Page 1
Published in Small Group Research (2011), 42(4): 405-457
2
Review of Virtual Environment Studies on Social and Group Phenomena
Anu Sivunen
BIT Research Centre Aalto University School of Science and Technology Finland
[email protected]
Marko Hakonen
BIT Research Centre Aalto University School of Science and Technology Finland
Abstract
This article provides a review of previously published studies on virtual environments (VEs),
focusing especially on empirical articles on social and group phenomena in VEs and their
methodological and theoretical trends. Virtual environments can be defined as
communication systems in which interactants share the same three-dimensional digital space
and can navigate, manipulate objects, and interact with one another via avatars. When
examining the methodological and theoretical choices of these studies, four trends could be
identified that characterize group studies on virtual environments: 1) testing the applicability
of real-life, social behavior norms in VEs, 2) a lack of work group studies using VEs, 3) the
micro-level treatment of social and groups and 4) a lack of covering theory. Finally,
propositions for future research are presented.
Keywords: computer-mediated communication, social psychology, virtual environment,
virtual group, virtual world
Page 2
Published in Small Group Research (2011), 42(4): 405-457
2
Review of Virtual Environment Studies on Social and Group Phenomena
Virtual environments (VEs), such as Second Life and Entropia Universe, have
gained massive popularity during the past decade (e.g. Gillath, McCall, Shaver, Blascovich,
2008). Millions of users interact daily in 3D VEs, which have been developed to resemble the
real world more and more and simultaneously provide possibilities to surpass real-life
obstacles (e.g., flying, teleporting). Virtual environments can be defined as communication
systems in which multiple interactants share the same three-dimensional digital space despite
occupying remote physical locations and can navigate, manipulate objects, and interact with
one another via avatars (Sallnäs, 2005; Yee & Bailenson, 2007). Avatars are flexible and
easily transformed digital self-representations in a graphic 3D form (Yee & Bailenson, 2007).
In addition to 3D worlds developed primarily for massive multi-user online game-
playing, there are also environments in the market that have no specific focus on particular
activities but that provide their users with a modifiable environment for collaboration. This
relatively new forum for human interaction has inspired scholars to study the social aspects of
VEs. In their seminal article, Blascovich et al. (2002) suggest that VEs provide an intriguing
new platform for empirical research in social psychology. They argue that the limitations of
real life studies such as the trade-off between realism and control, the impossibility of perfect
replications and the use of non-representative samples, especially in experimental studies, can
be avoided in VEs. Current VE technology enables realistic set-ups which are nevertheless
controllable, in which the conditions can be simply copied over time, and the population
interacting in VEs is highly representative. These points underline that the research potential
of VEs is enormous.
VE research has taken many directions, from technology-driven approaches to
individual psychology to studies of educational possibilities in VEs. Studies have been
Page 3
Published in Small Group Research (2011), 42(4): 405-457
3
conducted on themes such as presence and co-presence felt in VEs (Axelsson, Abelin, Heldal,
Schroeder & Wideström, 2001; Hoyt, Blascowich & Swinth, 2003; Slater, Sadagic, Usoh &
Schroeder, 2000), the verbal and nonverbal communication of the avatars operating in them
(Bailenson & Yee, 2006), the effects of avatar appearance on individual perceptions (Yee &
Bailenson, 2007), and the effects of different communication media (voice, text) on virtual
environments (Williams, Caplan & Xion, 2007). Despite the growing interest in the area,
little is known on the costs and benefits of using these environments for group work,
especially for organizational groups and teams.
The goal of this article is to integrate the findings of studies of virtual environments
and to identify directions for future research. We do this by examining peer-reviewed articles
published in academic journals, concentrating especially on studies that cover social
(psychological) and group phenomena in VEs. However, our review is not a conventional one
in that, rather than categorizing the articles studied according to their contents, we focus more
on methodological and theoretical trends to be found in them. The reason for this is that
research on virtual environments is not yet very robust, and conclusive classifications based
on study results are still difficult to make. Instead, our aim is to review articles in a
constructive manner and find the opportunities and challenges for further research within this
rather young field of study.
One of the main currents in VE studies deals with massively-multiplayer online role-
playing games (see e.g., Yee, 2006). We do not fully exclude game-playing studies, but due
to our interest in broadening the scope of research towards the potential of VEs for work
groups, the fully anonymous, fun-motivated game-playing VEs are not in the center of our
scrutiny.
Despite the burgeoning literature on virtual environments, there are no
comprehensive reviews on the use of VEs from a social scientific point of view. Our article
Page 4
Published in Small Group Research (2011), 42(4): 405-457
4
tries to fill this gap by providing a thorough exploration of empirical articles published in the
area. The main contribution of our review arises from the scrutiny of VE studies on social
and group phenomena within the context of four trends we observed in the literature
reviewed. Here we must highlight two points: we do not wish by any means to downplay
current VE research, and we recognize that steps have already been taken in the directions we
propose later. However, in the light of our review, these four trends prove to be rather
dominant and raise many questions and propositions for future research. First, many studies
assume or try to demonstrate that findings from real-life studies and “laws of social behavior”
apply also in VEs. Notwithstanding the importance of this endeavor, this trend raises a need
to study VE-specific dynamics in order to understand the possibly unique nature of these new
platforms of social action. Second, the lack of attention to VEs as potential platforms for
work groups, especially for virtual teams, prompted us to discuss the importance of this
notion and to give examples of how to broaden VE research to this realm. Third, it seems that
the dominant approach to social phenomena in current research is that researchers focus on
micro-level social phenomena (e.g., proximity, gaze, etc.), sometimes at the expense of
broader group-level phenomena like social identification with, or leadership in, groups.
Fourth, following from the last point, we argue that this concentration on interesting
phenomena has often overshadowed the efforts to bridge the research and findings with
covering and well-established (even dominant) theoretical approaches in social sciences.
The structure of this review is as follows. In section two we present the criteria and
methods of our literature review. Section three is devoted to supporting the argument on the
four trends observed and illustrating them with examples from the studies reviewed. In
section four, we suggest how to go further from each dominant trend and build propositions
for further VE research into social and group phenomena. In the last section, we summarize
our findings and discuss the conclusions of our review.
Page 5
Published in Small Group Research (2011), 42(4): 405-457
5
Methodology
The studies presented in this review were identified by various means. First, we
searched the main electronic databases (e.g. Ebsco, PsycINFO, Sage, JSTOR, Emerald) for
studies in communication, social psychology, business, and information systems using an
extensive list of relevant terms (e.g. “virtual environment and collaboration”, “virtual world
and communication”). From the articles found, we selected only the peer-reviewed empirical
studies. Next, we scanned through several leading journals in psychology, communication,
information systems, and management using similar key words. Third, we manually scanned
the reference lists of the articles identified by the first two methods. This procedure provided
us with 67 articles, of which 47 were selected for the final review (see Appendix). Articles
that were left out from the final review were exclusively game-, education- or clinical
psychology-related or focused only on individual, not social, phenomena (see a review on
e.g. clinical psychology and training-related VE studies on Fox, Arena & Bailenson, 2009).
The results of the review are presented in the following section.
Trends Identified in the Literature
Appendix presents the 47 articles studied in this review, including a short summary
of their methods and results. From these 47 articles, we identified four different trends which
characterize studies on virtual environments: 1) testing the applicability of real-life, social
behavior norms in VEs, 2) a lack of work group studies using VEs, 3) an emphasis on micro-
level treatment of social and groups and 4) a lack of covering theory. Table 1 shows how we
categorized each of the 47 articles in terms of these four trends.
Page 6
Published in Small Group Research (2011), 42(4): 405-457
6
Table 1 Classification of the Reviewed Articles According to the Four Trends Outlined Above
Authors Year Trend 1 Trend 2 Trend 3 Trend 4 Antonijevic 2008 X Axelsson, Abelin, Heldal, Schroeder & Wideström 2001 X X Axelsson, Abelin & Schroeder 2003 X Bailenson, Beall & Blascovich 2002 X X X X Bailenson, Beall, Loomis, Blascovich & Turk 2005 X X X Bailenson, Blascovich, Beall & Loomis 2003 X X X X Bailenson, Blascovich & Guadagno 2008 X X Bailenson & Yee 2005 X X X X Bailenson & Yee 2006 X X X Bailenson, Yee, Merget & Schroeder 2006 X X Bente, Rüggenberg, Krämer & Eschenburg 2008 X X Diehl & Prins 2008 X Dotsch & Wigboldus 2008 X X X Feldon & Kafai 2008 X X Fiedler 2009 X X Fiedler & Haruvy 2009 X X X Fox & Bailenson 2009 X X Friedman, Garmiel & Dinur 2009 X X X Gillath, McCall, Shaver & Blascovich 2008 X X X Gong & Nass 2007 X X Guadagno, Blascovich, Bailenson & McCall 2007 X X X Hindmarsh, Heath & Fraser 2006 X X Hoyt, Aguiular, Kaiser, Blascovich & Lee 2007 X X Hoyt & Blascovich 2003 X X Hoyt & Blascovich 2007 X X Hoyt, Blascovich & Swinth 2003 X X X Jackson, Zhao, Qui, Kolenic, Fitzgerald, Harold & Von Eye 2008 X X Jung & Lee 2009 X X X Lantz 2001 X Linebarger, Janneck & Kessler 2005 X X X Nakanishi 2004 X X X Nowak 2004 X X Nowak & Biocca 2003 X X Nowak & Rauh 2005 X X Nowak & Rauh 2008 X X Peña, Hancock & Merola 2009 X X X Sallnäs 2005 X X Scheck, Allmendinger & Hamann 2008 X X Slater, Sadagic, Usoh & Schroeder 2000 X X Talamo & Ligorio 2001 X Vasalou & Joinson 2009 X X X Vasalou, Joinson, Bänzinger, Goldie & Pitt 2008 X Vinayagamoorthy, Steed & Slater 2008 X X X X Yee & Bailenson 2007 X X X Yee & Bailenson 2009 X X Yee, Bailenson & Duchenaut 2009 X X X Yee, Bailenson, Urbanek, Chang & Merget 2007 X X X X
We estimated the accuracy of our categorization of the articles by conducting an
inter-rater agreement test. After our own coding we recruited two researchers who had not
Page 7
Published in Small Group Research (2011), 42(4): 405-457
7
been involved in the writing of this article. We gave them both all 47 articles, explained our
four trends and their inclusion/exclusion criteria and asked them to read and code the articles
independently. We used our own ratings as a third “coder” in the test. We had to count the
inter-rater agreement for each trend separately because the trends are not mutually exclusive.
However, by coding each article dichotomously as either belonging to the trend in question
(value = 1) or not belonging to it (value = 0) we were able to create a table for the calculation
of three-rater, two-category Fleiss’s Kappa for each trend (Fleiss, 1971). The Kappas were
.60 (SE = .08) for trend one, 1.0 (SE = .08) for trend two, .71 (SE = .08) for trend three .75
(SE = .08) for trend four. This indicated moderate agreement for trend one, perfect one for
trend two and strong agreement for trends three and four. The strongest disagreements
pertained to the trend one which our external raters noted to be sometimes slightly
ambiguous. Trend two was, as expected, totally unambiguous due to its rather factual nature.
Finally, the rather small disagreements in coding were negotiated between us and the two
external raters and they resulted in no changes. Overall, the inter-rater agreement was very
satisfactory. We trust that these agreement rates are sufficient for the purposes of categorizing
articles into trends which is our way of pointing out interesting lines in current VE research –
not finding clear-cut facts as often is the case when many coders are used.
Trend One: Testing the Applicability of Real-Life, Social Behavior Norms in VEs
As virtual environments make it possible to study social phenomena that would be
difficult or even impossible to study in real life, several scholars have made an effort to
construct complex set-ups and conduct laboratory studies to find out whether social norms
and behavior transfer from real life to VEs. Sixteen studies out of 47 explored the transfer of
known real-life social norms to a virtual environment. These norms include, for example,
interpersonal distance and eye contact (Yee, Bailenson, Urbanek, Chang & Merget, 2007),
Page 8
Published in Small Group Research (2011), 42(4): 405-457
8
interpersonal distance and discrimination (Dotsch & Wigboldus 2008), the presence of others
and its impact on task performance (Hoyt et al., 2003), and reactions to persons in need
(Gillath et al., 2008). This trend does not seem to be very strong based on the number of
articles categorized to it. Had we included all the articles that referred to real life-based
theories, studies or phenomena, almost all reviewed articles would have been included in this
category. Instead, we decided to include only those articles which clearly tested whether
some phenomenon found in real life (e.g., interpersonal distance) applied also to VE. Hence,
for example, most articles on avatar characteristics were not included even though their
theory sections included references to real-life research.
In the study of Yee and his colleagues (2007), the authors found that in a virtual
environment, a decrease in interpersonal distance was compensated for by avoiding eye-
contact with the interacting partner. In Dotsch and Wiboldus’s study (2008), subjects
maintained a larger personal distance in a VE with avatars that belonged to a different ethnic
background and looked different from their own avatars (and themselves in real life). A
questionnaire also revealed the subjects’ implicit prejudice against this same ethnic minority.
In a classic replication of the impact of the presence of others on the performance of
individuals, Hoyt and co-authors (2003) showed that social inhibition effects also occur in
VEs. Participants performed a novel task significantly worse when other avatars were
observing their performance than when performing the same task alone. Furthermore, basic
emotions, such as empathy towards persons in need, can be triggered in the same vein in
experimental set-ups in VE’s as in real life (Gillath et al., 2008).
