Industry 4.0, digital transformation, computer-mediated communication (CMC), e-learning, social media, big data, cloud computing, virtual/augmented reality (VR/AR), or Internet of things (IoT) are some of the current trending topics in the business field. Technology is transforming business, and there is a need to successfully utilise new technologies, adopting processes that allow company leadership and staff to strategically apply new technologies on an ongoing basis (Herbert, 2017).
The integration of enterprise social media, which enables workers to communicate online and create digital spaces sharing and exchanging information (Leonardi, Huysman, & Steinfield, 2013) – and consequently the transformation of the business and education structures supporting this digitalised era – is within a dynamic area where ubiquitous technologies and social aspects intersect. In this sense, different technologies might ascribe socio-psychological features related to people's behaviour and personality (Lytras et al., 2013).
Social media and CMC are understood as remote interactions that take place through the use of computer networks (Baym, 2015), allowing a user to overcome challenges in regard to distance, through the development of tools and processes supporting cooperation in distributed teams (Giuffrida & Dittrich, 2015). The key barriers to global collaboration are geographic, temporal, linguistic and cultural distance (Noll, Beecham, & Richardson, 2011). Modern technologies offer primary solutions in overcoming barriers with the provision of site visits and synchronous communication technologies that enable real-time communication (Walther, 1992) and a knowledge-sharing infrastructure (Noll et al., 2011). Another of the traits of global collaboration is the inclusion of a diverse workforce. Diversity, understood as the presence of difference, would include dimensions such as ethnicity, gender, age, religion, ability (mental and/or physical), sexual orientation and socioeconomic status (Blaine & Brenchley, 2018). Managing a diverse workforce through CMC requires very specific knowledge and much preparation.
Nevertheless, CMC is essential for global collaboration. CMC enables users to reach out to a vast number of receivers simultaneously, allowing to notify a large group of people about the same thing, and it helps to break down the barriers of communication such as physical limitations, shyness, or appearance (Althaus, 1997) While the main media for communication in many globally distributed teams has traditionally been telephone, email and video conferencing systems (Giuffrida & Dittrich, 2015), alternative technologies such as VR offer some potential, since VR allows for the construction of digital simulations that people can experience as virtually real. VR dissipates the separation between what is digital and what is real, and it enables the user to step into digital worlds and feel a full envelopment (Tredinnick, 2018). Until recently, VR has found application in tele-presence, remote collaboration, data visualisation, medicine (including surgical simulation), entertainment and education, among others (Fisher et al., 2018), but the very extent of how VR might develop and change the world still remains unclear (Slater & Sanchez-Vives, 2016). VR has a wide range of applications in psychology and neuroscience, but there has been little research linking socio-psychological investigations in the area of human behaviour with education or industry contexts. Thus far, the key contexts in which VR is being applied in business are simulation, situations requiring co-presence and training applications (Tredinnick, 2018).
In this paper, we present findings from an empirical study on stereotypical perception in a VR setting. The study was a co-present situation, where participants had to interact and negotiate different roles in a virtual room. In the negotiation process, it was observed whether gender had an influence on role assignment. Traditionally, women have been negatively stereotyped and perceived as less competent within digital communities (Kaye, Pennington, & McCann, 2018). Knowing how far sexism and gender stereotypes are present in virtual environments, or on social media will be relevant for diversity management when integrating VR in education and business encounters. Therefore, within the framework of these criteria, the paper aims to provide insights into understanding the extent to which gender stereotypes are present in avatar-based VR, inviting a discussion about the impact it could have when using VR in an educational setting.
The paper is structured as follows. We first present the hypotheses of the study based on research on avatars and social perception, demonstrating how broad and intricate the issue of stereotypes is and how it relates to avatar-based VR. Subsequently, the research design is presented to determine the extent to which gender stereotypes are present in avatar-based VR. In conclusion, a discussion about how avatar creation can affect social learning situations in virtual settings will be initiated.
Our perception of other people is – at least partially – informed by stereotypes. A stereotype offers a simplification of a complex social reality; it comprises knowledge about attributes and behaviour of people (Denmark & Paludi, 2007). As socio-cognitive knowledge structures, stereotypes are images that people carry within themselves, yet they are socially shared, created and recreated through social situations. This dual structure of stereotypes leads to a continuous validation through consensus, leading to alleged knowledge of a social group being perpetuated in every act of repetition (Eckes, 2008; see also Ashmore & Del Boca, 1979; Schneider, 2004).
