM use has been shown to influence perceptions of technology generally and anthropomorphism specifically (Nowak & Rauh, 2008), Fig. 1 predicts that more frequent IM users will select images that are higher on anthropomorphism than those using IM less frequently, explaining the positive predicted path between IM use and anthropomorphism. The next variable in the predicted model is perceptual realism. Perceptual realism of digital stimuli can mean the rendering quality of the image itself. In that sense, it would range from cartoon like to photorealistic, though photorealism is not the only dimension of realism (Bailenson et al., 2006; Garau et al., 2003). Certain images may be photorealistic but not able to exist in the offline world as they may portray something from a fantasy world meaning that a photorealistic image could lack probable realism, defined here as something probable, or likely to exist in a non-mediated context (Busselle, Ryabovolova, & Wilson, 2004; Lee, 2004a, 2004b). Previous research has shown that individual differences influence perceptions of avatars and that perceptions of anthropomorphism influence perceptions of realism (Hamilton & Nowak, 2010; Nowak et al., 2009) and that people prefer avatars where the levels of realism and anthropomorphism are complementary (Garau et al., 2003; Seyama & Nagayama, 2007). This leads to the prediction that more frequent users of IM will have more realistic icons and that more anthropomorphic avatars will be higher in realism explaining the predicted paths from IM use, and anthropomorphism, to realism. The model also predicts that those who use IM more will feel more identity with their Buddy Icons. Frequent IM use have seen and engaged with a wider variety of avatars and perhaps have spent more time thinking about possible avatars to select. Further, time with IM where they are using the Buddy Icon will likely make the feel more connected to it, so they are also predicted to feel a higher sense of identification with the Icon than less frequent users of IM, explaining the predicted path from IM use to identification. The next section examines the predicted relationship between realism, anthropomorphism and homophily. Perceptions of avatars on these variables have been shown to predict liking and credibility of other people and objects in survey research (Nowak et al., 2009) and this project tests the extent to which the process influencing information processing and person perceptions match the process predicting Buddy Icons people actually selected for selfpresentation and use in interactions. 3.2. Perceived anthropomorphism and realism drive homophily As discussed above, individual differences have been shown to influence people’s perceptions of and reported likelihood to choose

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

avatars as well as their perceptions of those being represented by the avatar (Bailenson et al., 2005, 2006; Garau et al., 2003; Seyama & Nagayama, 2007; Yee, Bailenson, & Rickertsen, 2007). As shown in Fig. 1, it is the perceptions of realism and anthropomorphism that influence homophily (Hamilton & Nowak, 2010; Nowak et al., 2009), explaining the direct and indirect predicted paths from anthropomorphism and realism to both appearance and psychological homophily. These predictions are explained in this section. Research has shown that people want to represent honest depictions of the self online (Ellison et al., 2006; Yee et al., 2009) and that appearance homophily is a strong predictor of avatar choice (Hamilton & Nowak, 2010; Nowak et al., 2009). Given that Buddy Icons and avatars are used for self presentation of identity (Taylor, 2002), it is likely that people will select those that they feel represents some aspect of them, or how they want to be perceived, meaning that people will select homophilous Buddy Icons. Homophily is the degree of perceived similarity to the self and one can be either visual or psychological (McCroskey, Richmond, & Daly, 1975). Visual or appearance homophily refers to a perceived similarity of physical appearance, whereas psychological homophily refers to a perceived similarity of attitudes or beliefs. As described above, people can select a Buddy Icon that is cartoon like, realistic, or represents an object or fictional character. Previous research has shown that perceived anthropomorphism influences perceived appearance homophily of avatars (Hamilton & Nowak, 2010; Nowak & Rauh, 2008; Nowak et al., 2009). Thus, the model predicts that anthropomorphism will directly predict appearance homophily because those that select highly anthropomorphic icons are likely to select them because they resemble some the physical aspects of the self that they wish to display to others. When users pick human like avatars, they are predicted to select those that they feel visibly resembles them in some way. Alternatively, those who choose to self-present the psychological self or aspects of the personality are likely to select less anthropomorphic avatars such as objects, accessories, or symbols, which lack human-like features. Icons that are lower in visual anthropomorphism are likely to be selected because of the symbolic meaning ascribed to the objects by users (Lasswell & Parshall, 1961). This is why the model predicts that users selecting less anthropomorphic avatars, such as objects, will rate them higher on psychological homophily than those who select more anthropomorphic avatars, explaining the positive path from anthropomorphism to appearance homophily. As explained above, realism is predicted to be influenced by perceptions of anthropomorphism and to directly predict both psychological and appearance homophily. Images that are highly realistic, or authentic, are likely to be more accurate in representing the self. The model predicts that users who select icons that represent important elements of their physical or psychological identity will feel a stronger sense of identification with the Buddy Icon and that they will be more likely to select it to provide social presence as discussed in the next section. 3.3. Identification and social presence with Buddy Icons Impression Management Theory predicts that individuals manage self presentation based on their goals and understanding of what is expected in a situation (subjective and normative beliefs) as well as the feedback they receive from others (Goffman, 1963). The use of Buddy Icons in Instant Messaging is an example of how users can present their identity online in unique ways (Birnbaum, 2008; Van Der Werf, 2007). The use of Buddy Icons, or any type of visual representation, can enhance engagement with the interaction and may make interactions seem more ‘social’ and make the partner more salient, increasing social presence (Nass,