Despite the eagerness to test whether social laws persist in VEs, there are also
several studies that have explored phenomena that would be very difficult or even impossible
to test in real life, such as avatar androgyny (Nowak & Rauh, 2008) or interaction with one’s
photographic self-representation (Bailenson et al., 2008). A good example is the seminal
Page 9
Published in Small Group Research (2011), 42(4): 405-457
9
study of Yee and Bailenson (2007) on the appearances of avatars and their effects on social
behavior. The influence of facial attractiveness and height showed that looks matter in social
relationships. Participants interacting in a VE with attractive-looking avatars walked closer to
the confederate and revealed more information than unattractively represented avatars.
Moreover, participants with tall avatars more often split money in their own favor than
participants with short avatars, who were more likely to accept an unfair offer than
participants in normal and tall condition. These phenomena occurred independently of the
real height and appearance of the participants.
In conclusion, scholars have put much effort into proving that several social
phenomena familiar from real environments also exist in virtual environments. Our need for
personal space, gaze transfer and perceptions of persons of specific appearance in real life
seem to persist when we interact as avatars in a virtual world. However, on the basis of the
studies reviewed, we do not know much about new types of behavior or about the
characteristics that would be specific to virtual world interaction, as the studies have focused
less on the new phenomena occurring in these environments.
Trend Two: The Lack of Work Group Studies Using VEs
A massive trend in the articles reviewed was the use of student subjects and the lack
of groups that had a common history or background, such as organizational teams. Thirty-six
out of 47 studies used student or pupil participants, and five studies reported that their
subjects had been recruited from a university campus, making the total number of studies
using university or school-related subjects as high as 41.
Only one of the 47 articles reviewed studied organizational work groups. This study,
conducted by Lantz (2001), observed the implementation of a virtual environment for a
workgroup that had four geographically distributed members and had earlier been using only
Page 10
Published in Small Group Research (2011), 42(4): 405-457
10
face-to-face meetings. The study found that meetings held in a VE were considered to be
more task-oriented than face-to-face meetings. It was also discovered that graphic avatars
played an important role in turn-taking and making the meetings more enjoyable in the virtual
environment.
As virtual environments can be seen as new collaboration tools or workspaces for
geographically dispersed work teams and organizations, the absence of empirical, peer-
reviewed studies on VE use in such contexts is rather surprising. The development of VEs,
such as Second Life, provides many opportunities, especially for geographically distributed
work groups. Such groups are also known as virtual teams, that is, groups of people striving
toward a common goal, dispersed in many locations, and communicating with each other
predominantly via information and communication technology (e.g., Axtell, Fleck, & Turner,
2004). VEs can be seen as a new media for virtual teams that struggle with communication
problems over distance (see Hertel, Geister, & Konradt, 2005; Martins, Gibson, & Maynard,
2004 for reviews of virtual team literature). The tools available even in many desktop-based
VEs, such as chat, voice and whiteboards, not to mention the avatars that can point, express
emotions and use tools, are powerful competitors to more traditional means of remote
interaction through expensive and static videoconferencing systems or simple but
asynchronous e-mails.
As working life is becoming more and more global and as companies are
simultaneously striving to minimize their travel costs and to find new ways of collaboration,
we attribute great importance to the research on VEs as potential media for distributed
organizational teams. Virtual environments can also provide a more inspiring experience for
collaborators than standard practices by enabling the team members to do several
unconventional activities together in a 3D environment, such as flying, diving or sitting
around a camp fire. Such activities might serve as team building functions and further
Page 11
Published in Small Group Research (2011), 42(4): 405-457
11
strengthen the participants’ identification with their virtual team (Sivunen, 2006).
Furthermore, meetings in VEs can be more enjoyable as the team members can flexibly
change and modify the meeting environment as well as their avatars’ graphic appearance.
Trend Three: The Micro-Level Treatment of Social and Group Phenomena
We classified 33 out of 47 articles as representing a trend in which the scope of
social phenomena could be characterized to be at micro- or individual level rather than at
macro- or group level. Our argument is based on two notions. First and most importantly, all
of these 33 articles studied some social tendency (e.g., interpersonal distance) of individuals
or some more VE-specific social construct (e.g., avatar appearance) of individual behavior or
attitudes in the presence of one or two others, thus representing a somewhat limited view of
groups. Second, the fact that 31 out of these 33 studies were conducted in laboratory settings
gave indirect support to our claim as to the micro-social trend because these set-ups are rarely
used in the study of larger groups and macro-level social or group phenomena. Nevertheless,
our identification of a trend does not imply that these studies may not be relevant or of high
quality. For example, many studies (e.g., on communication modalities) ingeniously utilized
the specific features and conditions of VEs to broaden the scope and findings of traditional
computer-mediated communication (CMC) literature (e.g. Daft & Lengel 1984; Rice &
Gattiker, 2001). Nevertheless, the trend reveals the need for more studies of broader-scale
social phenomena (e.g. leadership or social identity) occurring in real and larger groups than
dyads within VEs.
The two examples of this mainstream trend and one exception described below
illustrate the difference between micro and macro levels or individual- and group-centered
approaches. An example of the micro-level scrutiny of social behavior is the excellent study
on interpersonal distance and eye gaze transfer in VEs (Bailenson, Blasovich, Beall &
Page 12
Published in Small Group Research (2011), 42(4): 405-457
12
Loomis, 2003). In this laboratory experiment, a total of 160 undergraduate students
participated under two experimental conditions. Using fine-grained and impressive data
collection techniques, including head-mounted displays (HMDs) and tracking devices, the
researchers found that participants, through their avatars, were inclined to keep interpersonal
distance in VE (as in real life) and that they compensated for shorter interpersonal distance by
avoiding looking at the other avatar or agent.
Another example of this trend is a study that examined the reliance on visual
characteristics in online avatar perception and especially the relationships between avatar
androgyny, anthropomorphism and credibility (Nowak & Rauh, 2008). In a two-step research
process, the first sample of university student participants selected eight maximally differing
avatars in terms of the above-mentioned dimensions, which were used for the step 2
experiment. In the second step, a different set of student participants rated their text chat
partners, represented visually by the previously-chosen face images, based on the three
dimensions. Path analysis revealed that perceived avatar androgyny negatively predicted
perceived avatar anthropomorphism, which in turn positively predicted perceptions of partner
credibility.
A counter-trend example is the study by Talamo and Ligorio (2001) on the
construction of identities in a VE using ethnographic and conversational methodology. The
task of the multicultural participant group (N = 48) was to build a community in a VE called
Active Worlds in eight months. Two types of data were collected: 1) open responses to a
questionnaire, 2) recorded chat. It was found that identity construction was heavily context-
dependent and identities were not static but negotiated in interactions. Many strategies for
playing with different identities and making them distinct were observed – most commonly
that of trying different avatars. Here the scope was behavior in a rather large group. The set-
Page 13
Published in Small Group Research (2011), 42(4): 405-457
13
up did not limit the observations to any particular level (individual or group) and the
longitudinal approach ensured rich data and variations in behavior over time.
All three examples are well-designed and rigorously reported; in other words, they
represent sophisticated academic research. However, the given examples of the micro-social
trend clearly focus on individual reactions to rather limited social stimuli. For example, in
Nowak and Rauh’s (2008) study, the stimuli were the visual cues provided by static pictures
of individual faces. The micro-level, individual-centered treatment of social phenomena in
these and the other 31 studies showing a similar trend simply raises the question of how we
could be better informed about larger-scale, group-level dynamics in VEs. The counter-
example presented above gives some clues as to how to enrich research on social and group
phenomena in VEs. However, we would like to go further and encourage VE scholars to
study, for instance, whether group-level social identification predicts pro-social behavior in
VEs just as it has been shown to do in the real world (e.g., Tyler & Blader, 2001) and in
distributed teams (Hakonen & Lipponen, 2007).
Trend Four: The Lack of Covering Theory
The question of the lack of larger-scale theoretical approaches among the articles
reviewed (18 out of 47) is related to, but not the same as, the previous trend, namely the
concentration on micro-social phenomena. It seems to us that often the concentration of
interest upon an in-depth scrutiny of fairly similar, interesting but rather narrow research
fields (e.g., proximity) tends to lead to theoretical discussion that is more concerned with
previous findings and small-scale theories than with bridging the findings to more covering
and even dominant larger-scale theoretical approaches in social sciences. Again, this is not a
value statement. Deepening research certainly produces valuable accumulating knowledge on
the constructs studied. Furthermore, it must also be noted that different journals have
Page 14
Published in Small Group Research (2011), 42(4): 405-457
14
differing demands and space available for theoretical discussion. Hence, the lack of covering
theory in an article should not be attributed to lack of theoretical understanding on the part of
the authors but rather to the style of particular journals publishing VE research. Still,
interestingly, in their seminal conceptual article, Blascovich and his co-authors (2002) do
consider VEs to be a “methodological tool for social psychology.” The authors consider how
VE might help the study of different social and group phenomena but pay fairly little
attention to how to link this methodological promise to theoretical discourses. Our point here
is that research on social and group phenomena in VEs might benefit and be enriched by
linking future studies to well-established theoretical discussions and paradigms within social
sciences.
The study of cultural differences in morality in real and virtual worlds (Jackson et
al., 2008) is an example of this lack-of-theory trend. The authors cite Kohlberg’s theory of
moral development (1984) and briefly refer to cross-cultural research (e.g., Bond, 1986) in
the introduction. They found that Chinese youth emphasized conventional moral off-line
behaviors but accepted morally questionable on-line behavior more than their US
counterparts. However, while discussing this counterintuitive result showing inconsistent
moral norms on-line and off-line, the authors do not refer to any of the theories or sources
they presented in their introduction. The interesting result is left open to speculation and not
rooted in, for example, discussion on dissonance of social cognitions. In fact, one rather
commonly used theoretical framework was the social cognition research and its different
varieties (e.g., Gong & Nass, 2007; Hoyt et al., 2007). These tend to fall in the mid-range of
our trend: the hypotheses and results are grounded in a well-specified theory, but their
anchoring to broader theoretical approaches is not always very clear. However, the potential
of VEs for social cognition research was noted as early as 2002 by Groom, Sherman and
Conrey. We coded social cognition-based articles as having a covering theory, because social
Page 15
Published in Small Group Research (2011), 42(4): 405-457
15
cognition is a major theoretical field in communication and social psychology. Consequently,
this trend does not seem to be very strong. Had we decided otherwise and judged whether an
article had rooted its social cognition theorizing "well enough" to be merited as having a
covering theoretical basis, the coding criteria would have become too ambiguous.
A good but unfortunately rare example of the type of theory-use we recommend VE
scholars to adopt is exploring of the construction of cultural identity and intercultural literacy
in Second Life (Diehl & Prins, 2008). The authors based their study firmly on Cultural
Historical Activity Theory (e.g., Engestrom, 2001) and Heyward’s model of cultural literacy
(2002). It was found that Second Life interaction increased intercultural literacy and that
avatar appearance was used in cultural identity construction. Throughout the article, the
reasoning and findings are rooted to these two theoretical frameworks. Cultural Historical
Activity Theory may not be the best known tradition in social sciences, but it certainly fulfills
the criteria of a covering (meta)theory that has been used to explain various social
phenomena. Another example comes from the experimental research tradition and goes a
long way in the direction we would recommend: the study by Yee and Bailenson (2007) on
the so called Proteus Effect – that is, the effects of different self-representations (e.g., avatar
height) on behavior. The introduction and theoretical discussion goes deeply into CMC
literature (e.g., Postmes, Spears, & Lea, 1998; Walther, 1996), the hypotheses are firmly built
on well-developed theory, and the results are discussed in terms of the theoretical framework
presented.
Notwithstanding the promising counter-examples to this trend presented above, we
want to go further and suggest a particular, concrete theoretical approach for the study of
social and group phenomena in VEs. The Social Identity Approach (SIA; Tajfel & Turner,
1979) provides a theoretically coherent and covering view of many social psychologically
relevant phenomena, such as conformity, cohesion and leadership. Thus, we see no reason
Page 16
Published in Small Group Research (2011), 42(4): 405-457
16
why the SIA should not be utilized in the study of VEs. More specifically, the SIA has been
found to be useful in understanding social phenomena in work contexts (Haslam, 2004;
Riketta 2005), and as such it provides a theoretical approach to our interests in broadening
VE research to work settings (trend two) and to group-level studies. We elaborate this
argument and present the SIA in detail in the next section.
Agenda for Further Research
In the space of a few years, there has been an increase in research interest in virtual
environments and their effects on our collaboration and communication. As virtual worlds
develop and gain wider popularity among groups and organizations, studies in the area will
probably also increase in the future. Researchers are only now beginning to understand the
effects of virtual environments on groups’ collaboration, and our review suggests that more
work remains to be done to understand the full potential of virtual environments for social
science as well as for distributed groups and organizations. We will now elaborate on the four
trends found in our review and propose directions for further research.
Exploring Virtual Environment-Specific Phenomena
The review shows that there is an accumulating body of knowledge in the area of
studies related to the transfer of real-life social norms to virtual environments. People act
similarly in many ways whether they collaborate with one another in real life or in virtual
worlds through avatars. However, there is a gap in research focusing on virtual environment-
specific features and their effects on group behavior. Therefore, virtual environments still
present great opportunities for empirical social science research.
