Content uploaded by Katrina Fong
Author content
All content in this area was uploaded by Katrina Fong on Jan 21, 2015
Content may be subject to copyright.
Personality and Social
Psychology Bulletin
1 –13
© 2014 by the Society for Personality
and Social Psychology, Inc
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0146167214562761
pspb.sagepub.com
Article
Introduction
Technological advances have produced virtual spaces where
individuals gather and interact (e.g., online video games,
chat rooms; Steinkuehler & Williams, 2006). In some virtual
environments, individuals rely upon a visual graphic to rep-
resent themselves, known as an avatar. An avatar is typically
an image that represents the self in the virtual world, ranging
from very simple drawings (e.g., Mii characters for the
Nintendo Wii) to quite detailed three-dimensional renderings
of characters (e.g., World of Warcraft). A growing body of
exciting research has begun to investigate these avatars and
the outcomes of their use (e.g., Yoon & Vargas, 2014). In
particular, avatars have received a considerable amount of
attention from researchers who are interested in identity
because avatars allow individuals to express (or suppress)
various physical and psychological traits (Hoffner, 2008;
Vasalou & Joinson, 2009; Williams, Kennedy, & Moore,
2011). This ability to selectively represent the self highlights
the importance of first impressions based on avatars.
Although there is evidence that avatars can convey accurate
personality information about their creators (Belisle &
Bodur, 2010), little is known about how individual differ-
ences might moderate the accuracy with which an individual
is perceived or the social consequences of judgments based
on avatars. Here, we investigate a number of questions
related to both the accuracy of personality perception within
the world of avatars and the social consequences of these
impressions.
First Impressions and Computer-Mediated
Communication (CMC)
Interacting via CMC can affect how we form impressions of
others. Because impression formation is an important aspect
of any social interaction, while online individuals identify
CMC-specific cues to form impressions and employ these
cues to make impressions on others. These cues can be mini-
mal and text based, such as emoticons, usernames, and even
writing style (Heisler & Crabill, 2006; McAndrew & De
Jonge, 2011; Walther, Loh, & Granka, 2005), but even these
minimal cues can be used to form impressions of useful
social categories such as gender and disposition (Cornetto &
Nowak, 2006; McAndrew & De Jonge, 2011). Overall, these
CMC-specific cues seem to convey accurate information
about individuals’ personality. Accurate impressions can be
formed based on text-based interactions (Rouse & Haas,
2003), even when only very simple cues such as an email
address are available (Back, Schmukle, & Egloff, 2008).
However, there is some evidence that this accuracy is lower
than that found in face-to-face interactions (Okdie,
Guadagno, Bernieri, Geers, & Mclarney-Vesotski, 2011).
562761PSPXXX10.1177/0146167214562761Personality and Social Psychology BulletinFong and Mar
research-article2014
1York University, Toronto, Ontario, Canada
Corresponding Author:
Katrina Fong, York University, 4700 Keele Street, Toronto, Ontario
M3J1P3, Canada.
Email: kafong@yorku.ca
What Does My Avatar Say About Me?
Inferring Personality From Avatars
Katrina Fong1 and Raymond A. Mar1
Abstract
Many individuals now meet and develop friendships while online. As a result, people must form impressions of online
acquaintances based on that person’s online representation. Here, we investigate personality inferences and intentions to
befriend based solely on simple avatars (i.e., customized cartoon representations of the self). Our data show that some traits
are more easily inferred from avatars than others, avatars can communicate accurate and distinctive information regarding
personality, and individuals with certain personality traits create avatars that are more likely to be perceived accurately. We
also found that agreeable and normative individuals created avatars that elicited more desire for friendship from others,
implying that the impression given by one’s digital avatars may have social consequences.
Keywords
person perception, avatars, accuracy
Received August 13, 2013; revision accepted November 8, 2014
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
2 Personality and Social Psychology Bulletin
Some online environments, such as social networking pro-
files (Back, Stopfer, et al., 2010; Stopfer, Egloff, Nestler, &
Back, 2013) and personal websites (Vazire & Gosling, 2004),
contain both verbal and nonverbal cues, and these have also
been associated with accurate personality judgments.
Beyond the question of accuracy, researchers have
explored how liking emerges based on first impressions
(e.g., Back, Schmukle, & Egloff, 2010, 2011; Küfner,
Nestler, & Back, 2013). Certain individual differences tend
to be associated with positive social outcomes in the real
world. For example, physically attractive individuals tend to
be evaluated and treated more positively (Dion, Berscheid,
& Walster, 1972), and narcissists self-present in ways that
tend to elicit liking from others on first impression (Back,
Schumkle, & Egloff, 2010). Most germane to the current
context, social network profiles that communicate commu-
nal traits and creativity elicit judgments of liking (Stopfer
et al., 2013). Overall, it seems that certain individuals tend
to present themselves in ways that increase the likelihood of
positive social outcomes at first impression. In the current
study, we focus on impressions based on avatars, in light of
their prevalence in online environments and the exciting
surge of research underscoring their importance with respect
to identity.