1459

Steuer, Tauber, & Reeder, 1993; Nowak & Biocca, 2003). This section explains why those selecting Buddy Icons that are high on homophily will feel more identification and social presence, explaining the predicted paths from both appearance and psychological homophily to identification and social presence. Social Presence is the perceived or intended connection between people interacting via telecommunication systems (Fulk, Steinfield, Schmitz, & Power, 1987; Rice, 1993; Short et al., 1976; Walther, 1996) and is generally understood to refer to extent to which people feel some sense of connection with, or access to, another person and the salience of interpersonal relationships and is influenced by perceived interpersonal involvement in the discourses or interactions taking place (Lee, 2004a, 2004b; Rice, 1993; Short et al., 1976). It is likely that people will select Buddy Icons and use impression management strategies to increase social presence, or the salience of the interpersonal relationships and people, in interactions. This is particularly likely to be true for those who select homophilous avatars and those who strongly identify with their Buddy Icon, explaining the positive predicted paths from homophily to identification and social presence. Given that selecting avatars and Buddy Icons is a type of self presentation, people should select Buddy Icons that will facilitate the goals of the interaction and their impression management strategies. It is thus predicted that people who select Buddy Icons for their IM interactions that accurately represent them either physically or psychologically, or both, will feel more identification with the Buddy Icon. Further, those who select more homophilous avatars will feel more identification with their Icons and will thus select Icons they feel represents them and will help others understand them. Finally, the model predicts that those who rate the Buddy Icons they are currently using as high on homophily and identification are more likely to select Buddy Icons they believe will increase the sense of social presence. 4. Methodology 4.1. Participants The participants in this survey were recruited from undergraduate communication classes at a large public university in the United States. While 93 users of IM responded, 7 cases were excluded from the all analysis because they did not complete the survey, leaving a total sample size of 86. They ranged between 18 and 25 years of age and were 52% male and 48% female. All participants reported using Instant Messaging, which was required to participate in the study. 4.2. Measurement instruments All scales were evaluated for acceptable item quality by a series of exploratory and confirmatory factor analyses. Three criteria were imposed on items: homogeneity of content, internal consistency, and external consistency. For homogeneity of content, items had to measure the same content, or have content validity and tap into the same underlying theme. This test examines whether the matrix of correlations among the items on the scale is relatively flat (when items are of equal quality) or whether it approximates a Guttman Simplex (when there is a gradient of item quality). Internal consistency was examined through standard score coefficient alpha reliabilities, also known as unidimensionality. Finally, for tests of external consistency, items were compared across all scales to determine how each item correlates with other scales in the study. Items were considered acceptable if they loaded at a minimum of .5 on their primary structure and did not have equal

1460

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

Table 1 Measurement instruments and scale means. Dependent variable

Items retained after tests of internal consistency and reliability

Frequency of IM use. M = 3.2, SD = 1.08

How frequently do you use IM?