One virtual environment-specific phenomenon that has not been studied
comprehensively is the development of groups in virtual environments. Studying groups that
Page 17
Published in Small Group Research (2011), 42(4): 405-457
17
collaborate solely in a virtual world for a longer period of time and do not see one another
face-to-face might reveal something new related to the development of roles, leadership and
power relations in such groups.
Another aspect that has been studied in more traditional virtual contexts, such as in
computer-mediated groups and communities, but not in virtual environments, is the effect of
anonymity on group members’ identities, trust and perception of others (see e.g. Jarvenpaa &
Leidner, 1998; Lea & Spears, 1991). It would be interesting to explore what kind of impact
the presentation of others as anonymous, 3D graphical avatars has on the development of
trust and identities between the members of a group working together in a virtual
environment. In addition to experimental studies, research is needed on longitudinal and
naturally occurring group behavior where members build their trust and perceptions of others
on the interactions taking place in a virtual environment.
Furthermore, there are several social occurrences that could be studied as virtual
environment-specific phenomena. The existing research has focused more on replicating the
“laws of real-life social behavior” in VEs, such as whether the presence of avatars can trigger
social inhibition effects (Hoyt et al., 2003) or certain emotions (Gillath et al., 2008).
Fascinating as the results of such studies are, it would be valuable to give attention to virtual
environments as their own type of communication environment and not only as a substitute or
comparison for real-life settings. Instead of comparing VEs to real-life contexts, future
research should focus on the specific social affordances these environments provide for
groups.
One example of a virtual environment-specific phenomenon that can be very
different from real-life experience is self-disclosure and impression management via 3D
avatars. Research on text-based computer-mediated communication shows that mediated self-
presentation differs from unmediated self-disclosure (e.g. Tidwell & Walther, 2002).
Page 18
Published in Small Group Research (2011), 42(4): 405-457
18
However, the way we present ourselves when collaborating via graphic 3D avatars and the
mechanisms by which first impressions are formed from others’ graphic representations may
uncover some unique features that could not be found by studying mediated settings with text
or video-based communication technology – nor real-life face-to-face groups. Studying
anonymous groups collaborating for the first time in a virtual world might illustrate some
social mechanisms that do not exist outside graphic 3D space. The circumstances of such a
collaboration environment may allow or promote certain social, cognitive and communicative
processes that could not occur in other environments (see also Walther, 2007).
Studying virtual environments as a collaboration space for distributed teams
One of the findings of our review that has direct practical implications is the
surprising lack of study of work groups using VEs as a collaboration space. Only one of the
articles reviewed studied a geographically distributed team that used virtual world for
common meetings (Lantz, 2001). It is understandable that obtaining data from field settings
can be very difficult, especially as many organizations do not use virtual environments yet in
their daily activities. However, it is clear that distributed teams and organizations would
benefit greatly from using virtual environments for performing their real-life, organizational
tasks and that studying such naturally-occurring teams would advance the literature through
asking and answering questions that cannot be sufficiently tested in laboratory settings. Some
interesting questions for future research include: What are the implications of collaborating in
a virtual environment for the productivity of distributed teams? Can virtual environments be
efficiently used for organizational training or team-building activities? How do virtual
environments affect the distributed team members’ identification with their team?
Studying the use of virtual environments in organizational settings would also
provide new insights for inter-group and inter-organizational collaboration research. Thus far,
Page 19
Published in Small Group Research (2011), 42(4): 405-457
19
micro-social behavior occurring in isolated laboratory groups has received a fair amount of
attention in VE studies (see also trend three). If VE research were more oriented toward
conducting field studies on distributed teams and organizations, the relations and networks of
the groups and organizations under study could better be taken into consideration. As virtual
worlds, such as Second Life, consist of public and private collaboration spaces, these
environments can provide better possibilities for inter-organizational collaboration than the
closed private collaboration tools existing today. For social science research, this would give
a tremendous opportunity to study inter-group as well as inter-organizational collaboration in
naturally occurring contexts.
Finally, virtual environments provide new methods for studying distributed teams.
Thus far, studies of distributed teams have tended to use either survey or interview methods
to collect data for team members’ perceptions and attitudes (see e.g. Chidambaram, 1996;
Mortensen & Hinds, 2001; Walther & Bunz, 2005). Other methods, such as observation, have
been used much more rarely due to the geographical distance between the team members.
When studying distributed teams that use VEs for collaboration, observing the members’
nonverbal and verbal behavior through avatars is easier, as the members and the researcher
can share the same virtual collaboration space, which can be accessed through the internet all
over the world. Moreover, team members’ textual communication in VE is saved
automatically by the system regardless of where they are or at what time they work. This
gives richer opportunities for data collection than traditional survey and interview methods.
Toward a Broader Social Scope
The articles reviewed do include studies of groups in VEs that broadened the usual
micro-social trend. Many of these used qualitative methodologies, like ethnography. In
addition to the above-mentioned study by Talamo and Ligorio (2001), another encouraging
Page 20
Published in Small Group Research (2011), 42(4): 405-457
20
example is Antonijevic’s (2008) study of naturally-occurring nonverbal communication,
which covers larger groups than dyads. Such approaches allow researchers to observe larger
groups than is usually possible in laboratory settings. Moreover, qualitative methods lend
themselves well to the study of naturally occurring groups in VEs, and since VEs are a
relatively new area of scientific research, it would be advisable (see e.g. Denzin & Lincoln,
1994 on the benefits of qualitative research) to use qualitative and exploratory set-ups to find
out whether there is anything unique or novel in these new contexts of human interaction (see
also trend one). This call for broadening the methodological spectrum by no means
downplays the importance of laboratory studies. An intriguing example of socially broader
scale laboratory studies is that of Hoyt and Blascovich (2003) comparing groups under
transactional and transformational leaders, measured in different communication modalities
(face-to-face, audio and IVE). This study also demonstrates that the methodology is not
necessarily an obstacle if the aim is to extend the framing of the phenomena studied to group
level. This study could have been conducted, for example, using individual subjects instead
of triads, but this would have excluded the interaction aspect of “subordinates”. This, in turn,
would have made the measurement of, for example, cohesion rather awkward if not
impossible, since cohesion is by nature a group phenomenon (e.g., Hogg, 1993).
The above-mentioned examples also underline the importance of researchers’ own
interests in shaping the nature of studies. If the community of VE scholars becomes more
interested in broader-level phenomena, such as inter-group relations in VEs, the social scope
will broaden in a natural way. The environment or methods are not the obstacles. For
instance, the study of prejudice by Dotsch and Wigboldus (2008) could quite easily have
been reframed from an individual level of scrutiny and explanation to a more social level.
The study found that white student participants kept a greater distance from Moroccan-
Page 21
Published in Small Group Research (2011), 42(4): 405-457
21
looking avatars than white-looking avatars. This avoidance was explained by participants’
implicit prejudices and skin conductance level. An alternative way of studying prejudice
could have included, for example, a longitudinal observation of interaction of avatar groups
representing two different ethnic groups. Repeated measurements of in-group bias and the
scope of group identification could have informed the researchers about the development and
dynamics of prejudice rather differently from the actual study. Again, the point of the
example is by no means to downplay the high quality of the study in question but to show
that changing the scope and framing might have yielded different interpretations of the same
very interesting phenomenon. Given the dominant trend of micro-level studies of social and
group phenomena, we suggest that broader conceptualizations might provide more variety to
the domain of VE studies of this type. This might also attract a larger community of social
scientists to engage in discussion and research on the subject of groups in this new
environment.
Use of Larger Scale Theoretical Approaches – The Social Identity Approach as an
Example
One of the dominant theoretical “umbrellas” in social psychology that has developed
in recent decades is the Social Identity Approach (SIA). We argue that it could be a useful
(meta)theoretical lens for the study of social and group phenomena in the context of VEs.
The SIA provides a theoretical framework for the relationship between individual
and group. Specifically, it consists of two distinct theories: the original social identity theory
(e.g., Tajfel & Turner, 1979), and the more recent self-categorization theory (Turner, Hogg,
Oakes, Reicher, & Wetherell, 1987). Despite certain differences, both theories share the same
fundamental assumption that individuals define themselves in terms of their social group
memberships and that group-defined self-perception produces distinctive effects on social
Page 22
Published in Small Group Research (2011), 42(4): 405-457
22
behavior and inter-group relations (Hogg & Abrams, 1988; Turner, 1999). This means that
the more an individual conceives of him- or herself in terms of membership in a group (i.e.,
identifies with the group), the more his or her attitudes and behavior are governed by this
group membership (Hogg & Abrams, 1988; Van Knippenberg & van Schie, 2000).
During the past ten years, social identity principles have been increasingly applied to
the study of organizational psychological processes (e.g., Haslam, 2004; Hogg & Terry,
2001). In this context, organizational or team membership is understood to reflect on the self-
conception in the same way as other social memberships do (Ashforth & Mael, 1989; Hogg
& Terry, 2001). Thus, organizational identification is defined as “the perception of oneness
with or belonging to a group” (Ashforth & Mael, 1989, p. 34). Moreover, it is proposed that
this group-based self-conception leads to activities that are congruent with this identity.
Accordingly, empirical studies have shown that group identification is linked to various
important outcomes, such as high levels of extra-role behaviors (e.g., Tyler & Blader, 2001;
see Riketta, 2005, for a review).
The work context orientation of the SIA also means that it could be usefully applied
to remedying the lack of VE studies of collaboration in work settings as underlined above
(trend two). Recent studies of geographically distributed teams have shown the applicability
of SIA to studies of this organizational form. For example, it has been found that fairness
perceptions are important antecedents of virtual team-level identification and that this social
identification, in turn, leads to advancement of group goals beyond work-role descriptions
(Hakonen & Koivisto, 2008; Hakonen & Lipponen, 2007), as predicted by the SIA. Scrutiny
of antecedents and consequences of group-level identification in general would certainly be
interesting in VE settings, too.
Insofar as the applicability of these and related findings in different settings has not
been thoroughly examined in the VE context, we would urge VE scholars to use the SIA or
Page 23
Published in Small Group Research (2011), 42(4): 405-457
23
other broad theoretical frameworks when considering further studies of social and group
phenomena in VEs. Covering theoretical framings of interesting group dynamics in VEs
might result in innovative cross-fertilization of ideas and publishing forums and hence
broaden and enrich the discussion within social psychology and across other disciplines.
Conclusions
Our goal in this article has been to present and evaluate the existing research on VEs
related to social and group phenomena. Based on our review, we have identified four trends
that prevail in social science research on these environments. It seems that scholars have
focused more on proving how social norms transfer from real life to virtual worlds than on
trying to identify phenomena specific to these environments. There has been a paucity of
field research using, for example, distributed teams and of studies conducted in
organizational settings. Instead, the study of micro-social behavior and experimental research
using student samples have gained more attention in VE studies. Furthermore, the use of
larger theoretical frameworks has not been as dominant in VE studies as in traditional social
science research.
Our review shows several openings for further research. The study of virtual
environment-specific phenomena, such as certain group behaviors that can be triggered only
in virtual environments, would be a fascinating possibility for social scientists. Field research
conducted in organizational settings with qualitative, longitudinal set-ups would also raise
new possibilities for virtual environment scholars. The framing of research questions
focusing on macro-level phenomena such as leadership and intergroup relations could
broaden the field even further. Finally, utilization of covering (meta)theories of social
sciences, such as the Social Identity Approach, would contribute to the development of
virtual environment research in the future.
Page 24
Published in Small Group Research (2011), 42(4): 405-457
24
An interesting question that this review raises is whether there are any social
phenomena about which VE research could not inform us or that would be extremely difficult
to study in this new environment. Even though this question cannot, naturally, be extensively
answered by speculation, we found a couple of illustrative examples. First, and perhaps most
clearly, it might be very difficult to study the natural development of social interaction
between a parent and his/her (genetic) offspring (e.g., attachment; see Bowlby, 1980) when
the child is very young. One could test parents’ reactions by programming babys’ movements
and voices to a VE, but that hardly informs us about the natural social process, which is
highly emotional and based on physical touch. If this example seems to be slightly off the
scope of our review, our second example of a hard-to-study-in-VE phenomenon is the long-
term development of social coping mechanisms for work-related stress or work-life balance
(see e.g., Theorell & Karasek, 1996). Work, as we have shown, is a much understudied
context in VE research, but that could change over time. What is problematic here is the long-
term process, which involves many actors and complicated processes at workplace and at
home. However, this speculation was a hard exercise, which suggests that VEs can indeed
inform us of a very broad spectrum of social phenomena relevant to human behavior.
Considering the fact that research on social and group phenomena in VEs is a
relatively new endeavor, it is natural that the scope of studies in the field is not as extensive
as in more mature research areas. Consequently, our review has presented four trends in
current VE research. More importantly, we have tried to open new perspectives and research
agendas that go beyond these trends. It is our hope that these openings encourage and inspire
VE researchers in a constructive manner. In the future, as the technology develops further,
VE platforms can be expected to provide even more opportunities for social scientists to
study communication and collaboration dynamics in several different contexts.
Page 25
Published in Small Group Research (2011), 42(4): 405-457
25
References
Antonijevic, S. (2008). From text to gesture online: A microethnographic analysis of
nonverbal communication in the Second Life virtual environment. Information,
Communication & Society, 11(2), 221–238.
Ashforth, B. E., & Mael, F. A. (1989). Social identity theory and the organization. Academy
of Management Review, 14, 20-39.