In CMC, avatars symbolise the self in interaction; therefore, when different users interact in digital environments, judgements, attributions and stereotypes will be based on the appearance and behaviours of others’ avatars; avatars may influence users’ behaviours in the interactions (Nowak & Rauh, 2005; Nakamura, 2009). The different representations of avatars have an influence on the perception and attribution of determined sources and characteristics (including stereotypes), having implications for digital interactions. Characteristics of avatars are determined by user preferences, experiences within different environments, social norms and technological capabilities – these characteristics in turn can augment or limit people's ability to self-present and engage in digital environments (Blascovich et al., 2009; Stromer-Galley & Martey, 2009). Among the characteristics included would be appearance, abilities, traits, or behaviours that can either reflect reality or not. Apart from providing a form of embodiment that enables navigation through virtual spaces, avatars can also facilitate non-verbal communication via gestures, body posture, proxemics and even haptics (Biocca, 1997). As avatars have an influence on beliefs, attitudes and behaviours in communication, differences in representations are notable (Nowak & Fox, 2018).
To understand avatar perception, information processing theory (Walther, 1992) provides insights about how people attribute emotions or intentionality based on categorisations such as age or gender (McGuire, 1968). The more human-like the avatars are perceived to be, the more likely it is that the findings and theories of human communication will also apply to avatar-based interactions. Therefore, understanding how people perceive avatars might help to predict processes and outcomes in CMC. Among the aspects that can influence users’ perceptions of avatars would be agency (referred to as people's perception of an entity as a human or not), anthropomorphism (referring to possessing human forms or behaviours) and realism (related to having accurate forms or behaviours).
Due to the fact that avatars are perceived as social entities, people engage similar categorisation processes as in reality, applying the same heuristics and stereotypes associated with human members of the different categories (Nowak et al., 2006). In relation with the information processing theory (Walther, 1992), the more social that potential avatars are perceived to be, the more likely it is that this categorisation would occur. One of the most common categorisations humans make of others is based on the other's gender. In understanding others, identifying appropriate interaction scripts and predicting behaviours, people believe that sex categorisation provides useful information (Lakoff, 2008). Thus, unsurprisingly, sex and gender attributions have remained salient in CMC contexts (Fox et al., 2015).
One of the most prominent models of stereotypical intergroup/interpersonal perception is the stereotype content model (SCM), which conceptualises social perceptions on two basic dimensions: warmth and competence (Fiske, 1999, 2002; Cuddy, Fiske, & Glick, 2008). ‘Warmth’ describes how benign, likable, und warm members of a certain group are considered to be, and ‘competence’ describes how able, competent and independent the group members are perceived. Generally speaking, women are perceived to be warmer than men, whereas men elicit higher ratings of competence (Fiske, 2002; Cuddy, 2007). High ratings on both dimensions are commonly reserved for the members of one's in-group or groups that are highly admired in society, such as celebrities and stars. Remarkably, the perception of out-groups is often not generally negative, but ambivalent, with high values on one dimension and low values on the other dimension (Fiske, 2007). Groups that are classified comparably on these dimensions usually provoke similar emotional and behavioural responses (Cuddy, 2007).
Typically, some groups are perceived as cold but competent (e.g., rich people) and are envied, whereas other groups are perceived as warm but incompetent (e.g., the elderly or housewives) and are pitied. The SCM reveals how references based on these two dimensions lead to subtle and complex patterns of stereotyping, prejudices and discriminating behaviour (Fiske, 2007; Cuddy et al., 2008). Notably, gender stereotypes have not only descriptive aspects (i.e., how members of a group
Research on stereotypes shows that people unwittingly ascribe characteristics to others based on the latter's physical characteristics such as biological sex and race (Nakamura, 2009; Lee, Nass & Bailenson, 2014). Regarding education or business attainment, gender stereotypes and bias, understood as cultural beliefs about gender (Ibarra & Petriglieri, 2016), are reflected in occupational segregation of men and women. In literature, vertical segregation, also known as the ‘glass ceiling’, refers to the invisible barrier and difficulties that women face in making their way up in the corporation while preventing them from fully developing professionally (Barreto, Ryan, & Schmitt, 2009). The word ‘ceiling’, denoting high distance, refers to the struggle for women in climbing up the organisational ladder, while the word ‘glass’ reflects the lack of a clear visibility of this barrier (Barreto, Ryan, & Schmitt, 2009). The newly coined expression has inspired the rise of many other terms: ‘glass walls’ refers to horizontal segregation, i.e., high concentration of women in certain departments, such as marketing, sales, or human resources (Miller, Kerr, & Reid, 1999); ‘glass slipper’ indicates the lack of ambition in women to leadership positions (Rudman & Heppen, 2003); and ‘glass cliff’ refers to the fragile and uncertain presence of women on top (Ryan & Haslam, 2007). The variety of terms confirms the gender stereotypes and bias that negatively affect women's leading role assumptions.