Impressions based on avatars. Avatars are ubiquitous in
online environments and range widely in their implementa-
tion, from static images to dynamic three-dimensional char-
acters (Belisle & Bodur, 2010; Holzwarth, Janiszewski, &
Neumann, 2006). They can also be an important tool to fos-
ter relationships, increasing feelings of social connectivity
and emotional involvement (e.g., Taylor, 2011). When
encountering an avatar, individuals appear to use visual
cues encoded in the avatar to form an impression of that
avatar and its user. As individuals often anthropomorphize
avatars (Nass & Moon, 2000), it is no surprise that perceiv-
ers respond to avatars in a manner similar to how they form
impressions of individuals in the real world (e.g., based on
physical appearance; Naumann, Vazire, Rentfrow, & Gos-
ling, 2009). Tattooed avatars, for example, are perceived as
being sensation-seeking and risk-taking (Wohlrab, Fink,
Kappeler, & Brewer, 2009). How androgynous and human-
like the avatar appears also influences the perceived credi-
bility and attractiveness of avatars (Nowak & Rauh, 2006),
and these evaluations extend to the credibility of the ava-
tar’s user (Nowak & Rauh, 2008). In addition, these same
cues influence whether users would like to be represented
by a given avatar (Nowak, Hamilton, & Hammond, 2009;
Nowak & Rauh, 2006). The visual characteristics of ava-
tars, therefore, play an important role in both how others
are perceived and also how individuals choose to represent
themselves. But do these cues accurately reflect and com-
municate an individual’s real-world traits?
There are reasons to expect that an avatar’s cues may not
lead to accurate impressions of its user. For one, virtual envi-
ronments are well suited for identity exploration, so users
might adopt identities different from their actual real-world
identity (Hoffner, 2008; Kafai, Fields, & Cook, 2010).
Because avatars can easily be customized, your avatar does
not need to match your own appearance and can embody any
characteristics you wish (Dunn & Guadagno, 2012; Vasalou
& Joinson, 2009). In addition, an avatar may provide an
opportunity to deviate from one’s social identity (Williams
et al., 2011), motivated by enjoyment, entertainment, or
desires to manage self-presentation (Dunn & Guadagno,
2012; Williams et al., 2011). The latter can be achieved by
emphasizing desired psychological traits or physical charac-
teristics, such as confidence, attractiveness, or intelligence
(Dunn & Guadagno, 2012; Vasalou & Joinson, 2009; Vasalou,
Joinson, Banziger, Goldie, & Pitt, 2008). Because the positive
evaluations of an avatar are often extended to the user (Nowak
& Rauh, 2008), individuals may be highly motivated to cus-
tomize avatars in ways inconsistent with reality, resulting in
inaccurate impressions formed by others.
However, there are reasons to believe that avatars may
contain valid identity cues. Individuals who are uncomfort-
able or marginalized in the real world may view virtual
worlds, and the avatars within them, as an opportunity to
express their “true selves” (Williams et al., 2011). Research
has borne out this idea, with individuals often customizing or
choosing avatars to reflect their true personalities, mental
states, and interests (Kafai et al., 2010; Park & Henley, 2007).
Individual differences such as self-esteem, gender, and per-
sonality guide avatar customization (Dunn & Guadagno,
2012). Similar to how our clothes in the real-world convey
information about ourselves to others (e.g., Gillath, Bahns,
Ge, & Crandall, 2012), the clothes we choose for our avatars
may serve a similar function and may even correspond to our
actual clothes (Borkenau & Liebler, 1992). Consistent with
the idea that avatars can accurately reflect identity, individu-
als choose and prefer avatars perceived to be similar to them-
selves (Nowak & Rauh, 2006, 2008). Overall, there is research
to support the idea that individuals may be motivated to create
and employ avatars that are representative of their true iden-
tity. Our study employs the Brunswik Lens Model to examine
questions regarding the accuracy of personality perceptions
based on avatars (Brunswik, 1956).
The Brunswik Lens Model. The Brunswik Lens Model postu-
lates that observable cues found in the environment (e.g.,
cues present in customized avatars) provide a lens through
which perceivers observe constructs that may not be directly
observable (e.g., an avatar creator’s true personality;
Brunswik, 1956). Accuracy in personality perception is
driven by two components: (a) cue validity, the relationship
between phenomena (e.g., personality) and observable cues;
and (b) cue utilization, the relationship between cues and
how they are employed by perceivers. Accuracy occurs when
there is a high degree of convergence between cue validity
and cue utilization. The Brunswik Lens Model can be used to
identify both good and bad sources of personality informa-
tion, across many types of stimuli.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
Fong and Mar 3
The Current Study
The primary research questions for the current study center
on the accuracy of personality impressions and the elicitation
of friendship intentions in others. As a starting point, we
wished to replicate the trait-level accuracy findings of Belisle
and Bodur (2010) using a novel set of avatars to examine the
generalizability of their findings. Building on trait-level accu-
racy, our study will examine the robustness of this accuracy
by taking into account a factor not considered in past work:
the role of sex stereotypes. In an avatar context, sex cues can
activate social categorization (i.e., gender), which may subse-
quently lead to the application of gender stereotypes (Cornetto
& Nowak, 2006). Women are often perceived as more agree-
able and less emotionally stable than men (e.g., Gosling, Ko,
Mannarelli, & Morris, 2002; Spence, 1993), and so percep-
tions of avatars might be influenced by judgments that an ava-
tar is female. These stereotypical judgments of others based
on sex might also contribute to accuracy if the stereotypes
about men and women correspond to actual sex differences. It
is especially relevant to examine these stereotypes in avatars
because the visual cues provided by avatars may provide
more opportunities to elicit sex categorization in perceivers
compared with more cue-lean CMC environments (e.g., user-
names; Cornetto & Nowak, 2006). Based on past research on
real-world perception, we hypothesized that sex stereotypes
would influence perceiver judgments such that ratings of
agreeableness and emotional stability would be predicted by
the sex of the avatar, above and beyond the actual self-
reported levels of these traits by creators.