Anthropomorphism (a = .85) M = 3.73, SD = 1.81

This This This This

Realism (a = .90) M = 4.14, SD = 1.34

This icon is real is unreal has cartoon-like features is photorealistic, is natural is artificial is realistic is authentic is genuine

Appearance homophily (a = .84) M = 3.38, SD = 1.65

This image is similar to me This image looks like me This image resembles me

Psychological homophily (a = .89) M = 3.61, SD = 1.41

This This This This This This

Identification (a = .89) M = 4.03, SD = 1.53

I relate to this image I identify with this image This image represents something in me This image has meaning for me I wish I could be more like the character in this image

Social presence (a = .90) M = 3.13, SD = 1.23

This Buddy Icon Gives others a good idea of who I am Helps others better understand my message Provides a sense of realism Helps others better understand me Makes me seem more real, makes it seem more like my communication partner and I are in the same room Makes it seem more like we are having a face-to-face conversation Would allow others to know me well even if I only met them online

image image image image

image image image image image image

looks human has human features has human-like expressions is life-like

represents represents represents represents represents represents

my attitudes how I feel my social status who I am inside my values my background

or higher loadings on other scales, also called parallelism. Items that loaded higher on other factors in the analysis of parallelism were deleted from the scale. See Table 1 for details on retained items. Frequency of IM use was measured using a one item 5-point scale (1 = almost never; 5 = constantly). Realism was measured using a nine item 7-point scale (not at all to very much) adapted from (Nowak et al., 2009). Anthropomorphism was measured using a six item 7 point scale (not at all to very much) adapted from (Nowak et al., 2009), with four items remaining after tests of reliability and validity. Homophily was measured two modified subscales measuring appearance and psychological homophily (McCroskey, Hamilton, & Weiner, 1974) using likert type items on a 7 point scale (not at all to very much). Identification (Eyal & Rubin, 2003) with the Buddy Icon was measured using was measured with a scale created for this project consisting of five items on a 7 point scale (not at all to very much).

Intended social presence the extent to which users reported selecting the Buddy Icon to increase social presence with Buddy Icons was measured using modified items from (Short et al., 1976). The prompt ‘‘This buddy icon’’ was given to the items. Following confirmatory factor analysis, one item (persuade others) was removed, leaving seven items. 4.3. Procedure Participants were contacted via email and asked to provide a copy of the Buddy Icon they were currently using in their IM interactions as an email attachment, and were asked to fill out an online questionnaire evaluating the Buddy Icon they provided in terms of its anthropomorphism, realism, physical and psychological homophily, identification, intended social presence and other variables in addition to some demographic questions. 5. Results The predictions discussed above and presented in Fig. 1 were tested using the structural equation modeling program PMOD (Hunter & Hamilton, 2003). Structural equation modeling allows the testing of multiple relationships together and reduces the errors resulting from doing independent tests for each predicted relationship because conducting multiple tests increases the type I error rate. Further, the model does not predict direct effects with independent variables. Instead the model predicts that there is covariation between some variables and that variables intervene between the description of the characteristics of the avatars and identification and social presence, which would not be seen in independent tests. While a sample size of more than 200 has been stated as the ideal for causal modeling, a smaller sample size in this model is acceptable because it meets three of the criteria outlined by Kenny (2012). First, it meets the goal of greater than a 5 to 1 ratio of sample size to free parameters, with an average of 6.6 in the predicted model. Second, it is a simple model with only eight variables, and third the correlations are moderate to large. For models such as the one presented here with between 75 and 200 cases, the chi square test is a reasonable measure of fit (Kenny, 2012). Also, this is a test of a predicted model based on previous research and derived from theory so other indicators such as size of path models and absence of missing links are important in evaluating the model as well. The overall goodness of fit of the predicted model was not acceptable RMSE = .12, X2(13) = 14.20, p = .36. More importantly, there were a few large errors between what was obtained and predicted on some paths and a few large missing links. The model was respecified by first removing four small and non-significant paths. In this step, the paths between psychological homophily and social presence (.04), between biological sex and IM use (.05), and between IM use and both identification (.07), and realism ( .04) were removed. The next step in respecifying a model is to examine the missing links. In structural equation modeling, the missing paths analysis provides meaningful information as it reveals indirect paths, interactions, and other connections. Adding these paths can improve model fit and provide a more clear illustration and understanding of how the process works. That resulted in the addition of three paths as illustrated in Fig. 2 by the dashed lines. One from IM use to appearance homophily and two from anthropomorphism to both and identification and psychological homophily, both of which are moderate and negative. The respecified model is a good fit with the data RMSE = .065, X2(14) = 4.86, p = .99. The goodness of fit of the respecified model was acceptable with no significant errors. The added paths involved direct paths where