Unsurprisingly, sex and gender attributions have remained salient in CMC contexts (Fox et al., 2015). Gender stereotypes also occur in CMC (Nowak, Hamilton, & Hammond, 2009) and, in comparison to male avatars, female avatars are subject to more sexual harassment (Behm-Morawitz & Mastro, 2009). Stereotypical or sexualised virtual representations of women elicit more sexism than non-stereotypical representations (Fox & Bailenson, 2009), encouraging the perception of women as less intelligent (Behm-Morawitz & Mastro, 2009), in line with the assumptions of the SCM that demonstrated the association of femininity with lower competence, as women generally are attributed a lower level of competence traits than males (Fiske, 2002; Cuddy, 2007).
One of the possibilities offered by avatar customisation is the creation of androgynous avatars in which gender categorisation is not possible. However, to our knowledge, no research has been done to understand how people process and categorise androgynous avatars. This additional research would be particularly interesting considering self-similarity and social identity approaches. Bearing in mind that the characteristics of the avatars influence the way in which the receivers interpret messages and sources, manipulating the characteristics of the avatars and playing with androgynous dyads in CMC environments would be interesting to see how outcomes such as social influence, communication satisfaction, or task performance are affected.
In relation to avatar selection as self-representation, it is important to consider that the way people present themselves has an impact on people's ability to achieve their social goals (Goffman, 1978) and on career success (Ibarra, 1999). Besides attributes such as gender, age and ethnicity, the physical height of a person has also found to be an influential factor in human communication and a potential source of stereotyping (Judge & Cable, 2004; Moeller, Robinson, & Zabelina, 2008). Being taller is associated with a number of qualities, such as being more dominant, composed, expressive, dramatic, self-assured, persuasive, having more leadership qualities and showing less self-censorship (Burgoon & Dunbar, 2000). Regarding AR settings, Walker, Szafier and Rae (2019) have found that the relative size of the other interaction partner avatar matters: they found equalsized avatars of a remote user to be significantly more influential (on the local user) than small-sized avatars.
In digital environments, the flexibility to modify self-representation is much greater than in face-to-face settings. As summed up in the Introduction section, the use of avatars for self-representation influences people's evaluation and judgement. As users make judgements of avatars, avatar selection is significant to meet interaction goals. Through avatar selection, people aim to identify and express the self to others. It is common among users to choose avatars that truthfully represent any physical or psychological trait about them (Nowak & Rauh, 2005). Even when users try to convey elements of their social identities, such as gender, age, or race, when selecting avatars (Nowak & Rauh, 2005; Nakamura, 2009), the tendency is to select avatars that depict more aspirational or idealised versions of the self (Bessière, Seay, & Kiesler, 2007). In this sense, avatars can accurately represent some characteristics of the users and inaccurately present other aspects of the self. However, this is not always a choice. Sometimes, technological constraints, social norms, or limitations on avatar options make it difficult to convey one's authentic identity (Nowak & Fox, 2018).