After exploring the robustness of trait-level accuracy,
we move to the novel question of whether individuals’
overall personalities are also accurately perceived, known
as profile-level accuracy. In other words, can an avatar pro-
vide accurate information about an individual across their
personality profile? To explore this question properly, we
additionally took into account the fact that accuracy in
personality perception can be driven by normative expec-
tancies (Biesanz, 2010; Furr, 2008). That is, reports of
personality are influenced to some degree by the tendency
for these reports to reflect the average profile of the popula-
tion (e.g., an expectation of what the typical person is like).
It is important to account for normative influences when
investigating profile similarity (e.g., the similarity between
self-reported and perceiver-rated personality profiles) as
such influences can inflate similarity estimates. Prior
research has not determined whether accuracy in person
perception based on avatars is due to reliance on normative
expectancies. Thus, we included two components of profile
similarity in our analysis: (a) overall accuracy, or the level
of agreement between the creator and the rater regarding
the creator’s personality; and (b) distinctive accuracy, or
how well raters can predict the personality of creators
above and beyond impressions of what the typical person is
like (e.g., how much more or less extraverted a creator
is beyond an average person’s level of extraversion;
Furr, 2008). We hypothesized that overall profile-level
accuracy would be possible in the form of a positive corre-
lation between the average rated personality profile and
self-reported personality profiles by creators. Furthermore,
we hypothesized that once the influence of normative
expectancies was taken into account, this correlation (i.e.,
distinctive accuracy) would be smaller in magnitude but
remain positive and nonzero. This would indicate some
influence of normative expectancies.
The previous set of questions and hypotheses addressed
whether traits and trait profiles can be accurately inferred
from avatars. Also of interest is a related question that has
not been previously examined: who creates an avatar that is
perceived with profile-level accuracy? Specifically, what
personality traits are associated with creating an avatar that is
perceived more accurately? This differs from the question of
what traits may be inferred accurately. For example, trait
openness may be perceived accurately from avatars across a
group (based on a trait-level analysis). However, an individ-
ual high in openness may also tend to choose atypical avatar
cues to represent their other personality traits, therefore
reducing the overall accuracy with which their entire person-
ality profile can be inferred. The fact that some personality
traits predict higher levels of discrepancy between real-life
appearance and the appearance of one’s avatar suggests that
personality may be an important moderator of accuracy (e.g.,
Dunn & Guadagno, 2012). Although too little is known
about who promotes accurate perceptions in online contexts
for us to form concrete hypotheses, we expected that indi-
viduals who are high on expressive traits will be most likely
to be perceived with greater accuracy (i.e., extraversion and
agreeableness; Ambady, Hallahan, & Rosenthal, 1995).
Finally, first impressions made online should not be con-
sidered in isolation, as they take place within a social context
often motivated by the development of new friendships.
Previous research using social network profiles or photo-
graphs has found that even very brief exposure to a target is
sufficient to influence individuals’ liking of the target (e.g.,
Back, Stopfer, et al., 2010; Stopfer et al., 2013). The current
study extends past research by examining the social conse-
quences of impressions based on avatars: What kinds of
people create an avatar that elicits a desire to befriend its
creator? As extraverted individuals tend to be more popular
(Back et al., 2011; Eaton & Funder, 2003), we hypothesize
that greater extraversion in creators will be positively corre-
lated with a greater desire for friendship in perceivers. In
addition, we investigated what specific cues, associated with
different creator personality traits, would predict greater
friendship intentions on the part of perceivers. Finally, hav-
ing a personality profile that closely matches a normative
profile can be indicative of psychological adjustment or
social desirability (Furr, 2008), and this may influence inten-
tions to befriend. We hypothesized that global normative-
ness, the similarity between a creator and the average
personality profile (e.g., what people in general are like),
would be positively correlated with intentions to befriend an
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
4 Personality and Social Psychology Bulletin
avatar creator. An additional factor to consider is how reveal-
ing an avatar is with respect to its creator’s personality.
Avatars that communicate more of the creator’s true person-
ality might be more liked because people would seem to
prefer others who present themselves in an open and honest
manner. Therefore, we hypothesized that avatar creators who
were perceived with higher levels of accuracy would also
elicit greater intentions to befriend.
Method
This study involved two phases: In Phase 1, participants cre-
ated customized avatars, and in Phase 2, a different set of
participants viewed and rated the avatars created in Phase 1.
Participants
Participants in both Phase 1 and Phase 2 were recruited from
the undergraduate research participant pool at a large
Canadian university and received partial course credit for
participation. There were 99 participants (50 male) in Phase
1, who ranged in age from 17 to 40 years, M = 19.76 years,
SD = 3.76 years. In Phase 2, 209 individuals (60 male), rang-
ing in age from 16 to 36 years (M = 19.42 years, SD = 2.68
years), participated.
Materials
Avatar creation task. Participants created an avatar using an
online tool: weeworld.com. This website allows people to
choose a basic form for their avatar (e.g., sex, skin tone) and
customize it along various dimensions, including hair, cloth-
ing, and accessories (Figure 1). All participants consented to
having these avatars presented to other research participants.
Big Five Inventory–44 (BFI-44). To assess personality, partici-
pants completed the BFI-44 (John, Donahue, & Kentle,
1991). The BFI-44 is based on the five-factor model of per-
sonality and assesses the five major traits: (a) openness, (b)
conscientiousness, (c) extraversion, (d) agreeableness, and
(e) neuroticism. This measure consists of 44 descriptive
phrases, which respondents rate with respect to self-
characterization. Responses are given using a 5-point Likert-
type scale that ranges from 1 (disagree strongly) to 5 (agree
strongly). Example items include “I see myself as someone
who is full of energy” (extraversion) and “I see myself as
someone who gets nervous easily” (neuroticism). The BFI-
44 is a reliable and valid method of measuring five-factor
personality (John & Srivastava, 1999). We used scores from
the BFI-44 as a comprehensive measure of personality in all
analyses that examined how creators’ personalities might
relate to being accurately perceived and the elicitation of a
desire to befriend.