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

1461

Fig. 2. Respecified model with dashed lines for added paths.

only indirect paths had been predicted and were not inconsistent with the theory or previous research on which the predicted model was based. The added paths both included anthropomorphism suggesting this variable has a direct effect on more variables than predicted and as discussed below, the reasons for this should be explored in future research.

6. Discussion Consistent with self-presentation theory, people report that the Buddy Icons they selected represented either their physical appearance or some aspect of their personality and they reported strong identification with the Buddy Icons they were using. This section will examine the results with respect to the predicted model and implications for theory development and future research. The data were mostly consistent with the predicted model, with a few exceptions to be discussed in this section. Beginning at the far left of Fig. 1, the first predicted path between biological sex and IM use was not retained. It was small and not significant and removing it from the model improved the fit suggesting there is not a direct path. This may reflect recent trends showing that males and females are using computers and the Internet with about the same frequency. These data also show a decline in the frequency of IM usage as compared to studies done just a few years ago (Nowak & Rauh, 2008), suggesting that computer users may be moving away from using Instant Messaging systems. However, males still reported higher social presence than females. This could reflect a difference in the types of avatars males and females are selecting, or differences in how they use IM, and research investigating these possibilities is underway. As predicted, more frequent IM users were more likely to rate their Buddy Icon as high on anthropomorphism (.13), though this

path was not as large as it has been in previous research examining perceptions of different avatars. This data had different users evaluating the Buddy Icons they were currently using and not the same icons. This means that those using IM more frequently rated their Buddy Icons as more anthropomorphic than those using IM less frequently. These data cannot evaluate whether the Buddy Icons actually were more or less anthropomorphic. The next predicted path directly from IM use to realism was small and not significant so was removed from the model, though this does not mean IM use does not influence realism. The relationship is moderated through anthropomorphism, which is consistent with predictions that perceptions of anthropomorphism drive expectations for the realism. These results suggest that people spending more time with IM select Buddy Icons they perceive to have higher social potential or are more anthropomorphic and more realistic emphasizes the effect of perceived anthropomorphism on perceived realism found in previous research. Previous research has shown realism and anthropomorphism have similar relationships to variables, which is not surprising, given their high paths and strong correlations. The model is consistent with previous experimental and survey research showing that perceptions of anthropomorphism are driving the choice of avatars in interactions. It further confirms that perceptions of anthropomorphism influence perceptions of and expectations for realism, and that those wanting an anthropomorphic image also want it to be high on realism. The separation of perceived psychological homophily, which relates to the sense that the Buddy Icon represents something about the self, and appearance homophily, which represents physical characteristics have shown an important distinction between the two constructs. First, the paths from anthropomorphism to appearance homophily and realism were consistent with predictions in that those who do select Buddy Icons that are high on anthropomorphism are likely to select icons