Alternatively, some users select different avatars with the intention of exploring different identities and see how it feels to be an ‘other’ (Bessière, Seay, & Kiesler, 2007). While identity exploration via avatars can contribute to gaining identity self-acceptance, to building relationships, or to self-disclose to others, it can also have negative outcomes, such as reinforcing negative stereotypes (Nakamura, 2013). As shown in some of the experiments described during the development of this thesis and in line with the Proteus effect (Yee & Bailenson, 2007), avatar embodiment enacts stereotype-consistent behaviours, i.e., the behaviour of the user conforms to the representation of the self, regardless of the real self. The avatar's characteristics may influence the user's online and offline communication. Being embodied by an attractive avatar leads to closer approaches by the partner's avatars and in the disclosure of more personal information, while being embodied by tall avatars elicits more confident and aggressive negotiating behaviours (Yee & Bailenson, 2007). Recent work by Peck, Good, and Bourne (2020) brings more evidence forward that gender stereotypes not only are effective in VR settings regarding the perceptions of others, but they also affect the behaviour and performances of those embodying an avatar and can even lead to the nullification of the – usually very robust effects of –stereotype threat. Female participants with female avatars showed working memory impairment under stereotype threat (which did not occur in the female/female avatar group). Interestingly, female participants embodying a male avatar did not show this memory impairment under stereotype threat, either. These findings suggest that using a male avatar was able to protect the female participants from the negative effects of stereotype threat, thus replicating a previous study. They found more evidence of the Proteus effect, as even men with a female avatar showed impairment in the stereotype threat condition, whereas men with a female avatar under no threat did not. Additionally, men's performance within male avatars suffered in the ‘no threat conditions’, in line with assumptions about the workings of positive stereotypes, i.e., ‘stereotype lift’ (Walton & Cohen, 2003), which in this case, was not activated. However, as Palomares and Lee (2010) demonstrated, users tend to adopt a gender-typical language that suits their avatar's gender. These findings suggest that communicative outcomes of avatar selection are based not only on receiver's impression but also on sender's experience.
To summarise, avatar appearances have an impact on self-perception, attitudes and behaviours, as well as on its perceptions by others. As mentioned, choosing an avatar that masks one's gender is a common practice for women in order to prevent harassment in video-gaming and virtual communities (Ducheneaut et al., 2006; Cooper, 2007; Williams, et al., 2009). Moreover, people of colour sometimes prefer not to disclose their race or ethnicity to avoid being stereotyped, discriminated, or harassed (Yee & Bailenson, 2007). But, as discussed above, limiting the visibility of targeted groups, such as women or people of colour, as if they were not present in virtual environments has more downsides than benefits.
Encounters with other people – whether face to face or in VR – are perceived as social interactions. People use similar cognitive processes to perceive virtual representations of a person, so avatars in VR can evoke stereotypes in a similar way that a bodily representation of a person does (Nowak & Fox, 2018). Virtual others that are represented as women are therefore subject to gendered perception and can trigger sexist reactions (Behm-Morawitz & Mastro, 2009). Gender stereotypes imply an association between men and leadership and women and follower (Braun et al., 2017). We therefore have the following hypothesis for our setting in which a male and a female avatar meet and have to agree on who takes the role of the manager and who takes the role of the team member:
For the understanding of gender stereotypes and bias, we refer to the SCM, a socio-psychological theory that postulates that all group stereotypes and interpersonal impressions are formed along two dimensions: (1) warmth and (2) competence (Cuddy et al., 2008). The warmth dimension would aim to assess a stranger's intent to either harm or help, while the competence dimension seeks to judge a stranger's capacity to act on a perceived intention. Stereotypically, men are perceived as (and expected to be) more competent, while women are perceived as (and expected to be) warmer (Cuddy et al., 2008).
In line with the SCM, we expected a generally perceived gender stereotypicality: We expect, considering the ratings of women and men on a general level, the stereotypical pattern of high ratings for females’ warmth and higher ratings for males’ competence to emerge. Regarding self-perception, we expect positive self-ratings on both dimensions, warmth and competence (in-group favouritism).
Research aimed at exploring how the physical characteristics of an avatar relate to the perception of the avatar has shown that appearance plays an important role in how similar to themselves people perceive the avatar to be. In online games, players who are able to identify with their avatars have a more positive experience and the actual physical similarity results in better performance with a task (Wauck et al., 2018). At the same time, it has been shown that the virtual self-representation has an influence on the behaviour and the cognitions of users, i.e., people behave in line with the way their avatar would stereotypically be expected to behave. This phenomenon has been called stereotype lift (Lee et al., 2014) or the Proteus effect (Yee & Bailenson, 2007). Research done specifically to analyse the effects of gendered stereotypes shows that independent of their gender, users tend to conform to female-typed behaviour when embodied in a female avatar (Palomares & Lee, 2010). We therefore hypothesise as follows:
In order to test our hypotheses, we ran a quasi-experimental study. The study took place in two VR laboratories on two different campuses of a university in Germany. From the two locations, the participants scheduled in the same time slot would meet in a virtual room and work on a task together. The task was to explore how suitable the VR room that they found themselves in was for a performance review talk between a manager and a member of staff. For this, one of them was to take the role of the manager, the other one was to take the role of the team member and explore the room from that perspective. The room itself was designed like a modern business-style flat, featuring several rooms with big glass windows, design chairs and other furniture. This cover story was used to explain why the participants had to agree among themselves who would be the manager and who the member of staff. The participants were randomly assigned a male or a female avatar, i.e., participants were not allowed to choose the gender of their avatars but were allowed to choose the roles they wanted to play.