Big Five Inventory–10 (BFI-10). Personality was also measured
using the BFI-10 (Rammstedt & John, 2007), an abbreviated
version of the BFI in which each of the five-factor traits is
measured by two items, resulting in a total of 10 items. Each
trait is measured by one true-scored item and one reverse-
scored item. For example, extraversion is measured by the
two items, “I see myself as someone who is outgoing, socia-
ble” and “I see myself as someone who is reserved.” Respon-
dents rate each statement on a 5-point Likert-type scale that
ranges from 1 (disagree strongly) to 5 (agree strongly).
Despite its brevity, the BFI-10 has demonstrated good test–
retest reliability, as well as good convergence with more
detailed assessments of personality such as the 44-item BFI
(Rammstedt & John, 2007). The BFI-10 was only used to
determine profile accuracy, to allow for a direct comparison
between self-reported personality profiles and perceived per-
sonality profiles.
Desired friendship. Whether an individual was interested in
becoming friends with the creator of an avatar was measured
using a single item, “I would like to be friends with the per-
son who created this avatar.” Responses to this item were
provided using a 5-point Likert-type scale ranging from 1
(strongly disagree) to 5 (strongly agree).
Procedure
Phase 1 was conducted in a computer lab where participants
created an avatar and subsequently completed the BFI-44. All
participants were given the following instructions: “Please
create an avatar representation of yourself.” Half the partici-
pants were provided with an additional instruction, “Your
avatar should represent who you really are (e.g., your person-
ality); remember, your avatar does not need to look like you!”
Initial analyses indicated no differences between these two
groups, so these groups were collapsed for the final analysis.
There were no cross-sex or gender atypical avatar representa-
tions in this sample. Avatar creators also completed the BFI-10
Figure 1. Example avatars.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
Fong and Mar 5
because this was the measure that perceivers would later
employ to infer personality from the avatar. Having the cre-
ators’ BFI-10 scores allowed us to make a direct comparison
between self-rated personality and inferred personality, based
on the same measure, when exploring profile-level accuracy.
Finally, demographic information was collected.
Data for Phase 2 were collected online using the Qualtrics
survey client (www.qualtrics.com). A second set of partici-
pants, with no overlap from Phase 1, were shown a subset of
15 to 16 avatars created in Phase 1. We created seven subsets
by randomly distributing the avatars from Phase 1.
Participants in Phase 2 were randomly assigned to rate a sub-
set when they were recruited. These participants were given
the following instructions:
You will see a series of digital avatars and be asked to rate each
one based on the personality of its creator. The questionnaire
provided lists a number of characteristics that may or may not
describe the individual you’ve been asked to rate . . . Examine
each avatar and try to predict the personality of the person who
created that avatar.
Each avatar was rated by a minimum of 20 different peo-
ple. For each avatar, participants also indicated whether they
would like to befriend the creator of the avatar.
To evaluate the consensus among raters, overall mean
single-perceiver interrater consensus (i.e., agreement on BFI
personality ratings among all 209 raters and across all ava-
tars) was calculated using an intraclass correlation, ICC(2, 1)
= .19. Overall mean average-perceiver interrater consensus
(i.e., agreement of personality ratings across raters within
each subset) was also calculated, ICC(2, k) = .87, where k
was between 24 and 33 (the number of participants who
rated each subset).
A set of 111 potential cues was identified based on the
avatar customization options, and the number of avatars pos-
sessing any given cue was noted (Table 3; Online Appendix
A). All 99 avatars were then coded for these cues by two
research assistants who acted as independent raters. These
same raters also rated the avatars on three additional dimen-
sions based on overall appearance: stylishness, casualness,
and formalness. For all continuous cues (e.g., rated stylish-
ness), coder ratings were averaged. Mean interjudge agree-
ment was calculated by correlating the two raters’ scores on
each continuous item and then averaging correlations across
items. Interjudge agreement across items averaged .63. For
binomial cues (e.g., brown hair), any disagreement between
raters was resolved by the first author. Cue utilization was
calculated by correlating the coded physical cues of the cre-
ated avatars with the average perceived score for each trait.
Calculating cue validity followed a similar format but
employed the avatar cues and self-reported personality traits
from the BFI-44. Avatar cues, their cue utilization, and cue
validity values can be found in Table 3 and Online Appendix A.
Results
Can Individual Personality Traits Be Accurately
Inferred From Avatar Cues?
Trait-level accuracy was calculated by correlating the aver-
age rating of each trait with centered self-reported creator
scores on the BFI-44. Because subsets of avatars were rated
by subsets of perceivers (i.e., a planned missing design), a
multilevel approach was utilized. The fixed effect from the
model was standardized and represents the average relation-
ship between the creator self-report and perceiver ratings of
that trait, on average across perceivers (Table 1, column 1).
According to this analysis, avatars can provide accurate
information regarding trait extraversion, agreeableness, and
neuroticism but not conscientiousness or openness.