Axelsson, A.-S., Abelin, A. & Scroeder, R. (2003). Anyone speak Spanish? New Media &
Society, 5(4), 475–498.
Axelsson, A.-S., Abelin, Å, Heldal, I., Schroeder, R. & Wideström, J. (2001). Cubes in the
cube: A comparison of a puzzle-solving task in a virtual and a real environment.
CyberPsychology & Behavior, 4(2), 279–286.
Axtell, C. M., Fleck, S. J. & Turner, N. (2004). Virtual teams: Collaborating across distance.
In C. L. Cooper & I. T. Robertson (Eds.) International Review of Industrial and
Organizational Psychology, vol. 19. Chichester: John Wiley, 205-48.
Bente, G., Rüggenberg, S., Krämer, N. C. & Eschenburg, F. (2008). Avatar-mediated
networking: Increasing social presence and interpersonal trust in net-based
collaborations. Human Communication Research, 34(2), 287-318.
Bailenson, J. N., Beall, A. C. & Blascovich, J. (2002). Gaze and task performance in shared
virtual environments. The Journal of Visualization and Computer Animation, 13(5),
313-320.
Page 26
Published in Small Group Research (2011), 42(4): 405-457
26
Bailenson, J. N., Beall, A. C., Loomis, J., Blascovich, J. & Turk, M. (2005). Transformed
social interaction, augmented gaze, and social influence in immersive virtual
environments. Human Communication Research, 31(4), 511–537.
Bailenson, J. N., Blascovich, J., Beall, A. C. & Loomis, J. M. (2003). Interpersonal distance
in immersive virtual environments. Personality and Social Psychology Bulletin,
29(7), 819–833.
Bailenson, J. N., Blascovich, J., Guadagno, R. E. (2008). Self-representations in immersive
virtual environments. Journal of Applied Social Psychology, 38(11), 2673–2690.
Bailenson, J. N. & Yee, N. (2005). Digital chameleons: Automatic assimilation of nonverbal
gestures in immersive virtual environments. Psychological Science, 16(10), 814-819.
Bailenson, J. N. & Yee, N. (2006). A longitudinal study of task performance, head
movements, subjective report, simulator sickness, and transformed social interaction
in collaborative virtual environments. Presence, 15(6), 699–716.
Bailenson, J. N., Yee, N., Merget, D., Schroeder, 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 Interaction.
Presence, 15 (4), 359–372.
Blasovich, J., Loomis, J. M., Beall, A. C., Swinth, K. R., Hoyt, C. L. & Bailenson, J. N.
(2002). Immersive virtual environment technology as a methodological tool for social
psychology. Psychological Inquiry, 13(2), 103-124.
Bond, M. H. (1986). Lifting one of the last bamboo curtains: review of the psychology of
Chinese people. Hong Kong: Oxford University Press.
Page 27
Published in Small Group Research (2011), 42(4): 405-457
27
Bowlby, J. (1980). Attachment and loss: Loss, sadness, and depression. Basic Books: New
York.
Chidambaram, L. (1996). Relational development in computer supported groups [1]. MIS
Quarterly, 20(2), 143–165.
Daft, R. L. & Lengel, R. H. (1984). Information richness: A new approach to managerial
information processing and organization design. In L. L. Cummings & B. M. Staw
(Eds.) Research in organizational behaviour, Vol. 6. Greenwich: JAI Press, 191–234.
Denzin, N. K. & Lincoln, Y. S. (1994). Introduction: Entering the field of qualitative
research. In N. K. Denzin & Y. S. Lincoln (Eds.) Handbook of Qualitative Research.
London: Sage, 1–17.
Diehl, W. C. & Prins, E. (2008). Unintended outcomes in Second Life: Intercultural literacy
and cultural identity in a virtual world. Language & Intercultural Communication,
8(2), 101–118.
Dotsch, R. & Wigboldus, D. H. J. (2008). Virtual prejudice. Journal of Experimental Social
Psychology, 44(4), 1194–1198.
Engestrom, Y. (2001). Expansive learning at work: Toward an activity theory
reconceptualization. Journal of Education at Work, 14(1), 133-156.
Feldon, D. F. & Kafai, Y. B. (2008). Mixed methods for mixed reality: Understanding users’
avatar activities in virtual worlds. Educational Technology Research and
Development, 56, 575–593.
Fiedler, M. (2009). Cooperation in Virtual Worlds. Schmalenbach Business Review, April
2009, 173-194.
Page 28
Published in Small Group Research (2011), 42(4): 405-457
28
Fiedler, M. & Haruvy, E. (2009). The lab versus the virtual lab and virtual field—An
experimental investigation of trust games with communication. Journal of Economic
Behavior & Organization, 72, 716–724.
Fleiss, J. L. (1971). Measuring nominal scale agreement among many raters. Psychological
Bulletin, 76, 378–382
Fox, J., Arena, D. & Bailenson, J. N. (2009). Virtual reality: A survival guide for the social
scientist. Journal of Media Psychology, 21(3), 95–113.
Fox, J. & Bailenson, J. N. (2009). Virtual virgins and vamps: The effects of exposure to
female characters’ sexualized appearance and gaze in an immersive virtual
environment. Sex Roles, 61(3–4), 147–157.
Friedman, D., Karniel, Y. & Dinur, A. L. (2009). Comparing group discussion in virtual and
physical environments. Presence, 18(4), 286–293.
Gillath, O., McCall, C., Shaver, P. & Blascovich, J. (2008). What can virtual reality teach us
about prosocial tendencies in real and virtual environments? Media Psychology,
11(2), 259–282.
Gong , L. & Nass, C. (2007). When a talking-face computer agent is half-human and half-
humanoid: Human identity and consistency preference. Human Communication
Research, 33(2), 163-193.
Groom, C. J., Sherman, J. W. & Conrey, F. R. (2002). What Immersive Virtual Environment
Technology Can Offer to Social Cognition. Psychological Inquiry, 13, 125–145.
Guadagno, R. E., Blascovich, J., Bailenson, J. N. & McCall, C. (2007). Virtual humans and
persuasion: The effects of agency and behavioral realism. Media Psychology, 10(1),
1–22.
Page 29
Published in Small Group Research (2011), 42(4): 405-457
29
Hakonen, M. & Lipponen, J. (2007). Antecedents and consequences of identification with
virtual teams: Structural characteristics and justice concerns. The Journal of E-
working, 1, 137-153.
Hakonen, M. & Koivisto, S. (2008). Antecedents of identity in virtual teams. Nordic
Organization Studies, 2/08, 54-73.
Haslam, S. A. (2004). Psychology in organizations: The social identity approach. (2nd
edition) London: Sage.
Heyward, M. (2002). From international to intercultural: Redefining the international school
for a globalized world. Journal of Research in International Education, 1(1), 9-32.
Hertel, G., Geister, S. & Konradt, U. (2005). Managing virtual teams: A review of current
empirical research. Human Resource Management Review, 15(1), 69–95.
Hogg, M. A. (1993). Group cohesiveness: A critical review and some new directions.
European Review of Social Psychology, 4, 85-111.
Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology of intergroup
relations and group processes. London: Routledge.
Hogg, M. A., & Terry, D. J. (2001). Social identity theory and organizational processes. In
M. A. Hogg & D. J. Terry (Eds.), Social identity processes in organizational contexts
(pp. 1-12). Philadelphia, PA: Psychology press.
Hoyt, C. A., Aguiular, L., Kaiser, C. R., Blascovich, J., & Lee, K. (2007). The self-protective
and undermining effects of attributional ambiguity. Journal of Experimental Social
Psychology, 43, 884–893.
Hoyt, C. L. & Blascovich, J. (2003). Transformational and Transactional Leadership in Virtul
and Physical Environments. Small Group Research, 34(6), 678–715.
Page 30
Published in Small Group Research (2011), 42(4): 405-457
30
Hoyt, C. L. & Blascovich, J. (2007). Leadership efficacy and women leaders' responses to
Stereotype Activation. Group Processes & Intergroup Relations, 10(4), 595–616.
Hoyt, C. L., Blascovich, J. & Swinth, K. R. (2003). Social inhibition in immersive virtual
environments. Presence, 12(2), 183–195.
Jackson, L. A., Zhao, Y., Kiu, W., Kolenic, A., Fitzgerald, H. E., Harold, R. & Von Eye, A.
(2008). CyberPsychology & Behavior, 11(3), 279–286.
Jung, K. H. & Lee, J.-H. (2009). Implicit and explicit attitude dissociation in spontaneous
deceptive behavior. Acta Psychologica, 132, 62–67.
Jarvenpaa, S. L. & Leidner, D. E. (1998). Communication and trust in global virtual teams.
Journal of Computer-Mediated Communication, 3(4). doi: 10.1111/j.1083-
6101.1998.tb00080.x
Kohlberg, L. (1984). The psychology of moral development. San Francisco: Harper & Row.
Lantz, A. (2001). Meetings in a distributed group of experts: comparing face-to-face, chat
and collaborative virtual environments. Behaviour & Information Technology, 20(2),
111–117.
Lea, M. & Spears, L. (1991). Computer-mediated communication, de-individuation and
group decision making. International Journal of Man-Machine Studies, 34(2), 283–
301.
Linebarger, J. M., Janneck, C. D. & Kessler, G. D. (2005). Leaving the world behind:
Supporting group collaboration patterns in a shared virtual environment for product
design. Presence, 14(6), 697–719.
Martins, L. L., Gilson, L. L. & Maynard, M. T. (2004). Virtual Teams: What do we know and
where do we go from here? Journal of Management, 30(6), 805–835.
Page 31
Published in Small Group Research (2011), 42(4): 405-457
31
Mortensen, M. & Hinds, P. (2001). Conflict and shared identity in geographically distributed
teams. The International Journal of Conflict Management, 12(3), 212–238.
Nakanishi, H. (2004). Free Walk: A social interaction platform for group behaviour in a
virtual space. International Journal of Human-Computer Studies, 60(4), 421–454.
Nowak, K. L. (2004). The Influence of Anthropomorphism and Agency on Social Judgment
in Virtual Environments. Journal of Computer-Mediated Communication, 9. doi:
10.1111/j.1083-6101.2004.tb00284.
Nowak, K. L. & Biocca, F. (2003). The effect of the agency and anthropomorphism on users’
sense of telepresence, copresence, and social presence in virtual environments.
Presence, 12(5), 481–494.
Nowak, K. L., and Rauh, C. (2005). The influence of the avatar on online perceptions
of anthropomorphism, androgyny, credibility, homophily, and attraction. Journal of
Computer-Mediated Communication, 11(1). doi: 10.1111/j.1083-
6101.2006.tb00308.x
Nowak, K. L & Rauh, C. (2008). Choose your 'buddy icon' carefully: The influence of avatar
androgyny, anthropomorphism and credibility in online interactions. Computers in
Human Behavior, 24(4), 1473-1493.
Peña, J., Hancock, J. T., & Merola, N. A. (2009). The priming effects of avatars in virtual
settings. Communication Research, 36, 838-856.
Postmes, T., Spears, R., & Lea, M. (1998). Breaching or building social boundaries?
Communication Research, 25, 689–699.
Page 32
Published in Small Group Research (2011), 42(4): 405-457
32
Rice, R. E. & Gattiker, U. E. (2001). New media and organizational structuring. In F. M.
Jablin & L. L. Putnam (Eds.) The new handbook of organizational communication.
London: Sage, 544–581.
Riketta, M. (2004). Organizational identification: A meta-analysis. Journal of Vocational
Behavior, 66(2), 358–384.
Sallnäs, E.-L. (2005). Effects of Communication Mode on Social Presence, Virtual Presence,
and Performance in Collaborative Virtual Environments. Presence, 14(4), 434–449.
Scheck, S., Allmendinger, K. & Hamann, K. (2008). The effects of media richness on
multilateral negotiations in a collaborative virtual environment. Journal of Media
Psychology, 20(2), 57–66.
Sivunen, A. (2006). Strengthening identification with the team in virtual teams: The leaders’
perspective. Group Decision and Negotiation, 15, 345–366.
Slater, M., Sadagic, A., Usoh, M. & Schroeder, R. (2000). Small-Group Behavior in a Virtual
and Real Environment: A Comparative Study. Presence, 9(1), 37–51.
Tajfel, H. & Turner, J. (1979). An integrative theory of intergroup conflict. In W. G. Austin
& S. Worchel (Eds.) The social psychology of intergroup relations. Monterey:
Brooks/Cole, 33–47.
Talamo, A. & Ligorio, B. (2001). Strategic identities in cyberspace. CyberPsychology &
Behavior, 4(1), 109–122.
Theorell, T., & Karasek, R.A.. (1996). Current issues relating to psychosocial job strain and
cardiovascular disease. Journal of Occupational Health Psychology, 1, 9-26.
Page 33
Published in Small Group Research (2011), 42(4): 405-457
33
Tidwell, L. C., & Walther, J. B. (2002). Computer-mediated communication effects on
disclosure, impressions, and interpersonal evaluations: Getting to know one another a
bit at a time. Human Communication Research, 28(3), 317-348.
Turner, J. C. (1999). Some current issues in research on social identity and self-categorization
theories. In N. Ellemers, R. Spears, & B. Doosje (Eds.), Social identity: Context,
commitment, content (pp. 6-34). Oxford, UK: Blackwell.
Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987).