The experiment was conducted with students of the Bachelor's programme in Software Engineering at a University in the year 2018. The students participated in the study for extra credit. There were
The study featured a 2 (avatar gender: male vs. female) × 2 (performance role: manager vs. member of staff) design. Dependent variables were as follows: (a) the role (manager or member of staff) assigned to the female or the male avatar; (b) evaluation of gender stereotypes based on the SCM; (c) enjoyment of the VR experience; and (d) happiness with the avatar; additionally, (e) evaluation of the suitability of VR for performance review was measured.
At the beginning of the session, participants were shown the system and the head-mounted VR headset HTC Vive was put on the participant's head. They were made familiar with the functions of the controllers and were provided with an accommodation time of 2 minutes, during which users could test the controls and ask questions. Following this, the audio headset was put on and switched on. The maximum time that participants were in the VR room was set to 10 minutes. After exiting the VR, participants completed a questionnaire administered on a tablet computer using standard survey software (keyingress).
Based on the SCM, using the items by Cuddy and Fiske (2002), three variables were combined to determine warmth: ‘likeable’, ‘good-natured’ and ‘warm’; and three variables were included to evaluate competence: ‘competent’, ‘competitive’ and ‘independent’. Participants were asked to rate the other participant that they were negotiating roles and exploring the room with on these dimensions. Responses were recorded on a seven-point Likert scale anchored between ‘1’ (disagree) and ‘7’ (agree).
The effect of variables avatar gender and adopted role on happiness with the given avatar was examined (‘I was happy with the avatar I had’). Responses were also recorded on a seven-point Likert scale anchored between ‘1’ (disagree) and ‘7’ (agree).
The participant's engagement, enjoyment and immersion were assessed with 14 questions. Enjoyment was understood to be a feeling of pleasure or contentment during the VR experience (Witmer & Singer, 1998). The questions were based on Makransky's multimodal presence scale for VR environments (Makransky, Lilleholt, & Aaby, 2017), with the following additional items: ‘I enjoyed myself during the virtual reality experience.’; ‘I felt sad when the virtual reality experience was over.’; ‘I would like to repeat the experience I just had.’; and ‘The virtual reality experience was interesting.’ Responses were recorded on a seven-point Likert scale anchored between ‘1’ (disagree) and ‘7’ (agree).
Data were analysed using SPSS. Outlying rates, determined by stem-and-leaf plots, were replaced with the highest or lowest extreme rates in the data set (Tabachnick & Fidell, 2013). This outlier exclusion did not alter the reported results.
A two-way 2 (avatar gender: male vs. female) × 2 (performance role: manager vs. member of staff) analysis of variances (ANOVAs) and chi-square tests were used for analysis. The dependent variables were as follows: (a) the role (manager or member of staff) assigned to the female or the male avatar; (b) evaluation of gender stereotypes based on the SCM; (c) enjoyment of VR experience; and (d) happiness with the avatar.
Positive effect sizes denote that the results are consistent with experimental predictions, whereas negative effect sizes denote that they are contrary to predictions.