1462

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

that they feel have some similar physical characteristics to the self and more realistic avatars were also more likely to be rated higher on psychological homophily, though this was a smaller path than that between realism and appearance homophily. While the strong connection between anthropomorphism and appearance homophily was predicted, no direct path from anthropomorphism to psychological homophily or identification was predicted because it was thought that those selecting highly anthropomorphic avatars would select those that they feel resemble them physically and those who selected avatars they felt were high on psychological homophily would select objects or other things that would be less anthropomorphic. The direct negative path from anthropomorphism to identification was not predicted as this relationship was predicted to go through homophily. The negative path from anthropomorphism to psychological homophily in the respecified model suggests that not only are people who want to select an avatar high on appearance homophily more likely to select an anthropomorphic avatar, but also that people selecting avatars high on psychological homophily are less likely to select an anthropomorphic avatar. Controlling for these negative relationship between anthropomorphism and psychological homophily and identity raised the path between appearance homophily and psychological homophily from .58 to .80, clearly displaying the importance of this path and that psychological and appearance homophily are not unidimensional even though they are highly correlated. The presence of these unpredicted negative paths and the absence of a path from appearance homophily to identification suggests that those who chose an anthropomorphic image will do so because it is physically similar to the self but that they identify most with those that also represent them psychologically. It further suggests that those selecting more anthropomorphic Buddy Icons felt less identification, except when they rated them as physically or psychologically similar to them. Consistent with this explanation, the predicted direct path between appearance homophily and identification was also not retained. IM use did not directly predict identification with the Icon as predicted, suggesting that more time with IM alone does not increase identification but there is an indirect path through anthropomorphism. The direct predicted path from appearance homophily to identification was not significant. This relationship is indirect through psychological homophily, likely because of the negative relationship between anthropomorphism and identification. Finally, as predicted those who selected buddy Icons that they identified with, or that were either physically or psychologically homophilous reported that they selected their icons to increase social presence.

7. Conclusions This project asked IM users to evaluate the Buddy Icons they were using to examine the factors predicting the selection and the process of Impression Management online. Results are mostly consistent with predictions based on the theory and previous experimental and survey research. People seek to establish and maintain favorable impressions and present information to make these impressions believable and to make them more salient or socially present to the ‘audience’ or communication partners. People work to self-present in ways they feel will accurately reflect them either physically or psychologically that will help them reach their interaction goals, which can be to increase social presence and a sense of connection with partners when online. In this project, users were asked to indicate their perceptions of their Buddy Icon. Most users reported that their Buddy Icon accurately represented them either psychologically, physically, or both. This is consistent with previous findings that users prefer to select

images for interactions that closely resemble the self. The confirmation of this process model also shows that people use a very similar process model in evaluating the Buddy Icons they are currently using that they use to evaluate the Buddy Icons being used by others. If so, this could mean that when people first encounter a Buddy Icon, whether it represents someone else or they are considering it for themselves, they begin the evaluation of what it stands for and how it might be interpreted. They may then make a comparison to their self-concept and the identity they want to portray to maximize similarity to the self, explaining why Buddy Icons that are perceived as similar to the self they want to portray are most likely to be selected for interactions (Hamilton & Nowak, 2010; Nowak & Rauh, 2008; Nowak et al., 2009). It seems clear that people are utilizing the ability to select their own image online to personalize their IM Buddy Icons. They are not using them to create fantasy characters or to be something they are not, but they are instead using the technology and options to personalize the image to present fairly realistic parts of their identities, images they see as high on homophily, to others. Further research should continue to explore this process. This is also consistent with previous research showing that people seek to be relatively honest and accurate in their online selfpresentations (Ellison et al., 2006). The data here cannot determine whether people actually were accurate or honest, only that people reported that their Buddy Icons accurately represented some part of the self. Future research should ask other people to evaluate the Buddy Icons and indicate whether they agree that the Buddy Icons are accurate representations of the people who have selected them. Researchers should also examine the boundary conditions of this finding and which types of characteristics people include, and which ones they leave out, when selecting Buddy Icons. Also, future research should further examine the possibility that people use Buddy Icons in online settings to present aspects of their true selves they are unable or unwilling to present offline (Bargh et al., 2002). This also does not answer the question of how people adapt the Impression Management process in light of the fact that information stays online across audiences and over long periods of time. Some important questions along these lines is how frequently users select anthropomorphic avatars, or how often they alter the Buddy Icon they select for representation across different contexts or with different interaction partners, and research on these questions is currently underway using the Buddy Icons selected for this project. Finally, the research supports the prediction that those who felt their Buddy Icons resembled them physically or psychologically felt a much stronger sense of identification with them and believe these Buddy Icons will increase social presence. In this context it seems that people are choosing anthropomorphic avatars when they desire to be represented accurately in terms of physical characteristics but less anthropomorphic avatars when they wish to represent a psychological aspect of the self. Further, it suggests that those selecting images that look like them identify with them less and are less likely to have selected their Buddy Icons that would enhance social presence. The differing effects and relationship to identification adds to our understanding to this process and confirms that this process holds with the Buddy Icons people actually select for interactions. This is consistent with both Impression Management Theory and predictions of social presence scholars arguing that people will manage self presentation and their media selection and use to best improve the relationship and to meet the interaction goals, as well as to manage impressions. It shows that people want to select media and use it in ways that will be consistent with their understanding of the social norms (Fulk, Schmitz, & Steinfield, 1990) and to self present in ways that are expected and socially appropriate (Goffman, 1959).