Rediscovering the social group: A self-categorization theory. Oxford, UK: Blackwell.
Tyler, T. R., & Blader, S. L. (2001). Identity and cooperative behavior in groups. Group
Processes and Intergroup Relations, 4, 207-226.
Van Knippenberg, D., & van Schie, E. C. M. (2000). Foci and correlates of organizational
identification. Journal of Occupational and Organizational Psychology, 73,137-147.
Vasalou, A., Joinson, A. N. (2009). Me, myself and I: The role of interactional context on
self-presentation through avatars. Computers in Human Behavior, 25, 510–520.
Vasalou, A., Joinson, A., Bänziger, T., Goldie, P. & Pitt, J. (2008). Avatars in social media:
Balancing accuracy, playfulness and embodied messages. International Journal of
Human-Computer Studies, 66 (11), 801–811.
Vinayagamoorthy, V., Steed, A. & Slater, M. (2008). The impact of a character posture
model on the communication of affect in an immersive virtual environment. IEEE
Transactions on Visualization and Computer Graphics, 14(5), 965–981.
Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and
hyperpersonal interaction. Communication Research, 23(1), 3–43.
Page 34
Published in Small Group Research (2011), 42(4): 405-457
34
Walther, J. B. (2007). Selective self-presentation in computer-mediated communication:
Hyperpersonal dimensions of technology, language, and cognition. Computers in
Human Behvior, 23(5), 2538–2557.
Walther, J. B. & Bunz, U. (2005). The rules of virtual groups: Trust, liking, and performance
in computer-mediated communication. Journal of Communication, 55(4), 828–846.
Yee, N. (2006) The demographics, motivations, and derived experiences of users of
massively multi-user online graphical environments. Presence, 15(3), 309–329.
Yee, N. & Bailenson, J. (2007). The Proteus Effect: The effect of transformed self-
representation on behavior. Human Communication Research, 33(3), 271-290.
Yee, N. & Bailenson, J. N. (2009). The difference between being and seeing: The relative
contribution of self-perception and priming to behavioral changes via digital self-
representation. Media Psychology, 12, 195–209.
Yee, N., Bailenson, J. N. & Ducheneaut, N. (2009). The proteus effect: Implications of
transformed digital self-representation on online and offline behavior.
Communication Research, 36, 285–312.
Yee, N., Bailenson, J., Urbanek, M., Chang, F. & Merget, D. (2007). The unbearable likeness
of being digital: The persistence of nonverbal social norms in online virtual
environments. CyberPsychology & Behavior, 10(1) 115-121.
Page 35
Published in Small Group Research (2011), 42(4): 405-457 Appendix
Studies included in the review Authors Year Aim / Research
question Methods Subjects / Sample Main findings Journal
Antonijevic 2008 Nonverbal
communication in
a virtual
environment.
Six month-long, ethnographic field
study in the virtual world Second
Life. Data were gathered from 108
publicly available areas by
observing and recording the
naturally-occurring interaction in
both dyadic and group settings.
Approximately 1000
users of Second Life.
Nonverbal communication in
Second Life can be divided into
four categories of nonverbal cues:
1) user-defined, 2) predefined, 3)
blended, and 4) missing cues.
User-defined cues show
adaptability to the setting, whereas
predefined (system generated) and
blended cues often represent
stereotypical, culture- and gender-
biased nonverbal acts.
Information,
Communication &
Society
Axelsson, Abelin,
Heldal, Schroeder,
Wideström
2001 Comparison of
pairs in two
situations: some try
to carry out a task
in a virtual
In the VE setting, one participant
was collaborating on an immersive
system and the other using a
desktop system. Their task was to
collaborate on a Rubik’s cube-type
44 participants in 22
dyads.
In the virtual setting, only 6 pairs
out of 22 completed the task within
the time limit, whereas in the real
setting, all pairs could complete the
task in the given time. Participants
CyberPsychology
& Behavior
Page 36
Published in Small Group Research (2011), 42(4): 405-457
environment and
the others in a real
environment.
puzzle. In the real setting,
participants collaborated face-to-
face with blocks of similar size as
in the VE. A time limit of 20
minutes was set for each
experimental run, first in VE and
then in the real world. After the
experiments, participants answered
questions on presence, co-
presence, their contribution to the
task, and collaboration.
reported a significantly stronger
feeling of presence in the
immersive VE than the desktop
system. Both participants
perceived the person in the
immersive VE to be more active
than the participant at the desktop.
In the real-world setting, there was
no difference between the
participants in perceived
contribution to the task.
Participants in the real-world
setting reported stronger feelings
of collaboration than participants
in VE setting.
Axelsson, Abelin
& Schroeder
2003 How people using
different languages
interact in a virtual
Participant observation of the
virtual world Active Worlds,
consisting of 50 hours of field
Users of a virtual
world Active Worlds.
The phases of a language
encounter situation are
introduction, responses and
New Media &
Society
Page 37
Published in Small Group Research (2011), 42(4): 405-457
environment. observations and notes as well as
chat logs. The observations were
made in the central and most
populated places of the
environment, in language-specific
areas, and in themed areas.
consequences. The reasons for the
introduction phase are 1) finding
out whether there are any fellow
speakers present, 2) having a
language game, or 3) disturbing the
ongoing conversation. The
responses to a language encounter
situation are either acceptance,
rejection, neutral or mixed. The
direct consequences of a language
encounter are that the language
introduced either survives or
disappears. Indirect consequences
are shifts in the atmosphere or in
topics of conversation.
Bailenson, Beall,
Blascovich
2002 Gaze and task
performance in a
virtual
Experimental study in which three
participants sat in physically
remote rooms with HMDs, entered
27 undergraduate
students.
Participants felt involved in the
game, truly interacting with the
other partners, and enjoyed the
The Journal of
Visualization and
Computer
Page 38
Published in Small Group Research (2011), 42(4): 405-457
environment. a common virtual room, and
played games of 20 questions.
Avatar behavior was measured
under three different conditions: 1)
low behavior condition, where
there was no avatar but participants
collaborated through voice, 2)
medium behavior condition, where
each participant could see others’
avatars and hear their voices, and
3) high behavior condition, where
participants’ head movements were
rendered. Each participant played
three separate games of 20
questions, one under each of the
three conditions. The three games
played under each condition
resulted in nine games in total.
experience more under the high
behavior condition than the other
conditions. Most head movements
(in a horizontal plane) occurred
under the high behavior condition.
The highest percentage of speaking
occurred under the low behavior
condition.
Animation
Page 39
Published in Small Group Research (2011), 42(4): 405-457
After each game, participants took
off their HMDs and responded to a
questionnaire.
Bailenson, Beall,
Loomis,
Blascovich & Turk
2005 Social influence of
gaze.
Experimental study in which a
presenter read a persuasive passage
to two participants using HMDs
with three gaze conditions: natural,
augmented or reduced. In the
augmented condition, the
presenter’s avatar directed his/her
gaze toward each of two
participants 100% of the time,
whereas in the reduced condition,
the presenter’s avatar gazed down
100% of the time. Both gaze
conditions were implemented by
transforming the veridical head
movements of the presenter.
72 psychology
students interacting
with one another in
groups of three: two
participants and a
presenter
(confederate).
Women were overall less
persuaded than men as a group.
Agreement in the augmented gaze
condition was significantly higher
than in the other gaze conditions
with female participants but not
with male participants. There was
no significant main effect of gaze
condition on how participants
remembered the persuasive
passage. Participants reported
lower scores of social presence in
the augmented condition than in
the other conditions. Overall,
participants did not detect that the
Human
Communication
Research
Page 40
Published in Small Group Research (2011), 42(4): 405-457
veridicality of the gaze had been
breached.
Bailenson,
Blascovich, Beall
& Loomis
2003 Proxemic behavior
study.
Two experiments using IVE with
HMD: 1) walking participant’s
avatar (distance) approached
virtual human and 2) virtual human
approached standing participant’s
avatar (avoidance). Gaze and
position were tracked.
Altogether 160
undergraduate
students.
Generally, similar distance
behavior was detected as in real
life. In study 1, greater distance
was kept from avatars than agents,
and this was enlarged if avatars
looked at the participants. In study
2, some avoidance of approaching
avatars was found, but less than
expected.
Personality and
Social Psychology
Bulletin
Bailenson,
Blascovich &
Guadagno
2008 How does
interaction differ if
individuals interact
with virtual
representations of
themselves or of
others in an IVE?
Experimental study with 2
(participant gender) X 3 (agent
identity: high-similarity self-
representation vs. low-similarity
self-representation vs. other
representation) factorial design.
Participants in similarity self-
64 university
students.
Participants in the other
representation condition remained
twice as distant from the agent as
participants in high-similarity self-
representation condition.
Additionally, the difference in
minimum distance between the
Journal of Applied
Social Psychology
Page 41
Published in Small Group Research (2011), 42(4): 405-457
representation condition were told
to walk around an animated agent
who represented him- or herself,
and in other representation
condition, they were told to walk
around of a virtual representation
of someone else. In high-similarity
self-representation condition, the
agents bore a photographic
resemblance to the participant,
whereas in low-similarity self-
representation condition, they did
not. In other representation
condition, the agent ostensibly
represented an unfamiliar stranger.
Participants’ location was recorded
during the experiment and they
were administered questions to
other representation condition and
the low-self-representation
condition was significant, but there
were no differences in distance
between the two similarity self-
representation conditions. The
main effect of gender to distance
was not significant. Participants
were more willing to commit
embarrassing acts in front of the
agent in both similarity self-
representation conditions than in
other representation condition but
there was no difference between
the high- and low-similarity self-
representation conditions and no
main effect of gender. Both
similarity self-representation
Page 42
Published in Small Group Research (2011), 42(4): 405-457
gauge liking for the virtual
representation and willingness to
commit embarrassing acts in the
presence of it.
conditions were also preferred to
the other representation condition
in terms of attraction and liking,
but there were no differences
between the two similarity self-
representation conditions.
Bailenson & Yee 2005 Social influence of
mimicry
(chameleon effect).
Two randomized conditions with
IVE and HMD: 1) agent mimicked
the participant; 2) agent acted as
previous participant. In both
conditions, the agent read a
persuasive message.
69 undergraduate
students.
Mimicking agents were perceived
to be more persuasive and rated
more positively than non-
mimickers, even though the
participants were aware that the
agent was a computer-directed
non-human. Moreover, few
participants noticed the mimicking.
Psychological
Science
Bailenson & Yee 2006 Effects of
similarity on task
performance
(different task
Longitudinal laboratory
experiment using collaborative VE
and HMD. Manipulation of facial
and behavioral similarity
Nine undergraduate
university students in
triads.
Performance was best in the facial
similarity condition, followed by
the mimicry and realistic
conditions. The more familiar the
Presence
Page 43
Published in Small Group Research (2011), 42(4): 405-457
types) and other
DVs over time.
(mimicry) of avatars to each other.
In a repeated problem-solving task,
presence, co-presence, entativity (
cohesion) and simulator sickness
were measured.
participants were with the system,
the less they looked at each other
and the less they reported
simulator sickness. Entativity
(cohesion) was high and increased
over time as in FTF interaction.
Bailenson, Yee,
Merget &
Schroeder
2006 How different
communication
modalities affect
verbal and non-
verbal self-
disclosure,
co-presence,
transmission and
identification of
emotions.
Experimental study with three
different conditions:
videoconference,
voice only and
“emotibox” (system that renders
the dimensions of facial
expressions abstractly in terms of
color, shape, and orientation on a
rectangular polygon).
30 university students
in dyads.
Verbal and non-verbal self-
disclosure was lowest in the
videoconference condition.
Nonverbal disclosure was highest
in the voice-only condition. Self-
reported co-presence and success
of transmission and identification
of emotions were lowest in the
emotibox condition. However,
people seemed to be able to
identify certain emotions even if
expressed in abstract fashion
Presence
Page 44
Published in Small Group Research (2011), 42(4): 405-457
(emotibox).
Bente,
Rüggenberg,
Krämer &
Eschenburg
2008 Influence of
different
communication
modalities on
interpersonal trust,
social presence,
communication
quality, nonverbal
behavior and visual
attention.
Laboratory experiment with 71
dyads unknown to each other
beforehand in two separate rooms
using PC interface. Five
conditions: 1) text, 2) audio, 3)
audio-video, 4) low fidelity avatar,
and 5) high fidelity avatar. Each
dyad conducted a job candidate
selection task. Tracking sensors
were used to record body
movements and questionnaires to
measure other DVs.
142 university
students in dyads.
The results indicated that text
mode was inferior to all other
communication modes in
predicting co-presence, emotional
closeness and affect-based trust.
Moreover, video and avatar modes
were rather similar with respect to
nonverbal activity and visual
attention to the other via the
computer’s communication
window.
Human
Communication
Research
Diehl & Prins 2008 Exploration of the
construction of
(individual)
cultural identity
and intercultural
The data consisted of observations
and interviews of Second Life
characters by the first author’s own
avatar via chat. Moreover, 29
habitants completed a survey on
Characters of Second
Life representing a
wide range of
demographic and
social categories.
Drawing on Activity Theory and
Heyward’s model of cultural
literacy, it was found that SL
interaction increased intercultural
literacy (e.g., through the use of
Language and
Intercultural
Communication
Page 45
Published in Small Group Research (2011), 42(4): 405-457
literacy in Second
Life.
their activities in the Second Life. many languages, cross-cultural
friendships and understanding).