According to Hypothesis 1, and in line with stereotypical social beliefs regarding gender and roles, we expected that the manager role would be mostly played by the male avatar and the team member role would be mostly played by the female avatar. Figure 1 shows the frequencies of the adopted roles in relation to avatar gender: The differences were not significant (χ2=0.201,
There were no significant differences between the groups; so Hypothesis 2 was rejected, i.e., there was no link between the gender of the other avatar and perceived warmth or competence traits as the ANOVA showed no significant effect of the gender of the avatar on warmth or competence ratings in our sample, WELCH <0.247,
Regarding differences between genders and confirming the predictions of the SCM, the ANOVA of the within-subject factor ‘Competence’ showed that women (
Considering the warmth dimension, women (
In testing Hypotheses 3 and 4, the effect of the variables avatar gender and adopted role on happiness with the given avatar was examined. The two-way 2 (avatar gender: male vs. female) × 2 (performance role: manager vs. member of staff) ANOVA of the variable ‘happiness with avatar’ showed that no main effects emerged, but the interaction between role and avatar gender was significant (
For those who had the manager role, the difference between male and female avatars was not significant (
Regarding avatar gender, participants who were represented by the male avatar showed differences in happiness when assuming different roles (
The ANOVA of the variable ‘Sadness when the VR experience was over’ with between-subject factors ‘avatar gender’ and ‘role’ showed a significant effect of avatar gender:
The ANOVA of the variable ‘the environment was beautiful’ with between-subject factors ‘avatar gender’ and ‘role’ showed a significant effect of avatar gender:
The ANOVA of the variable ‘I was comfortable in the virtual environment’ with between-subject factors ‘avatar gender’ and ‘role’ showed a significant effect of role:
The ANOVA of the variable ‘The virtual reality experience was immersive’ with between-subject factors ‘avatar gender’ and ‘role’ yielded a significant effect of ‘role’:
The ANOVA of the variable ‘The virtual reality experience was exciting’ with between-subject factors ‘avatar gender’ and ‘role’ yielded a significant effect of ‘role’:
The ANOVA of the variable ‘I was completely captivated by the virtual world‘ with between-subject factors ‘avatar gender’ and ‘role’ yielded a (marginal) significant effect of role:
The ANOVA of the variable ‘my experience in the virtual environment was pleasant’ with between-subject factors ‘avatar gender’ and ‘role’ showed a (marginally) significant effect of ‘role’:
The ANOVAs of the other variables from Makransky's multimodal presence scale for VR environments regarding embodiment, sense of presence and ease of figuring things out in the VR environment, as well as the other additional items (‘wish of repeating the experience’, ‘the VR experience was interesting’, and ‘I enjoyed myself during the experience’), with between-subject factors ‘avatar gender’ and ‘role’, did not yield any significant effects.
In this study, participants were randomly assigned a male or female avatar and were instructed to negotiate a role – manager or member of staff. The analysis examined the influence that avatar gender may have on role negotiation with variables based on the SCM (Cuddy et al., 2008), on the theories of the Proteus effect (Ducheneaut et al., 2006), as well as on other experiments conducted in the fields of avatar self-similarity (Wauck et al., 2018). Our study results did not find any significant effect of the gender of the avatar on role negotiation, nor did the gender of the other person's avatar have an effect on either the warmth or the competence ratings of the other. As expected, when looking at the scores of self-perception, people rated themselves highly in both dimensions, but when rating others, women (as a social group) received higher scores regarding their warmth, but lower competence ratings than men, and vice versa.
Concerning the satisfaction relating to roles and avatar gender, the results showed significant interaction between gender of the avatar and role. Contrary to our hypotheses, participants with a staff role and a male avatar were considerably more pleased with their avatar than female avatars in the same role. Participants in managerial roles, however, were equally content with their roles, regardless of their avatar gender. Furthermore, on the level of avatar gender, participants embodied in a male avatar were also happier with their avatar when adopting a staff role instead of a manager role, while role did not affect participants with female avatars in regard to their happiness. Furthermore, participants who were provided with the male avatar and who had assumed the member of staff role were the happiest with the avatar they had. They were more content with their avatars than participants with male avatars in the manager role and participants with female avatars in the staff role.
Avatar gender and role affected the evaluation of the VR experience in various ways. Participants in the role of a ‘member of staff’ reported more enjoyment regarding the VR experience, being more comfortable in the VR environment, rating the experience as more exciting and more immersive and pleasant, and being more captivated by the virtual world than participants in the manager role.
Again, in reference to avatar gender, participants with male avatars reported more sadness when the VR experience had concluded and rated the VR environment as more ‘beautiful’ than participants with a female avatar.
These findings can be interpreted with regard to perceived psychological or visual similarity to the self and its influences on perception and attribution in interactive and communicative contexts (Wauck et al., 2018). Similar to outcomes in face-to-face contexts, in which people prefer to maintain relationships with those individuals they perceive to share any kind of social identity with (Slater et al., 2013), in avatar-based VR, people also tend to select self-similar avatars and prefer to interact with other avatars they identify as similar to them (Nowak & Rauh, 2005; Nowak et al., 2009; Nowak & Fox, 2018). The fact that, in our experiment, the majority of the participants were male students may explain the greater enjoyment of the VR experience when having the male avatar and member-of-staff role. This interpretation is further supported by the evaluations of the experience. Due to the self-similarity, participants in the staff role could identify more with their avatar and thus, experienced more immersion and excitement and were more captivated by the VR than the other group. This may also be supported by the findings regarding male avatars, who were associated with being more ‘comfortable’ in the virtual world.