To provide some insight into the process of how trait-level
accuracy might be achieved, we examined the relation
between cue utilization and cue validity. Using vector-
column correlations (Funder & Sneed, 1993), we were able
to examine whether cue choices associated with creator per-
sonality were also utilized by perceivers. Cue utilization and
cue validity correlations were first transformed using Fisher’s
r-to-z formula to form vectors. Cue utilization and cue valid-
ity vectors were then correlated across all 114 cues for each
of the Big Five traits. This procedure characterizes the extent
to which cue utilization and cue validity are congruent (Table 1,
column 2). Vector correlations for extraversion, agreeable-
ness, and neuroticism were all significant and positive,
whereas the vector correlations for conscientiousness and
openness did not reach threshold for significance (although
openness fell just above threshold, p = .06). This indicates
that the way individuals customize avatars to reflect their
Table 1. Trait-Level Accuracy, Cue-Based Trait-Level Vector Correlations, and Regressions of Accuracy Controlling for Gender.
Trait
Trait-level
accuracy (β)
Vector-column
correlations Gender (β)
Accuracy (β),
controlling for gender
Extraversion .24* .43* −.04 .24*
Agreeableness .13* .41* −.09 .11*
Conscientiousness .03 .15 −.29* .01
Neuroticism .10* .40* −.09 .08
Openness .04 .18 −.12* .04
*p < .05.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
6 Personality and Social Psychology Bulletin
own traits is congruent with how perceivers use these avatar
cues to infer personality (for all traits except for conscien-
tiousness and perhaps openness). To further illustrate these
vector correlations, we plotted scatterplots of cue utilization
against cue validity for each trait and fitted each plot with a
loess curve (Figure 2).
Figure 2. Scatterplots for cue utilization and cue validation by trait with fitted loess curve.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
Fong and Mar 7
To investigate whether the valid and utilized cues could
help explain the accurate perception of each trait, we con-
ducted a series of bootstrapped multiple mediation analyses.
For each of the Big Five traits, we conducted a regression
with creator personality predicting rated personality traits
and entered avatar cues that were both utilized and valid as
potential mediators (Table 2). Mediation was observed in
terms of a nonzero total indirect effect of the cues for extra-
version (shorts and jewelry) and agreeableness (open eyes
and a neutral expression). The total indirect effect of the cues
for neuroticism (gray or beige shoes) and openness (number
of accessories) approached significance, with the lower
bound of the bootstrapped 95% confidence interval (CI) just
including 0. No cues for conscientiousness were both valid
and utilized.
Do Sex Stereotypes Affect the Accuracy of
Personality Perception From Avatars?
To investigate the influence of sex stereotypes on percep-
tion, we conducted separate regressions for each mean-
rated personality trait with avatar sex (dummy coded as
females = 0 and males = 1) and self-reported creator per-
sonality included as predictors. Including self-reported cre-
ator personality allowed us to examine whether avatar sex
predicts the rated personality trait above and beyond the
actual self-reported personality of the creator (Table 1, col-
umns 3 and 4). In the case of only two regressions, for con-
scientiousness and openness, was avatar sex a significant
predictor after taking into account self-reported creator per-
sonality. This demonstrates that ratings of conscientious-
ness and openness were driven to some extent by the
perceived sex of the avatar. Specifically, male avatars were
seen as less conscientious and less open to experience.
Impression formation based on avatars does appear some-
what driven by gender-based stereotypes, although the
observed traits for which this was the case did not corre-
spond to our hypotheses. Importantly, in light of the fact
that female creators did not report more conscientiousness
or openness relative to males, it is unlikely that the stereo-
types applied by perceivers improved accuracy in personal-
ity perception, conscientiousness: t(97) = −1.14, p = .26, d
= −0.23; openness: t(97) = 0.28, p = .78, d = 0.06.
Can Personality Profiles Be Accurately Inferred
From Avatar Cues, in Light of Normative
Expectancies?
Moving beyond the accuracy associated with individual
traits, we subsequently examined whether entire personality
profiles could be accurately inferred from avatars. For this
analysis, accuracy is considered a profile-level correlation
(Funder, 1999), where each target’s BFI-10 responses were
correlated with the mean BFI-10 profile provided by the per-
ceivers to directly compare perceived and self-reported
personality. The raw associations were considered a measure
of overall accuracy and were subjected to a single sample
t test with the null-hypothesis being no correlation between
self-reported creator personality and rated personality (test
value of 0). Overall accuracy was statistically different from
0, r = .26, t(98) = 7.75, p < .001, 95% CI [.19, .32]. To parse
out the effect of normative influence on personality judg-
ment, we calculated distinctive and normative accuracy
using a multilevel model following the procedures outlined
by the Social Accuracy Model (Biesanz, 2010). Similar to
the trait correlations, the multilevel model allowed us to
account for the fact that subsets of avatars were being rated
by subsets of perceivers. The fixed effects from this model
were considered. On average, across avatar creators and per-
ceivers, there was statistically significant agreement between
self-reported and rated personality profiles after accounting
for normative influences (i.e., distinctive accuracy), b = .04,
p = .03. In addition, the results for normative accuracy also
reached statistical significance, b = .31, p < .001, indicating
that personality inference from avatars has a normative com-
ponent. These results were consistent with our hypotheses.
Table 2. Bootstrapped multiple mediation analyses examining whether valid and utilized cues mediate the association between self-
reported and perceived personality for each trait.
Trait Relevant cues Indirect effect, βIndirect effect, 95% CI p
Extraversion Total .05 .00, .11 .05
Shorts .03 –.01, .07 .15
Jewelry .02 –.01, .06 .20
Agreeableness Total .08 .03, .13 < .01
Open eyes .01 –.02, .04 .50
Neutral expression .07 .02, .12 .01
Conscientiousness - - - -
Neuroticism Grey or beige shoes .03 –.00, .06 .07
Openness Number of
accessories
.03 –.00, .06 .06
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
8 Personality and Social Psychology Bulletin
What Creator Traits Predict the Creation of an
Avatar That Is Accurately Perceived?