Moreover, avatar appearance was
used in cultural identity
construction although only 76% of
the avatars were said to resemble
the real life appearance of those
they represented. Identity shifts
and literacy were seen more as
opportunities provided by the
virtual world than as causes of it.
Dotsch &
Wigboldus
2008 How implicit
prejudice is related
to automatic
proxemic behavior
and to skin
conductance level
(SCL).
Laboratory experiment utilizing
IVE. White Dutch participant
walked individually among 12
avatars (white-Dutch and
Moroccan-Dutch looking). Implicit
prejudice was measured by implicit
association test (IAT) and explicit
33 university
students.
Participants had no explicit
(conscious) prejudices toward
Moroccan-Dutch persons, but IAT
revealed more negative
associations. They kept a greater
personal distance from Moroccan-
looking than white avatars and
Journal of
Experimental
Social Psychology
Page 46
Published in Small Group Research (2011), 42(4): 405-457
prejudice with a questionnaire. their SCLs were higher while near
Moroccan-looking avatars. In
addition, SCL fully mediated the
relationship between implicit
discrimination and distance kept.
Feldon & Kafai 2008 How users engage
with their avatars
in a virtual world
and in real life.
A field study of the virtual world
Whyville. Online tracking data and
surveys were collected from all
participants. In addition, a subset
of the users were observed offline,
and their behavior was recorded
while they were using the virtual
world. These participants were also
interviewed offline.
595 children (8-18
years old) using
virtual world
Whyville. A subset of
the users (88
participants) were
recruited from an
elementary school
science class and
after-school clubs.
Avatar-related behavior (such as
customizing the appearance of the
avatar) was the most commonly
used behavior in the virtual world.
In second place was socializing,
such as sending messages and
chatting. Two primary types of
interactions centered on avatars
who were critiquing or admiring
others’ avatars and donating or
selling face parts to others’ avatars.
Regardless of differences in
participants’ profiles, the
Educational
Technology
Research &
Development
Page 47
Published in Small Group Research (2011), 42(4): 405-457
proportion of online activity
related to avatars did not vary.
Avatar customization was an
important part of this specific
virtual world.
Fiedler 2009 A study on the
effects of
strategically
irrelevant
communication,
media richness,
experience with
the
communication
medium, social
distance,
and collective
orientation on
An experiment with three
randomly assigned dyads was
conducted. The communication
media were a) communication via
Second Life voice, b)
communication via Skype chat,
and c) no communication. In the
trust-game experiment, the
amounts of money sent (proposer)
and returned (responder) were the
dependent variables and proxies
for cooperation.
304 university
students in dyads.
It was found that the collective
orientation predicted cooperation
both in sending and returning
money. The strategically irrelevant
communication had a positive
effect on cooperation only for the
sender since s/he was the only one
expecting reciprocation.
Communication media had no
clear effects on cooperation. Other
empirical results remained
somewhat mixed.
Schmalenbach
Business Review
Page 48
Published in Small Group Research (2011), 42(4): 405-457
cooperation.
Fiedler & Haruvy 2009 A study of
differences
between
laboratory and
field experiment
in trust-‐games.
The effects of
communication
and allocation
motives were also
studied.
The trust-game was the same as
that of Fiedler (2009) above,
except that in all conditions the
game was played in SL. Three
conditions were created: 1) 40
university students played in lab
without prior communication, 2)
136 university students played the
game in lab after 15 minutes of
game-irrelevant discussion, and 3)
216 volunteer SL residents (field
experiment) played the game after
similar discussion as in condition
2). Multiple independent variables
were measured.
176 university
students and 216 SL
residents in dyads.
The main finding was that
interaction through a virtual world
interface (cond. 2 & 3) increases
the amount sent relative to
laboratory results (cond. 1), but
that subjects recruited in the virtual
world (cond.3) give and return less
than the laboratory control group
with the same virtual world
interface (cond. 2). The analyses of
such allocation motives as trust,
cooperative orientation, obligation
and utilitarianism produced mixed
results. Hence, the authors
speculate that different aspects of
laboratory and field conditions
might have triggered the found
Journal of
Economic
Behavior &
Organization
Page 49
Published in Small Group Research (2011), 42(4): 405-457
differences.
Fox & Bailenson 2009 Does the exposure
to a sexualized
dress and amount
of eye gaze of a
female agent in an
immersive virtual
environment affect
sexist and rape-
supportive
attitudes?
A 2x2x2 (Dress x Gaze x Gender)
between-subjects design was
employed. Participants were
placed in an IVE with HMDs. The
female agents were selected
based on a pretest. 30
participants from a separate pool
from the main experiment
viewed 16 female agents and
rated them on several qualities.
Two agents that demonstrated
extreme scores as sexy and
suggestively dressed and two
that ranked as not sexy and
conservatively dressed were
selected for the experiment. In
the high gaze condition, an
83 participants
recruited from a
university.
Participants in the sexualized
condition indicated that the
agent was sexier and more
suggestively dressed than
participants in the nonsexualized
conditions. Moreover,
participants in the high gaze
condition indicated that the
agent saw and acknowledged
them more than participants in
the low gaze condition. There
was no main effect for dress on
rape myth acceptance,
benevolent sexism or hostile
sexism. Those in the suggestively
dressed high
gaze condition (“vamp”) and the
Sex Roles
Page 50
Published in Small Group Research (2011), 42(4): 405-457
algorithm determining the
agent’s head movements
assessed where the participant
was in the room and
appropriately oriented the
agent’s head toward the
participant, whereas in the low
gaze condition, a different
algorithm with random eye gaze
was used. The participants were
told to walk toward the agent in
the IVE and examine it carefully
to feel prepared to answer
questions about the agent’s
appearance and behavior later.
After the experiment,
participants answered two
questionnaires regarding the
conservatively dressed low gaze
condition (“virgin”) exhibited
significantly higher rape myth
acceptance than those in the
suggestively dressed low gaze
and conservatively dressed high
gaze conditions.
Page 51
Published in Small Group Research (2011), 42(4): 405-457
avatar appearance, gaze, sexism
against women and rape myth
acceptance.
Friedman, Garmiel
& Dinur
2009 How group
discussions in FTF
situations and in
VEs differ from
one another in
dynamics and
content.
Participants of four groups carried
out a conversation on same topic,
global warming. Two of the groups
interacted FTF and two of the
groups communicated via chat in
the virtual world Second Life. The
participants were instructed that
they had one hour to reach
consensus whether they were ready
or not to decrease their car usage to
address climate change. Before the
experiment, each subject filled in a
pre-experiment questionnaire and a
short questionnaire assessing their
view of global warming. In the
Four groups of 7–12
student participants,
altogether 36
subjects.
In all groups, there was an increase
in the concern for global warming
after the discussion. The number of
sentences was almost the same in
both conditions, but the sentences
in VE chat were much shorter and
the number of words per minute
smaller than in FTF condition. The
number of on-topic sentences per
speaker, normalized by group size,
was bigger in the FTF than in the
VE chat condition. The most
dominant speaker in the FTF
condition was also dominant in the
number of on-topic sentences, but
Presence
Page 52
Published in Small Group Research (2011), 42(4): 405-457
FTF condition, the participants
were located in a meeting room,
and in the VE condition they were
sitting in the same room using PCs
and collaborating via text chat in
Second Life. Both types of
sessions were recorded and
transcribed. After the experiment,
all participants filled in the same
questionnaire assessing their
opinions on global warming.
this was not the case in VE chat
condition. FTF groups had more
topics discussed in-depth, whereas
VE chat groups had more topics
discussed superficially.
Gillath, McCall,
Shaver &
Blascovich
2008 How participants
react in VE to
people in need.
Two experiments using IVE and
HMD when participants navigated
in VE. Study 1: emotional
reactions towards a blind man in an
accident. Study 2: proxemic
behavior towards a beggar vs. a
businessman on the street.
Altogether 107
university students.
Dispositionally compassionate
people reacted in a more
sympathetic way to the person in
need than the others (36%
emphatic) – similar to results for
real-life studies. Moreover,
dispositionally compassionate
Media Psychology
Page 53
Published in Small Group Research (2011), 42(4): 405-457
participants were more inclined to
look at and stay near the beggar
than the others.
Gong & Nass 2007 How people react
to inconsistencies
between human vs.
humanoid voice
and face and
whether humanoids
(i.e., computer-
generated artificial
agents ) are
considered to be a
different category
from natural
humans.
In two laboratory studies using
“talking faces” shown on a
computer screen, the four different
conditions (human face and voice,
humanoid face and voice =
consistent, human face and
humanoid voice, humanoid face
and human voice = inconsistent)
were tested for males and females.
In study 1, the task was self-
disclosure and in study 2
comprehension of a text read
aloud. The level of trust, negative
attitudes toward the human or
humanoid, used processing time
Altogether 160
undergraduate
university students.
In line with (social cognition)
literature, study 1 showed that
inconsistency of voice and face
(the stimulus seen and heard from
computer) led to lower levels of
trust and longer processing time of
responses. Moreover, female
participants showed more negative
attitudes toward, and were more
sensitive to, inconsistency than
males. Study 2 replicated the
processing time and females’
negative attitude effects regarding
inconsistency. Due to the
comprehension focus of study 2,
Human
Communication
Research
Page 54
Published in Small Group Research (2011), 42(4): 405-457
and in study 2 comprehension of
the message were the DVs
measured.
voice clarity overshadowed the
inconsistency effect, that is, the
human voice was preferred and it
was associated with more positive
judgments of the stimuli than the
humanoid voice, regardless of the
face. Overall, the studies indicated
that humans were perceived to be
qualitatively a different category
from humanoids.
Guadagno,
Blascovich,
Bailenson &
McCall
2007 Attitude change
after listening to a
persuasive
communication by
an in-group
member.
Laboratory experiment using IVE
and HMD. Study 1 investigated
gender and behavioral realism
effects on attitude change after a
persuasive communication. Study
2 extended study 1 by
manipulating the participants’
perceptions of agency.
Altogether 239
undergraduate
university students.
There was a greater attitude change
for same-gender virtual humans
(in-group) than for those of
different gender. Moreover,
behaviorally realistic in-group
(same gender) virtual humans were
perceived to be more influential,
but this applied only to male
Media Psychology
Page 55
Published in Small Group Research (2011), 42(4): 405-457
participants.
Hindmarsh, Heath
& Fraser
2006 How CVEs support
object-focused
collaboration.
Qualitative, ethnomethodological
study including a series of short
trials with novice users in a CVE.
Users’ task was to collaboratively
organize the layout of furniture in a
virtual room. The interaction of the
users was analyzed by
conversation analysis methods.
Six trials of two
participants and two
trials of three
participants (N=18).
Most participants
were students.
Due to the narrow field of view
available and the inflexibility of
avatars, referencing objects and
spaces became a topic in and of
itself within CVEs. The technical
limitations of CVE made it
difficult for participants to assess
what others in the environment
were able to see and how they
engaged in the activities. However,
during the course of trials,
participants became familiar with
some of the problems related to
inflexibility and the narrow view
and learned to manage them
through their talk.
The Sociological
Review
Hoyt, Aguiular, 2007 Attributional Laboratory experiments using Altogether 99 The studies confirmed the Journal of
Page 56
Published in Small Group Research (2011), 42(4): 405-457
Kaiser, Blascovich
& Lee
ambiguity effects
on well-being for
White vs. Latino
(socially
stigmatized
identity) in a
feedback-giving
situation.
IVET and HMD. In study 1,
participants performed an
employee-hiring task as leaders
with two white virtual subordinates
giving feedback. The participants’
virtually represented ethnicity was
varied (known by the participants)
and their well-being and
attributions to discrimination were
measured. Study 2 replicated study
1 but the participants were all
Latinos.
undergraduate
students, half White
and half Latino.
hypotheses, according to which 1)
a real or portrayed stigmatized
ethnicity (i.e. Latino) provoked
attribution of negative feedback to
discrimination and consequent
high-levels of well-being. The
reverse was true for positive
feedback, i.e., the stigma was
perceived to result in overtly
(normatively) positive evaluations
and post-feedback well-being was
thus rated lower. Attributions to
discrimination mediated the
relationship between stigma and
well-being.
Experimental
Social Psychology
Hoyt & Blascovich 2003 Comparisons of
groups under
transactional
Laboratory experiment with 2 X 3
between-subject factorial design
(leadership style: transformational
144 undergraduate
university students in
groups of three
Post-experimental manipulation
check confirmed clear differences
in the perception of leadership
Small Group
Research
Page 57
Published in Small Group Research (2011), 42(4): 405-457
leaders and
transformational
leaders based on
performance,
leadership
satisfaction, group
cohesiveness, trust,
and value
congruence when
operating in face-
to-face or virtual
environments.
or transactional and group setting:
face-to-face, immersive virtual
environment, or intercom [audio-
only]). Participants performed two
tasks collaboratively under a
confederate leader who had been
trained to exhibit behaviors
associated with either the
transactional or
transformational leadership style.
Participants also answered two
questionnaires.
including the
confederate group
leader.
style across leadership style
conditions, and these perceptions
did not differ as a function of
group setting. The followers of
transformational leaders produced
less quantity, greater quality, were
more satisfied, and had greater
cohesiveness than those led by
transactional leaders. Group
members were more satisfied with
their leader when interacting face-
to-face than in the two other
conditions, but there were no
differences in qualitative or
quantitative group performance or
group cohesiveness across settings.