The study holds limitations due to the small size of the sample, which may weaken statistical validity in terms of findings. In addition, the sample presented disparity on participants’ real genders, as – in line with the gender distribution in the computer science student population – the majority consisted of male participants. As the participants were assigned to the groups at random, it does not skew the results; it does – however – lower the ecological validity and we cannot yet draw conclusions of gender differences, or the lack thereof, regarding male and female participants. Effects of gender stereotypes on one's behaviour might also have been enhanced if the participants had had more time to become acquainted with their avatar – they only had an initial 10 minutes to familiarise themselves with the setting and the avatar.
Another limitation is that some of the participants knew each other and, hence, their real genders, while others in the group did not know each other, which may have limited the effect of the avatar's gender on the stereotypical perception of another person. Even if they did not know each other, the voice link gave cues about their partner's gender, which might override the effect of the avatar. Alternatively, being with a friend or classmate in the VR Room may cause participants to feel more relaxed and less focussed on the task, negatively affecting performance and engagement. All of the participants were undergraduates of software engineering and already familiar with VR, and the fact that they were participating for extra academic credits might have also influenced participants’ enjoyment in the VR experience and satisfaction with the given avatar.
Keeping in mind that people conform to stereotypes based on what they see and that, in VR, avatars might serve as the primary identity cue, it can be concluded that people will be stereotyped based on the characteristics of their virtual self-representations (Lee et al., 2014) without asking who is the person hidden behind the avatar.
In this sense, being represented with socially acceptable or desirable characteristics might help to thrive in the digital world. However, the creation of ‘ideal selves’ in avatar customisation may have a negative impact on gender equality, as well as the integration of other disadvantaged or marginalised groups based on disabilities, homosexuality, skin colour, etc. The creation of ‘ideal selves’, which – in the beginning – may aim to avoid negative gender-related stereotypes and prejudices also prevalent in the offline world, can result in the long term in a poorer integration (and even higher discrimination) toward those groups that might not correspond to the white–young–attractive–competent–men considered the societal reference par excellence.
In the Introduction section, the terms ‘glass ceiling, glass walls, glass slipper and glass cliff’, which represent the obstacles that women face in the workplace, were briefly explained, and therein originated the many cultural prescriptions for leadership identity and behaviour that many women found unattainable (Ibarra & Petriglieri, 2016). When assuming new leadership positions or more senior roles, people have to engage in a new image and identity (Roberst, 2005) which conforms to role-related expectations (Ibarra & Petriglieri, 2016). Role transitions are often followed by feelings of anxiety, which can be a symptom of identity threat and are often followed by a gender-based stereotype threat. This stereotype threat stems from cultural beliefs that are associated with the traditional masculine leader image, creating an additional identity threat to women. Women have to bridge the gap between current and desired identities as senior professionals, and concerns about identity are challenged simply due to the fact of being women (Ibarra & Petriglieri, 2016).
When relating these strategies of the SCM (Cuddy et al., 2008), protective self-presentation styles would be aligned with warmth traits, and acquisitive strategies are more likely linked to competency features. In an attempt to avoid disapproval, women tend to behave kind and friendly, trying to gain respect by showing their capability, technical competency, as well as integrity and sincerity. In doing so, they need to master a balancing act between being seen as too feminine to be competent and too masculine to be likable (Williams & Multhaup, 2018). In contrast, men will tend to make use of their intelligence and confidence to build positive first impressions or to take more aggressive stands in order to achieve their goals. Our results reaffirm that warmth traits are more related to relationship-oriented leadership styles, while competence traits are more linked to task-oriented styles.
Even when men and women possess comparable analytic skills, the choice of protective self-presentation is consequential for women. It has been found that the ability to win confidence may facilitate career advancement, while the use of protective self-presentation confirms gender stereotypes that can lead to career stagnation (Ibarra & Petriglieri, 2016). Moreover, the majority of organisational hierarchies are dominated by men, providing few role models for women and perpetuating intrinsic beliefs that leadership behaviours are more appropriate and common in men (Eagly et al., 2007).