To identify which personality traits of a creator were asso-
ciated with creating an avatar perceived accurately, the pre-
vious multilevel model was expanded to include each
self-reported creator personality trait in turn (as measured
by the BFI-44). Individuals who were more extraverted
(b = .07, p < .001), more agreeable (b = .08, p < .001), and
more conscientious (b = .04, p < .001) were more likely to
be perceived with distinctive accuracy, whereas individuals
who were more neurotic were less likely to be perceived
with distinctive accuracy (b = −.06, p < .001). Creator
openness was not related to being perceived with distinc-
tive accuracy (b = .01, p = .69). To examine which of these
traits uniquely predicted accuracy, all five traits entered
simultaneously as potential moderators, along with gender
and age as covariates (Table 2). Creator extraversion, agree-
ableness, and neuroticism remained unique predictors, with
extraversion and agreeableness predicting greater distinc-
tive accuracy and neuroticism predicting less accuracy.
How Is Creator Personality Related to Intentions
to Befriend That Creator Based on an Avatar?
We began by asking whether certain types of individuals are
more likely to create an avatar that others wish to befriend.
To do so, we analyzed multilevel models with the creator’s
self-reported personality (i.e., trait scores from the BFI-44)
predicting perceivers’ average rated likelihood of befriend-
ing that individual, reporting the standardized fixed effects.
More agreeable creators made avatars more likely to evoke
friendship intentions in perceivers (b = .09, p = .02). No
other personality trait for creators predicted an increased
likelihood of eliciting friendship in perceivers, extraversion:
b = .04, p = .27; conscientiousness: b = .02, p = .63; neuroti-
cism: b = −.02, p = .66; and openness: b = .02, p = .63.
Because we hypothesized that more extraverted creators
would be more likely to elicit friendship intentions, this finding
was inconsistent with our hypothesis.
To explore the question of friendship intentions further,
we engaged in an exploratory vector correlation analysis to
examine whether cues associated with certain traits were
linked to a desire to befriend. This is akin to the previous
analysis examining the convergence between cue validity
and utilization, but in this case, friendship intentions take the
place of cue utilization. We calculated correlations between
avatar cues and scores on the befriending item and applied a
Fisher’s r-to-z transformation. We then calculated a vector
correlation between cue validity and friendship intentions
across the Big Five traits. This allowed us to examine how
avatar customizations were associated with perceivers’ inten-
tions to befriend, with regard to the creator’s personality. The
vector correlation between creator agreeableness and friend-
ship intentions was strongest (r = .57, p < .001), but creator
conscientiousness was also statistically significant, r = .31,
p < .001. In other words, agreeable and conscientious indi-
viduals tended to customize their avatars with cues that were
also associated with the elicitation of friendship intentions.
The vector correlations between friendship intentions and
creator extraversion (r = .20, p = .09), neuroticism (r = .10,
p = .31), and openness (r = −.04, p = .65) did not reach
threshold for statistical significance. Cues that were either
valid or utilized in determining friendship intentions are
summarized in Table 3. Avatars with open eyes, a smile or
grin, an oval face, brown hair, and/or a sweater were more
likely to elicit friendship intentions. In contrast, avatars with
a neutral expression or any other expression (other than a
smile), black hair, short hair, a hat, and/or sunglasses were
less likely to elicit friendship intentions (Figure 1). Most of
Table 3. Cue Validity and Cue Utilization for Cues Pertaining to Friendship Intentions.
Target
BFI_E
Target
BFI_A
Target
BFI_C
Target
BFI_N
Target
BFI_O Avatar cue
Rated
BFI_E
Rated
BFI_A
Rated
BFI_C
Rated
BFI_N
Rated
BFI_O Friendship Base rate
−.02 .25* .12 .04 .03 Open eyes .32* .36* .40* −.03 .29* .50* 94
−.02 .20 −.01 .15 .06 Smile .14 .32* .26* −.17 .20* .32* 50
.00 .08 −.00 .06 .02 Oval face .01 .15 .28* −.02 .12 .31* 56
.24* −.04 .08 .08 .06 Brown hair .09 .20* .22* −.07 .12 .25* 37
.15 .07 .27* −.19 .03 Grin .39* .11 .10 −.30* .13 .22* 31
−.19 .15 .02 .05 −.05 Sweater −.01 .23* .34* .03 .09 .20* 18
.031 −.15 .06 −.26* .17 Short hair .06 −.16 −.34* −.18 −.16 −.22* 32
.17 .04 .05 .04 .10 Sunglasses .17 −.35* −.28* −.13 .19 −.23* 29
.13 .03 .04 −.05 .08 Hat .13 −.08 −.13 −.16 .29* −.26* 27
−.12 −.10 −.08 −.13 −.09 Black hair −.17 −.31* −.21* .23* −.25* −.30* 44
−.12 −.10 −.32* .09 −.03 Other mouth
expression
−.39* −.17 −.15 .57* −.12 −.32* 5
−.09 −.32* −.14 −.02 −.11 Neutral expression .48* −.52* −.41* .30* −.40* −.57* 13
Note. BFI_E = Extraversion; BFI_A = Agreeableness; BFI_C = Conscientiousness; BFI_N = Neuroticism; BFI_O = Openness to experience.
*p < .05.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
Fong and Mar 9
these cues were correlated with perceptions of conscientious-
ness and to a lesser degree agreeableness (positively or nega-
tively, consistent with the elicitation of intentions; Table 3).