Followers reported higher levels of
trust and value congruence when
Page 58
Published in Small Group Research (2011), 42(4): 405-457
led by a transformational as
opposed to transactional leader.
Trust mediated fully the
relationship between
transformational leadership style,
satisfaction and cohesiveness.
Hoyt & Blascovich 2007 How gender
stereotype
activation on
women leaders is
moderated by low
or high leadership
efficacy and how
stereotype
activation interacts
with perceptions of
leadership
performance, rated
Laboratory experiment using IVE
and HMD with between-subjects 2
X 2 design (leadership efficacy
[high or low] and stereotyped
activation [primed or not]).
Subjects were selected based on a
score they had got from a Self-
Efficacy for Leadership
prescreening. Before the
experiment, subjects in the
stereotype activation condition had
to examine a folder containing
53 female
undergraduate
students (study 1) and
72 female
undergraduate
students (study 2)
The findings show that the
stereotype activation serves to
increase high efficacy leaders’
perceived performance. High
efficacy leaders were also rated to
perform better in the stereotype
activation condition than in the
control condition. Additionally,
high-efficacy leaders evince
heightened identification with the
domain of leadership. Stereotype
activation has also a positive effect
Group Processes &
Intergroup
Relations
Page 59
Published in Small Group Research (2011), 42(4): 405-457
performance,
domain
identification and
well-being
responses.
stereotypical images of male
leaders and information on the
gender gap in top leadership roles,
whereas the subjects in the control
condition had to peruse a folder
containing material on the virtual
reality lab. The subjects ostensibly
participated in a leadership task
with two other participants (who
were in reality automated avatars).
They received instructions
regarding the task and were given
time to prepare for a recorded,
three-minute meeting with the
other ‘participants’. Their task was
to chair a selection committee
hiring a new employee, but the
meeting allowed only one-way
on the psychological well-being
(self-esteem and depressed affect)
of high efficacy leaders. Moreover,
stereotype activation on high- and
low-efficiency leaders’
identification with their domain as
well as their self-reported well-
being are mediated by the leaders’
perceived performance.
Page 60
Published in Small Group Research (2011), 42(4): 405-457
communication from the leader to
followers. After the meeting,
subjects completed the perceived
performance and domain
identification questionnaires. Study
2 was conducted with similar
methods and extended to well-
being responses (subjects filled out
a post-experiment questionnaire
measuring self-esteem and
depressed affect).
Hoyt, Blascovich
& Swinth
2003 Are social
facilitation and
social inhibition
exhibited in avatar,
agent or alone
conditions in
virtual
A 2 x 3 between-participants
experiment with varying task types
(novel or well-learned) and
audience types (alone, agent or
avatar). Before the experiment,
participants had a learning phase
consisting of a categorization task
39 undergraduate
students
Participants in the novel task
condition reported significantly
higher levels of task anxiety
compared to participants in the
well-learned task condition.
Participants in the avatar condition
reported significantly higher levels
Presence
Page 61
Published in Small Group Research (2011), 42(4): 405-457
environments? and a pattern recognition task in
VE with HMDs and with the help
of immediate audio feedback.
After learning the criteria,
participants were assigned to
different task conditions. In the
well-learned condition, the
participants performed the same
task they had just completed, and
in the novel task condition, they
performed another task.
Participants in the no-audience
condition performed the task alone,
participants in the agent-audience
condition were told that two
computer-controlled virtual
observers would join them in VE,
and participants in the avatar
of co-presence than participants in
the agent condition, as did the
participants in the well-learned
task condition compared to the
novel task condition. Participants
in the well-learned task condition
performed significantly better than
participants in the novel task
condition. Participants in the novel
task condition performed
significantly worse in the presence
of avatar observers compared to
when they performed in the
presence of the agent observers or
alone. Participants in the avatar
audience condition experienced
significantly greater co-presence
and performed significantly worse
Page 62
Published in Small Group Research (2011), 42(4): 405-457
condition were joined by two
assistants with HMDs observing
them in VE. Upon completion of
the experiment, participants
completed a short questionnaire.
in the novel task condition than
those in the agent condition.
Jackson, Zhao,
Qui, Kolenic,
Fitzgerald, Harold
& Von Eye
2008 Cultural
differences in
morality in real and
virtual worlds.
Cross-sectional survey
methodology.
600 US and 600
Chinese school pupils
(mean age about 12
years).
Chinese youth (especially girls)
emphasized more (conventionally)
moral off-line behaviors than US
counterparts. However, the
Chinese accepted more
questionable on-line behavior as
compared to their US counterparts.
Real world moral behavior
predicted VE moral behavior.
CyberPsychology
& Behavior
Jung & Lee 2009 Implicit and
explicit attitudes
and personality in
predicting
The experiment tested deception
by letting participants walk a
designated route in a VE. Lying
about following rules was the
60 undergraduate
students.
The results showed that
spontaneous deception was only
explained by implicit attitudes. The
explicit measures were argued not
Acta Psychologica
Page 63
Published in Small Group Research (2011), 42(4): 405-457
deceptive behavior. indicator of deception.
Manipulative personality (self-
report) and implicit and explicit
attitudes towards deception and
honesty and were measured as IVs.
to have explanatory power because
explicit favoritism of deception is
socially very undesirable.
Lantz 2001 Differences in
meetings held face-
to-face or via chat
or collaborative
virtual
environments
between
geographically
dispersed experts.
The participants held their regular
face-to-face meetings three times
through text-based chat and three
times in the virtual environment
Active Worlds. A questionnaire
was sent after these meetings.
4 distributed work
group members who
already had
experience of
continuous work
meetings face-to-face
Meetings via chat and
collaborative virtual environment
were perceived as more task-
oriented than face-to-face
meetings. Avatars were perceived
as enjoyable, and they helped turn-
taking in the meetings. The
participants perceived it as
important that the avatars were in
the same place during the meeting,
even though they could have used
the text-chat of the VE if located in
different parts of the environment.
Behaviour &
Information
Technology
Page 64
Published in Small Group Research (2011), 42(4): 405-457
Linebarger,
Janneck & Kessler
2005 How a virtual tool
that helps to
visualize and create
sub-groups affects
time, quality and
productivity of
simulated product
design task in a
CVE.
Experiment with 18 ad hoc groups
of six persons. Two in each group
used the VE through HMD and
four through PC. The task was for
groups to build a virtual roller
coaster from blocks. Three
experimental conditions were
tested using six groups for each: 1)
no sub-grouping support tool, 2)
manual use of it, and 3) automatic
version of the tool. All group
members were in the same room so
that comments could be heard.
108 undergraduate
university students in
groups of 6.
There were no differences in
completion time between the three
conditions, but quality and
productivity were superb when
group members could manually
use the sub-grouping tool. In fact,
without the tool, all groups failed
to complete the task, and in the
automatic sub-grouping condition
only one group succeeded.
Presence
Nakanishi 2004 Test of how
different features
of a VE called
FreeWalk 1)
affected different
Study 1 compared the VE as a
group discussion platform to FTF
and videoconferencing for seven
persons in three tasks (decision-
making, idea-shaping and free
University students:
1) 21 students in
groups of 7 and 2)
unknown number of
students in dyads.
In study 1, it was found that VE
initiated more turn-taking in
discussions than FTF or
videoconference. Moreover, the
length of utterances was most
International
Journal of Human-
Computer Studies
Page 65
Published in Small Group Research (2011), 42(4): 405-457
group discussion
topics and 2)
helped to overcome
awkward pauses in
the cross-cultural
discussion of pairs.
Note: This VE did
not use avatars but
placed the video
images of
participants’ faces
in 3D space.
discussion). FTF was videotaped
and used as a tracking tool for
tracking the movements and
speech of participants. In study 2,
an agent was developed for the VE
system. The agent (pictured as a
dog) detected awkward pauses in
discussions and presented
discussion-promoting questions to
both participants in a chat bubble.
Preferences and attitudes were
measured by means of a
questionnaire.
equal in VE and participants had
more non-task-related
communication than in other
conditions. Especially in the free
discussion condition, the
participants moved freely in the 3D
space. It was concluded that VEs
can support casual communication.
In study 2, the reactions to agent
intrusions of the American-
Japanese pairs’ discussion revealed
culture-specific effects. Americans
preferred safe discussion openings
by the agent (e.g. weather),
whereas the Japanese preferred
unsafe openings (e.g. religion).
Preferences also related directly to
perceptions of the other person and
Page 66
Published in Small Group Research (2011), 42(4): 405-457
stereotypical representation of the
other’s culture.
Nowak 2004 Study of the
influence of
anthropomorphism
and agency (human
vs. computer
operated virtual
image) on social
judgments in VE.
A 2 X 3 experimental set-up was
created: 1) participants were told
that they were interacting with
agent or avatar, and 2) they were
randomly given a virtual image of
their interaction partner that
represented three levels of
anthropomorphism (high, low, no
image). The task was 15 minutes
of voice interaction with
experimenter-controlled virtual
image in a desktop VE. Social
attraction, credibility and
uncertainty were measured.
134 university
students.
Contrary to hypotheses, the results
showed that low anthropomorphic
images were perceived as more
likable and credible than no image,
which was more credible and
likeable than the highly
anthropomorphic images. The
degree of agency had no effects on
any of the dependent variables and
no differences in uncertainty were
found.
Journal of
Computer-
Mediated
Communication
Nowak & Biocca 2003 Do people feel
more presence
A 2x3 between-participants
experiment with a variously
134 undergraduate
students.
The results showed no differences
in feelings of presence between
Presence
Page 67
Published in Small Group Research (2011), 42(4): 405-457
when interacting
with an avatar than
with an agent, and
is the sense of
presence lower
with partners
represented by less
anthropomorphic
images than with
partners
represented by
highly
anthropomorphic
images or no
images?
interactive participant (avatar
condition and agent condition) and
degree of anthropomorphism
(highly anthropomorphic image,
low anthropomorphic
image, and no image control).
avatars and agents. Existence of a
virtual image increased
telepresence. Participants
interacting with the less
anthropomorphic image reported
more co-presence and social
presence than those interacting
with partners represented by either
no image at all or by a highly
anthropomorphic image of the
other. This indicates that the more
anthropomorphic images set up
higher expectations that lead to
feelings of reduced presence when
these expectations are not met.
Nowak & Rauh 2005 Evaluation of
avatars in a
static context in
30 images were created from 3D
models which were divided into 4
image types: 10 male characters,
255 university
students
A test for a potential order effect
revealed that regardless of which
image participants saw first, the
Journal of
Computer-
Mediated
Page 68
Published in Small Group Research (2011), 42(4): 405-457
terms of their
androgyny,
anthropomorphism,
credibility,
homophily,
attraction, and
likelihood of
choosing the avatar
to represent
oneself.
10 female characters, 5 animal
characters and 5 objects. 4 human
images were purposefully children
and 4 were purposefully lower
rendering quality. All images
(including the objects) had
identifiable eyes and mouth to
ensure some level of consistency.
Measures were taken of
participants' gender, computer use,
computer efficacy, and the
perception of each presented
avatar's anthropomorphism,
androgyny, credibility, homophily,
attraction and their likelihood to be
chosen for an interaction.
first image shown was rated as
more androgynous and less
anthropomorphic than the
subsequent images, and less
homophily was felt towards the
first image than towards
subsequent images. Finally,
participants were less likely to
choose the first image to represent
themselves than subsequent
images. The most feminine avatar
was perceived as the most
attractive and the most masculine
avatar as the most
anthropomorphic. Human male
avatars and human female avatars
were rated as the most
anthropomorphic groups, followed
Communication
Page 69
Published in Small Group Research (2011), 42(4): 405-457
by animals and objects. Female
avatars were perceived as the most
attractive followed by men and
then by non-humans and objects.
Participants were more likely to
choose avatars that were human-
like and of the same gender as
themselves. The quality of the
image reduced the androgyny
perception.
Nowak & Rauh 2008 The study
examined the
reliance on visual
characteristics in
online avatar
perception and
especially the
relationships
In Step 1, respondents rated 30
static avatar representations on the
three dimensions in a survey. The
survey ratings were used to select
eight maximally differing avatars
for the experiment in Step 2. In the
second step, a different set of
participants rated their text chat
Altogether 485
university students.
Path analysis revealed that
perceived avatar androgyny
negatively predicted perceived
avatar anthropomorphism.
Perceived avatar
anthropomorphism, in turn,
positively predicted perceptions of
partner (represented by avatar in
Computers in
Human Behavior
Page 70
Published in Small Group Research (2011), 42(4): 405-457
between avatar
androgyny,
anthropomorphism
and credibility.
partners (represented visually by
randomly assigned avatars) on the
three dimensions.
chat experiment) credibility.
Overall, the experimental results
suggest that the visual cues of the
avatar were stronger predictors of
person perceptions than the
behavior (chat) of the person.
Peña, Hancock &
Merola
2009 Study of priming
effects in VEs.
That is, do
situational cues
activate
unconscious cue-
congruent thoughts
and deactivate cue-
incongruent
thoughts also in
VEs?
Two experiments were conducted
in desktop VEs. In experiment 1,
the priming was created by
dividing the participants into two
groups: those wearing black and
those wearing white cloaks.
Aggression and cohesion were
measured. In experiment 2, the
priming was done by randomly
dividing participants to use a) a
Ku-Klux-Klan (KKK) avatar, b) a
transparent avatar, or c) a doctor
Altogether 139
(51+88) university
students in groups
of three
(experiment 1) and
individually
(experiment 2).