Role modelling involves cognitive processes for the understanding of which specific self-presentation behaviours are effective, why they are effective and also the affective processes that determine whether the specific behaviour will produce positive consequences (Bandura, 1986). In this sense, self-presentation enables one to create, maintain, or modify a public self that is usually consistent with their ideal self (Baumeister, 1982). Career success relies importantly upon self-presentation, whereby the success factors generally include the ability to generate novel ideas, represent the firm, sell new business and develop relationships with clients (Ibarra & Petriglieri, 2016).
This paper has explored the extent to which stereotypes might be present in VR, but further research will be needed to analyse women's behaviour in VR in relation to role modelling, self-presentation strategies, leadership positions, or negotiation techniques. Would women also adopt protective self-presentation strategies in VR? Would women construct ideal selves through avatar customisation? In the present era of digital transformation, are we demanding a creation of what we term ‘women's impossible selves’? Could avatar-based VR help at role modelling and self-presentation allowing us to overcome such demands? Or is it actually harder for women to thrive in VR environments due to the negative stereotypes associated with women and technology? These are intriguing questions to propose for further exploration in the domain of gender differences in immersive VR. In its applicability to the business field, these questions might be of interest for the design of human resources practices and strategies that advocate for the inclusion of diverse workforce and promote gender equality within companies and organisations.
This study was designed assuming that avatar-based VR will play a growing role in remote learning and working. We also believe that the characteristics of current immersive VR displays still need to be developed in order to be sufficiently usable for educational as well as business purposes. Slater and Sanchez-Vives (2016) state that ‘…an ideal form of a shared VR would require real-time full facial capture, eye tracking, real-time rendering of subtle emotional changes such as blushing and sweating, subtle facial muscle movements such as almost imperceptible eyebrow-raising, the possibility of physical contact such as the ability to shake hands, or embrace, or even push and so on. Such a system does not exist today, though it is one to strive for’ (Slater & Sanchez-Vives, 2016). Considering that more human-like avatars can be perceived, it is more likely that theories of human communication will apply to avatar-based interactions. Future work in VR should be conducted to investigate how technological improvements might affect human behaviour.
Aside from the pertinent improvements in apparatus, other technical issues should be involved in how to realise such a system (Steed & Oliveira, 2009). Questions such as how and where to distribute the computation systems – in a distributed network or a setup in which one master machine broadcasts to all the others; or how to keep the different participant environments synchronised with one another so that they all will be able to perceive the same consistent environment needed for consideration are among these issues. With the use of immersive technologies in virtual business teams, a redesign of work practices and workplaces might be necessary (Steed & Oliveira, 2009).
This paper has aimed to provide insights into understanding the extent to which gender stereotypes are present in avatar-based VR, inviting a discussion and reflection about the possible impact that could result from using VR for business and educational purposes. Correspondingly, the possibilities offered by immersive VR to see whether this tool could benefit the domain of co-presence, facilitating remote communication and collaboration has been explored. As simulated in the experiment conducted, the virtual environment was a shared space among multiple participants, where each participant was represented by an avatar observing and interacting with the representations of the others. In the experiment, participants were randomly assigned a male or female avatar and were instructed to negotiate a role – as manager or member of staff. The analysis focussed on the possible influence that avatar gender could have on role negotiation with variables based on the SCM (Cuddy et al., 2008), on the theories of the Proteus effect (Yee, Bailenson, & Ducheneaut, 2009), as well as on other experiments conducted in the fields of avatar self-similarity (Wauck et al., 2018). For the empirical research, quantitative methods were used. A two-way 2 (avatar gender: male vs. female) × 2 (performance role: manager vs. member of staff) ANOVA and chi-square tests and statistics were used for the analysis of the effects of gender on the dependent variables. Findings indicate that the explicit gender of the avatar did not appear to have a significant detrimental impact on role negotiation. Additionally, avatars were not significantly affected by gender-related stereotypes. The data did reveal significant differences in enjoyment and satisfaction regarding roles and avatar genders; participants who were arbitrarily represented by male avatars and assumed the role of member of staff showed greater satisfaction. This pattern would point to the importance of self-similarity and social identity theories. These results contribute to theoretical and practical implications for social identity dynamics in avatar-based virtual environments.
This paper contributes to the literature and understanding of avatar-based VR and provides insights into understanding the extent to which gender stereotypes are present in avatar-based VR. The importance of this cannot be understated because the design, connectivity and functionality of new technologies, such as VR environments, will undoubtedly play a more significant role in the complex global post-coronavirus disease (COVID) context for both business and education.