In light of the fact that creator agreeableness was the only
unique predictor of intentions to befriend, we further exam-
ined whether specific cues could account for this relation-
ship. There were two cues related to both creator agreeableness
and friendship intentions: open eyes and a neutral expression
(negative predictor). A bootstrapped multiple mediation
analysis indicated a nonsignificant indirect effect of open
eyes (β = .03, 95% CI [−.00, .07]) and a significant indirect
effect of neutral expression (β = .07, 95% CI [.02, .12]), with
a significant total indirect effect, β = .10, 95% CI [.04, .16].
This means that the relationship between creator agreeable-
ness and intentions to befriend could be partially explained
by customizing an avatar to have open eyes and avoiding a
neutral expression.
Along with customization cues, other factors associated
with impressions of avatars might help to explain friendship
intentions. Two such factors are how accurately the avatar is
perceived and also how normative the avatar is perceived to
be. We included friendship intentions in the previous multi-
level model examining distinctive and normative accuracy to
investigate the linear relationship between these components
of accuracy and desired friendship. Distinctive accuracy was
not related to desired friendship (b = −.01, p = .39), but
global normativeness (b = .29, p < .001) predicted friendship
intentions. In support of our hypothesis, those who made
avatars perceived as typical—those whose self-reported
personality was close to the average personality—were per-
ceived more positively by perceivers.
Discussion
In this study, we addressed a number of questions related to
impressions formed of others based on digital avatars. Our
results confirmed that avatars can provide accurate personality
information at the level of individual traits, with extraver-
sion, agreeableness, and neuroticism accurately predicted
from avatars. Accuracy in assessing extraversion has also
been observed based on static real-world cues (e.g., pictures;
Borkenau & Liebler, 1992; Funder & Dobroth, 1987; Stopfer
et al., 2013). This suggests that extraversion is highly observ-
able in both online and real-world contexts, perhaps contrib-
uting to its high level of accurate assessment across
environments (John & Robins, 1993). The fact that we
observed accuracy for agreeableness and neuroticism in the
avatar context diverged from past work using real-world
thin-slice exposure, as cues for these traits are generally less
observable in those contexts (Borkenau & Liebler, 1992;
Funder & Dobroth, 1987). This agreeableness finding
appears to be reliable, however, as it replicates a past study
on avatars (Belisle & Bodur, 2010). Neuroticism, in contrast,
is not often inferred accurately in CMC contexts (Back,
Stopfer, et al., 2010; Belisle & Bodur, 2010; Gill, Oberlander,
& Austin, 2006; Wall, Taylor, Dixon, Conchie, & Ellis,
2013). We also found that conscientiousness and openness
were not accurately predicted from the avatars, consistent
with past work (Belisle & Bodur, 2010), although conscien-
tiousness is often accurately perceived in the real world
(Borkenau & Liebler, 1992; Funder & Dobroth, 1987;
Gosling et al., 2002). In interpreting these findings, it is
important to emphasize that the avatar context is a relatively
cue-lean context compared with the real world. This differ-
ence might differentially influence the visibility of certain
traits. As past work has shown, some traits become more
accurately perceived as cue-richness decreases (i.e., extra-
version and neuroticism), whereas other traits become less
accurately perceived (i.e., conscientiousness and openness;
Wall et al., 2013). Cues to personality in the real world can
come either directly in the form of identity claims (choosing
to display cues that reinforce one’s self-identity) or indirectly
in the form of behavioral residue (remnants of behavior
driven by personality; Gosling et al., 2002). In the avatar
context, it seems that only identity claims are available, with
raters being aware that every customization is deliberately
chosen. If an avatar is wearing a dirty shirt, the rater knows
that this was a deliberate choice and not the accidental out-
come of some personality-revealing behavior (i.e., the avatar
did not spill ketchup on her clean shirt). There are, therefore,
fewer sources of information available in the avatar context
relative to the real world, but the meaning of the available
information is also different across contexts as the rater
knows that the available cues are not present by chance.
When cue validity and cue utilization were considered,
there was a congruence observed for extraversion, agreeable-
ness, and neuroticism. The customization cues employed by
creators were related to these traits, and raters used these
same cues to inform their judgments of these traits. Much
less congruence was observed for conscientiousness and
openness. One possible explanation for this lower congruence
is that the customization choices made available did not pro-
vide the necessary options to convey information regarding
these traits. It was not possible to choose a dirty or rumpled
shirt, for example; all clothing appeared neat and tidy on the
avatar. Similarly, the clothing and accessory choices may not
have allowed for sufficient creativity to convey high levels
of openness. It is also important to stress that we may have
observed different results if we had employed more detailed
or dynamic avatars that incorporated movement.
Avatar sex characteristics also influenced how raters saw
avatar creators. Specifically, when rating avatars created by
females, perceivers tended to rate them as being more con-
scientiousness and open, even after taking into account the
creators’ true conscientiousness and openness. In other words,
the avatar’s sex was considered an important cue for predicting
conscientiousness and openness. Females, however, did not
report being more conscientious and more open, so relying on
sex categorization to influence these judgments might have
lowered accuracy for these traits overall; conscientious and
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
10 Personality and Social Psychology Bulletin
openness were the two traits for which trait-level accuracy
was not observed. It is somewhat surprising that avatar sex
did not influence judgments in typical gender stereotypic
directions (i.e., for agreeableness and emotional stability), as
we predicted. That said, our results are consistent with past
work indicating that sex categorization plays an important
role in person perception for CMC contexts (e.g., Cornetto &
Nowak, 2006).
Not only can avatars be a source of trait accuracy, but an
individual’s unique personality profile can also be perceived
accurately, even after accounting for normative expectancy.
In other words, avatars can provide accurate information
about how its creator is different from the average person.
Overall accuracy was larger than distinctive accuracy, how-
ever, indicating that expectations based on what the average
person is like boosted overall accuracy ratings. Furthermore,
some individuals are perceived more accurately than others.