After a three-person group decision
task (all having the same cloak),
the black-cloaked participants in
experiment 1 developed more
aggressive intentions and attitudes
but reported less cohesion than the
white-cloaked participants. In
experiment 2, persons with KKK
avatars wrote more aggressive and
less affiliative TAT stories than
persons with other avatars. The
results suggested that the automatic
Communication
Research
Page 71
Published in Small Group Research (2011), 42(4): 405-457
avatar. Participants wrote stories
based on Thematic Apperception
Test (TAT) pictures, which were
analyzed.
cognitive priming models are
applicable also in VEs.
Sallnäs
2005 How does
communication
mode affect
people’s
experience of
social presence,
presence, and
performance, and
how does it affect
their actual
collaboration in a
virtual
environment?
Two between-participants
experiments with three conditions:
VE text-chat, VE audio, and VE
video. The participants were
instructed to perform a decision-
making task in the VE. The time to
finish the task, number of words
used, and the number of words
used per second in
the dialogue between the
participants were measured.
Sixty subjects,
university students
and administrative
personnel.
Presence and social presence were
higher in audio and video
conditions than in the text
condition. No significant effect
was found in task performance, but
perceived performance was rated
highest in the video condition.
Significantly fewer words per
second were spoken in the text
condition than in other conditions.
In a web environment with audio
and video conditions, there was no
difference in presence, but
perceived presence was rated
Presence
Page 72
Published in Small Group Research (2011), 42(4): 405-457
higher in the video condition than
in the audio condition. There was
no difference in social presence or
performance.
Scheck,
Allmendinger &
Hamann
2008 Media richness
effects in a three-
person negotiation
task.
PC-based collaborative VE
experiment on multilateral and
multi-thematic negotiation.
Avatars with 1) text, 2) audio, and
3) audio + text conditions.
Measurement of objective
performance (time, quality) as well
as subjective satisfaction,
difficulty, concentration and
response speed.
Undergraduate
students in triads.
The performance quality was
similar in all conditions, but
solution was slower in condition 1
than in 2 or 3. No difference
between conditions 2 and 3 was
found. Moreover, higher ratings of
user satisfaction, lower difficulty
higher concentration and higher
response speed were reported for
conditions 2 and 3 as compared to
text condition.
Journal of Media
Psychology
Slater, Sadagic,
Usoh & Schroeder
2000 Comparison of
various aspects
(e.g. emerging
Experimental study that took place
over a two-week period. One
participant with HMD and two
Ten groups of three
people recruited by
an advertisement on
The immersed participant (with
HMD) was rated clearly as leader
and as most talkative after the
Presence
Page 73
Published in Small Group Research (2011), 42(4): 405-457
leadership and
group accord) of
small-group
behavior among
group members in
a virtual and a real
environment.
participants using workstation
displays met first in a shared VE
and carried out a collaborative
puzzle task of solving riddles. One
of the group members continually
monitored a particular one of the
other members in order to assess
social discomfort. After 15
minutes, the virtual session was
terminated, the participants
completed a questionnaire and
continued the same task in the real
world for another 15 minutes.
the university
campus.
virtual meeting, whereas after the
real meeting, each participant was
assigned approximately the same
leadership rating, and there were
no differences in perceived
talkativeness. Overall group scores
revealed a significant difference in
group accord, which was higher
after the real session than the
virtual session. However, this
difference could not be found at an
individual level. Women tended to
show higher accord scores than
men and the more riddles solved
the greater the accord was.
Talamo & Ligorio 2001 Construction of
identities in VE.
A community building a VE within
Active Worlds over eight months
for about four hours weekly both
38 students and 10
teachers from 7
schools in two
Identity construction was heavily
context-dependent, and identities
were not static but negotiated in
CyberPsychology
& Behavior
Page 74
Published in Small Group Research (2011), 42(4): 405-457
on- and off-line. Real-life identities
were known from the beginning.
Ethnographic and conversational
methodology was used. Identity
manipulation was possible through
changing avatar (fully), naming it,
and chat. Two types of data were
collected: 1) open responses to a
questionnaire, 2) recorded chat.
countries (age range
10 to 15 years)
interactions. Many strategies for
playing with different identities
and making them distinct were
found (most commonly trying
different avatars). Chat analysis
revealed different ways of
supporting one’s own and others’
identity construction (e.g. shifts
between real and virtual identity
talk).
Vasalou & Joinson 2009 The effects of
interactional
context on self-
presentation
through avatars.
First, the participants were divided
into one of three context groups
with a task of creating an avatar.
The contexts that were
hypothesized to create different
avatars were on-line 1) blogging
(supposedly neutral condition), 2)
dating and 3) detective gaming.
71 university
students.
The analysis of the experimental
data revealed no differences in
avatar attributes with one
exception: avatars for dating were
reported to be more attractive than
those for blogging. Moreover, the
participants created avatars that
were highly similar with their
Computers in
Human Behavior
Page 75
Published in Small Group Research (2011), 42(4): 405-457
The participants were led to
believe that their avatars would be
used later for the three respective
purposes. Second, after the
experiment, everyone was
interviewed and the data were
analyzed qualitatively.
features. No differences in
participants’ self-awareness across
the three groups were found. The
interviews revealed several
strategies of self-presentation but
these resulted in general tendency
to create self-representative
(realistic) avatars across contexts.
Vasalou, Joinson,
Bänziger, Goldie &
Pitt
2008 How users
negotiate their self-
presentation via an
avatar in social
media.
A thinking aloud method was
applied in two scenarios using
Yahoo! Avatars: 1) constructing an
avatar for a romantic date and 2)
making a postcard including the
avatar to send to a family member
regarding a nice surprise. The
participants’ speech and actions
were viewed and recorded by two
observers in an adjacent room.
20 volunteers from
university.
Three strategies or motives of self-
presentation via an avatar were
found independently of the two
scenarios. 1) An avatar was
customized to reflect a more or less
accurate self by using stable self
attributes or idealizing them to suit
social expectations. 2) Ambiguity
enabled playful avatar design in
order to engage the receiver in
International
Journal of Human-
Computer Studies
Page 76
Published in Small Group Research (2011), 42(4): 405-457
This data was used in qualitative
analysis.
reinvented interpretations of social
life. 3) Avatars were used as
proxies: they were designed to
embody and convey a literal or
symbolic message to the receiver.
Vinayagamoorthy,
Steed & Slater
2008 The impact of
character posture
on the
communication of
affect in IVE.
An experimental study of an IVE
where participants observed a
discussion of two virtual
characters. One character was
designed to be active using two
observable behavioral cues
(posture and facial expression) and
two emotional states (anger and
sadness), and the other character
was designed as passive,
portraying the same neutral state
throughout the experiment. The
verbal content of the discussion
49 male participants
recruited from a
university campus.
In angry conditions with postural
cues and facial expressions, the
participants were able to recognize
the emotional state of the active
character as anger, but this was not
the case in the sad condition.
Moreover, participants did not
perceive the affective state of the
passive character as being neutral,
even though the passive character
was designed to portray neutral
cues throughout the experiment.
The affective state of the passive
IEEE Transactions
on Visualization
and Computer
Graphics
Page 77
Published in Small Group Research (2011), 42(4): 405-457
was muffled in order to ensure that
the affective states could only be
inferred through the nonverbal
behavioral cues. The participant
was instructed to move toward the
characters and once s/he got close
enough, the active character
stopped the discussion, adopted a
neutral state and started an
experimenter-controlled discussion
with the participant while the
passive character remained a
listener. Two sets of between-
groups 2 (facial expression) X 2
(postural cues) factorial designs
were employed, as well as
interviews after the experiment.
character was instead perceived as
a function of the active character’s
behavioral animations. When
measuring personal distance, most
participants maintained smaller
interpersonal distance from the
active than the passive character.
Yee & Bailenson 2007 Does a change of Two between-subjects designs Altogether 82 In study 1, participants in the Human
Page 78
Published in Small Group Research (2011), 42(4): 405-457
one’s self-
representation
change one’s
behavior in a
virtual
environment?
with several conditions: study 1
avatars with attractive or
unattractive faces, and study 2
avatars of the same height, taller,
or shorter than the confederate.
undergraduate
students.
attractive condition walked closer
to the confederate and revealed
more information than
unattractively-represented
participants. In study 2,
participants represented as tall split
money more in their own favor
than those represented as short,
who were more likely to accept an
unfair offer than participants
represented as tall or of normal
height.
Communication
Research
Yee & Bailenson 2009 The relative
contribution of
self-perception and
priming to
behavioral changes
via digital self-
In study 1, participants interacted
with a confederate in a 2 (attractive
or unattractive avatar) X 2 (mirror
or playback condition) between-
subjects experiment. Participants
were assigned avatars with faces
73 undergraduate
students.
In the mirror condition,
participants in the attractive
condition were chosen more often
as a partner than participants in the
unattractive condition, and
participants in the unattractive
Media Psychology
Page 79
Published in Small Group Research (2011), 42(4): 405-457
representation that had been pretested for
attractiveness. In the mirror
condition, participants were
exposed to their avatar in a virtual
mirror whereas in playback
condition they saw a virtual
recording of the previous
participant in the same
attractiveness condition. In both
conditions, participants were
immersed in a virtual environment
by using HMD and asked to walk
closer to the confederate. The
confederate asked questions from
the participants to allow enough
time for social interaction. After
study 1, participants were asked to
pick two pictures from the photos
condition were significantly more
likely to increase their reported
height than participants in the
attractive condition. In the
playback condition, participants
in the attractive condition did not
have significantly different
reported height differences or
higher partner choice scores than
participants in the unattractive
condition. The means showed
that in the mirror condition,
participants in the attractive
condition walked closer to the
confederate than in the unattractive
condition, whereas in the playback
condition, the distance for
both attractive and unattractive
Page 80
Published in Small Group Research (2011), 42(4): 405-457
of nine people of opposite gender
in whom they were most interested
and who they thought would be
most likely be interested in them.
Participants were asked to tell their
height, but they were unaware that
their height was also automatically
measured in the IVE.
conditions was very similar.
Yee, Bailenson &
Duchenaut
2009 How avatar’s
appearance
changes user’s
online and offline
behavior.
In study 1, an automated script was
implemented in a multiplayer
online game (World of Warcraft,
WoW) which gathered information
on all individual characters, such
as their race, their location and
their current level within the game.
Screenshots of each race were
captured and the first four
randomly generated avatars for
In study 1 data was
collected from 76,843
players of WoW. In
study 2 the
participants were 40
undergraduate
students.
The findings of the study 1 show
that tall, attractive avatars are
expected to be the highest level in
the game, whereas the short,
attractive avatars are expected to
be the lowest level. Study 2
showed that the effect of height
was significant on the first split of
money both in virtual environment
and face-to-face trials but not
Communication
Research
Page 81
Published in Small Group Research (2011), 42(4): 405-457
every race and gender were
selected, resulting in 64 images of
WoW avatars. 22 undergraduate
students rated these images based
on their attractiveness. Height of
the race was calculated by
measuring the number of pixels of
their avatar’s height.
In study 2 participants were evenly
divided into short and tall
condition. Participants wore an
HMD and conducted a money
sharing task with a confederate in
IVE, where one designated a split
and the other would accept or
reject the split. After the
experiment, the participants
performed the same task again
significant on the second split or
on the final acceptance of money.
Page 82
Published in Small Group Research (2011), 42(4): 405-457
with the same person face-to-face.
Yee, Bailenson,
Urbanek, Chang &
Merget
2007 Do social norms of
gender and
interpersonal
distance (IPD) and
gaze transfer from
real life to VE?
Field experiment in Second Life.
Data gatherers carried a script that
saved information about avatars
(e.g. position, gaze direction).
From snapshots and script data
IPD and sum of gaze directions
were calculated for interacting
dyads.
835 dyads of Second
Life inhabitants
observed by
undergraduate
research assistants.
Norms were transferred to VE.
Male-male dyads maintained
greater IPD and maintained less
gaze-contact than female-female
dyads. However, mixed gender
dyads had the smallest IPD.
Decrease in IPD was compensated
for avoiding eye-contact. During
chat, the avatars were more likely
to look at each other than when not
chatting.
CyberPsychology
& Behavior
Abbreviations in the Appendix: DV = Dependent Variable; FTF = face-to-face; IV = Independent Variable; IVE = Immersive Virtual Environment; HMD = Head-Mounted Display
Page 83
Published in Small Group Research (2011), 42(4): 405-457
Biographies
Anu Sivunen, Ph.D. (Speech Communication), is Research Manager at Aalto
University, School of Science and Technology, Finland. Her research interests include global
virtual teams, collaborative virtual environments, group identity, and computer-mediated
communication. She has published articles of these issues in journals including IEEE
Transactions on Professional Communication and Group Decision and Negotiation.
Marko Hakonen, Ph.D. (Organizational Psychology), works as a researcher at Aalto
University, School of Science and Technology, Finland. He has studied virtual work and
virtual teams for over a decade. His specialties include research of social identification,
perceived justice and interpersonal trust in new ways of work.
Acknowledgement
The authors wish to thank the anonymous reviewers for their insightful comments as
well as Johanna Haapamäki and Petra Bosch-Sijtsema for their hard work on coding the
articles. This study was supported by the MIDE program, Aalto University School of Science
and Technology, Finland (http://mide.tkk.fi/en/).