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

Social presence was predicted by identification with the Buddy Icon as well as both appearance and psychological homophily. This means that those who selected Buddy Icons that looked like them or represented part of their personality did so because they wanted to enhance the sense of salience, or connection with their interaction partners. Further, those who identified more with their Buddy Icons reported that they felt it could enhance the sense of social presence in the interaction. It thus seems to be a key variable in understanding why people select their Buddy Icons and their goals in using online media more generally, and how they use online systems to represent the self and express identification. The results suggest that overall users feel more identification with avatars that look like them physically and also represent some aspect of the psychological and they feel reduced identification without both. Future research should examine this carefully to see if users most want an avatar to represent their gender, race, or other physical or personal interests or characteristics. It will be interesting to find out which a user would select if given a choice between an avatar that matches gender or race, and one that matches a hobby, or another aspect of the personality. This has potential implications for systems that provide limited choices of avatars for users to choose as people who do not feel any of the choices represent them both psychologically and physically are likely to feel less identification or social presence with the system or the interaction.

References Arkin, R., & Shepperd, R. (1990). Strategic self presentation: An overview. In M. Cody & M. McLaughlin (Eds.), The psychology of tactical communication (pp. 175–193). Clevedon: Multilingual. Aronson, E., Wilson, T. D., & Akert, R. M. (2010). Social psychology (6th ed.). Upper Saddle River, NJ: Prentice Hall. Back, M. D., Schmuckle, S. C., & Egloff, B. (2008). How extroverted is [email protected]? Inferring personality from email addresses. Journal of Research in Personality, 42, 1116–1122. doi: 1110.1016/ j.physletb.2003.1110.1071. Back, M. D., Stopfer, J. M., Vazire, S., Gaddis, S., Schmukle, S. C., Egloff, B., et al. (2010). Facebook profiles reflect actual personality, not self-idealization. Psychological Science, 21(3), 372–374. Bailenson, J. N., Swinth, K. R., Hoyt, C. L., Persky, S., Dimov, A., & Blascovich, J. (2005). The independent and interactive effects of embodied agent appearance and behavior on self-report, cognitive, and behavioral markers of copresence in immersive virtual environments. PRESENCE: Teleoperators and Virtual Environments, 14, 379–393. Bailenson, J. N., Yee, N., Merget, D., & Shroeder, R. (2006). The effect of behavioral realism and form realism of real-time avatar faces on verbal disclosure, nonverbal disclosure, emotion recognition, and copresence in dyadic interactions. Presence: Teleoperators and Virtual Environments, 15(4), 359–372. Bargh, J. A., McKenna, K. Y. A., & Fitzsimons, G. M. (2002). Can you see the real me? Activation and expression of the ‘‘true self’’ on the Internet. Journal of Social Issues, 58(1), 33–48. Biocca, F., & Nowak, K. (2002). Plugging your body into the telecommunication system: Mediated embodiment, media interfaces, and social virtual environments. In D. Atkin & C. Lin (Eds.), Communication technology and society (pp. 409–446). Cresskill, NJ: Hampton Press. Birnbaum, M. (2008). Taking Goffman on a tour of Facebook: College students and the presentation of self in a mediated digital environment. Unpublished doctoral dissertation, University of Arizona. Bortree, D. (2005). Presentation of self on the web: An ethnographic study of teenage girls’ weblogs. Education, Communication & Information, 5(1), 25–39. http://dx.doi.org/10.1080/14636310500061102. Boyd, D. (2004). Friends, Friendster and MySpace top 8: Writing community into being on social networking sites. First Monday, 11(12). . Busselle, R., Ryabovolova, A., & Wilson, B. (2004). Ruining a good story: Cultivation, perceived realism and narrative. Communications, 29, 365–378. Chandler, D. (1998). Personal homepages and the construction of identities on the web. In Aberystwyth post-international group conference, Wales, UK. . Compeau, D., Higgins, C. A., & Huff, S. (1999). Social cognitive theory and individual reactions to computing technology: A longitudinal study. MIS Quarterly, 23(2), 145. Ducheneaut, N., Wen, M., Yee, N., & Wadley, G. (2009). Body and mind: A study of avatar personalization in three virtual worlds. In Computer human interactions (CHI) conference, Boston, MA.

1463

Dwyer, C. (2007). Digital relationships in the ‘MySpace’ generation: Results from a qualitative study. In 40th Hawaii international conference on system sciences (HICSS), Big Island, HI. . Eastin, M. S., & LaRose, R. (2000). Internet self-efficacy and the psychology of the digital divide. Journal of Computer Mediated Communication, 6(1). Vol. 2002. Ellison, N., Heino, R., & Gibbs, J. (2006). Managing impressions online: Selfpresentation processes in the online dating environment. Journal of ComputerMediated Communication, 11(2), 415–441. Eyal, K., & Rubin, A. (2003). Viewer aggression and homophily, identification, and parasocial relationships with television characters. Journal of Broadcasting Electronic Media, 41(1), 77–98. Fulk, J., Schmitz, J., & Steinfield, C. (1990). A social influence model of technology use. In J. Fulk, J. Schmitz, & C. Steinfield (Eds.), Organizations and communication technology (pp. 117–140). London: Sage Publications. Fulk, J., Steinfield, C., Schmitz, J., & Power, J. (1987). A social information processing model of media use in organizations. Communication Research, 14(5), 529– 552. Garau, M., Slater, M., Vinayagamoorthy, V., Brogny, A., Steed, A., & Sasse, M. A. (2003). The impact of avatar realism and eye gaze control on perceived quality of communication in a shared immersive virtual environment. In Proceedings of CHI’03 – The conference on human factors in computing systems. ACM Press. Goffman, E. (1959). The presentation of self in everyday life. Garden City, NY: Doubleday. Goffman, E. (1963). Behavior in public places; notes on the social organization of gatherings. New York: The Free Press. Hamilton, M. A., & Nowak, K. L. (2010). Advancing a model of avatar evaluation and selection. PsychNology Journal, 8(1), 33–65. (retrieved 06.27.10). Hunter, J. E., & Hamilton, M. A. (2003). Path program to perform least squares path analysis. In Version 6.0 for windows. Kassin, S. M., Fein, S., & Markus, H. R. (2008). Social psychology (7th ed.). Boston, MA: Houghton Mifflin. Kenny, D. (2012). Measuring model fit. In Structural equation modeling. Retrieved December 2012. Lasswell, T., & Parshall, P. (1961). The perception of social class from photographs. Sociology & Social Research, 45, 407–414. Lee, J. R. (2003). Consuming virtual identity: Exploring the motivation based of participation in online avatar marketing. In International Communication Association conference, San Diego, CA. doi: ica_proceeding_12207.PDF. Lee, E.-J. (2004a). Effects of gendered character representation on person perception and informational social influence in computer-mediated communication. Computers in Human Behavior, 20(6), 779. Lee, K. M. (2004b). Presence, explicated. Communication Theory, 14(1), 27–50. McCroskey, J., Hamilton, P., & Weiner, A. (1974). The effect of interaction behavior on source credibility, homophily, and interpersonal attraction. Human Communication Research, 1(1), 42–52. McCroskey, J., Richmond, V., & Daly, J. (1975). The development of a measure of perceived homophily in international communication. Human Communication Research, 1(4), 323–332. Mitrano, T. (2006). A wider world: Youth, privacy, and social networking technologies. Educause Review, 41(6), 16–29. . Myers, D. G. (2008). Exploring social psychology (Vol. 5th). Boston, MA: McGraw-Hill. Nass, C., Steuer, J., Tauber, E., & Reeder, H. (1993). Anthropomorphism, agency, & ethopoeia: Computers as social actors. In InterChi ’93, Amsterdam, the Netherlands. Nowak, K. L., & Biocca, F. (2003). The effect of agency and anthropomorphism on users’ sense of telepresence, social presence and copresence in virtual environments. PRESENCE: Teleoperators and Virtual Environments, 12(5), 481–494. MIT Press. Nowak, K. L., Hamilton, M. A., & Hammond, C. C. (2009). The effect of image features on judgments of homophily, credibility, and intention to use as avatars in future interactions. Media Psychology, 12, 50–76. http://dx.doi.org/10.1080/ 15213260802669433. Nowak, K. L., & Rauh, C. (2008). Examining the perception process of avatar anthropomorphism, credibility and androgyny in static and chat context. Computers in Human Behavior, 24(4), 1473–1493. Parks, M. R., & Floyd, K. (1996). Making friends in cyber space. Journal of Computer Mediated Communication, 1(4). article 5. .. Paulos, E., & Canny, J. (1997). Ubiquitous tele-embodiment: Applications and implications. International Journal of Human-Computer Studies, 46, 861– 877. Pena, J., Hancock, J., & Merola, N. (2009). The priming effects of avatars in virtual settings. Communication Research, 36(6), 838–856. doi: 810.1177/ 0093650209346802. Rice, R. (1993). Media appropriateness: Using social presence theory to compare traditional and new organizational media. Human Communication Research, 19(4), 451–484. Schlenker, B. (1980). Impression management: The self-concept, social identity, and interpersonal relations. Monterey, CA: Brooks/Cole. Seyama, J., & Nagayama, R. S. (2007). The uncanny valley: Effect of realism on the impression of artificial human faces. Presence, 16(4), 337–351. Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications. London: John Wiley & Sons, Ltd..

1464

K.L. Nowak / Computers in Human Behavior 29 (2013) 1456–1464

Tamborini, R., & Westerman, D. (2006). The effect of avatars on uncertainty reduction. In International Communication Association conference, Dresden, Germany. Taylor, T. L. (2002). Living digitally: Embodiment in virtual worlds. In R. Schroeder (Ed.), The social life of avatars; presence and interaction in shared virtual environments (pp. 40–62). London: Springer-Verlag. Tidwell, L. C., & Walther, J. B. (2002). Computer-mediated effects on disclousre, impressions, and interpersonal evaluations: Getting to know one another a bit at a time. Human Communication Research, 28(3), 314–348. Trammell, K. D., & Keshelashvili, A. (2005). Examining the new influencers: A selfpresentation study of a-list blogs. Journalism and Mass Communication Quarterly, 82(4), 968–982. Turkle, S. (1995). Life on the screen: Identity in the age of the internet. New York: Simon & Schuster. Van Der Werf, M. (2007). Beware of using social-networking sites to monitor students, lawyers say. The Chronicle of Higher Education.

Vasalou, A., Joinson, A. N. & Pitt, J. (2007). Constructing my online self: Avatars that increase self-focused attention. In Computer human interaction (CHI) conference, San Jose, CA. Vazire, S., & Gosling, S. D. (2004). E-perceptions: Personality impressions based on personal websites. Journal of Personality and Social Psychology, 87(1), 123–132. doi: 110.1037/0022-3514.1087.1031.1123. Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and hyperpersonal interaction. Communication Research, 23, 3–43. Yee, N., Bailenson, J. N., & Ducheneaut, N. (2009). The proteus effect: Implications of transformed self-representation on online and offline behavior. Communication Research, 36(2), 285–312. doi: 210.1177/0093650208330254. Yee, N., Bailenson, J. N., & Rickertsen, K. (2007). A meta-analysis of the impact of the inclusion and realism of human-like faces on user experiences in interfaces. In Paper presented at CHI, San Jose, California, USA, April 28–May 3, 2007.