Identifying who is perceived with more accuracy (i.e., a
good target who is high in judgeability) is important because
this individual difference is associated with various positive
outcomes, including greater psychological adjustment and
higher social status (Human & Biesanz, 2013). In our study,
more agreeable individuals were more likely to be perceived
with greater distinctive accuracy, with high extraversion
also exhibiting a similar pattern (but falling just below
threshold for statistical significance). These findings are
consistent with real-world impression formation, with extra-
version and agreeableness predicting expressive or sociable
behaviors that are then accurately perceived by others
(Ambady et al., 1995; Human & Biesanz, 2013; John &
Srivastava, 1999). Individuals who are more social or
greatly value harmonious relationships (i.e., those high in
extraversion and agreeableness) may invest in accurate rep-
resentations of their personality as being perceived accu-
rately is related to positive socialization (e.g., Human &
Biesanz, 2013). In contrast, we found that individuals higher
in neuroticism were perceived with less distinctive accu-
racy. It is possible that highly neurotic individuals are less
well adjusted, which may correspond to a tendency to shield
one’s personality or needs from others (Human & Biesanz,
2013). This is purely conjectural, however, and this relation-
ship between neuroticism and accuracy should be more
directly explored in future research. Another issue to con-
sider is that customizing an avatar is an opportunity to control
self-presentation (e.g., Hoffner, 2008), and in this context,
personality traits may moderate whether people use this
opportunity to pursue accuracy, self-exploration, or impres-
sion management. Extraverted and agreeable people may be
more motivated by accuracy, whereas highly neurotic indi-
viduals may be more motivated by impression management.
This would be consistent with the patterns in our data, but a
direct test of this idea remains a future goal.
Finally, we explored the social consequences of first
impressions based on avatars, discovering that individuals
who were more agreeable were more likely to create an ava-
tar that others wanted to befriend. This result is consistent
with real-world findings where agreeable individuals are
more likely to be selected as friends (Selfhout et al., 2010). It
does, however, differ from a past study using social network
photographs, in which target agreeableness was not a predic-
tor of liking (Stopfer et al., 2013). It may be that agreeable
people are able to craft likeable avatars but are less capable
of creating or choosing likeable pictures of themselves.
Although purely speculative, this may be because one has
more control over one’s appearance when customizing stan-
dardized avatars compared with the realm of photography.
We also found that individuals reporting an average per-
sonality (perceived as such by perceivers) created avatars
that were more likely to elicit the desire to befriend in others.
In addition, distinctive accuracy was not found to predict
friendship intentions. Therefore, our data indicate that inten-
tions to befriend individuals based on their avatar are most
robustly determined by the degree to which the creator is
seen as an average or normal person. The preference to
befriend individuals who are perceived to be average might
reflect our greater comfort pursuing friendships within our
in-group (Jugert, Noack, & Rutland, 2011) or with those
whose normativeness is perceived as an indicator of social
adjustment (Furr, 2008).
Certain cues were also associated with friendship inten-
tions, most notably having to do with the face and facial
expression (e.g., eyes, mouth, hair, and head shape). This is
consistent with past research using photographs: One factor
that predicts liking at zero-acquaintance is a charming facial
expression (Back, Schmukle, & Egloff, 2010). Perhaps
related to the importance of these facial cues, accessories
that blocked the face (i.e., hats and sunglasses) led to lower
ratings of friendship intentions. (No other accessories
covered the face.) Only one clothing item was positively cor-
related with friendship intentions: the wearing of a sweater.
A purely speculative interpretation of these findings is that
avatars that contain cues communicating warmth, be they
through friendly facial expressions (e.g., smiling, open eyes)
or perhaps metaphorically through their clothing (e.g., a cozy
sweater), are more likely to elicit intentions to befriend in
others. In contrast, choosing to customize one’s avatar with
accessories that conceal the face (e.g., sunglasses, hats) may
come across as closed off or lacking in warmth, reducing this
likelihood. This would be consistent with the fact that these
cues were often related to perceived agreeableness, but it is a
bit more difficult to theorize why these same cues also pre-
dicted perceived conscientiousness. In light of the large
number of cues examined, these exploratory analyses should
likely be interpreted cautiously and await replication.
We feel that further studies are needed to investigate the
broader social outcomes of impression formation from ava-
tars. To facilitate research into this topic, we have decided to
release the full set of created avatars and associated personal-
ity and demographic data for the avatar creators (from Phase
1) to other researchers. This stimulus set should provide a
helpful tool for future work and will be made available to
researchers upon request to the first author.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
Fong and Mar 11
Limitations
The current study is subject to a number of limitations. The
avatars employed in this study were simple, two-dimensional,
and static (Figure 1). In general, these types of avatars may
be used by individuals in lieu of a real photograph on social
networking profiles or instant messaging applications or in
simple online social environments where the primary activities
tend to be chat oriented. However, in many virtual environ-
ments such as online games, avatars are much more detailed,
three dimensional, and move according to the wishes of the
user. These dynamic avatars would also provide personality
information in the form of behavioral residue (Gosling et al.,
2002). It would be interesting to see if the results found here
replicate with other forms of avatars, particularly because
patterns of avatar behavior have been found to be associated
with personality traits (Yee, Harris, Jabon, & Bailenson,
2011) and can be used to infer personality (Yee, Ducheneaut,
Nelson, & Likarish, 2011). Furthermore, because the partici-
pants who created these avatars were not intending to keep or
use them in any way, it is possible that they were less moti-
vated to treat the process seriously compared with when they
create avatars for use in their own lives. As a result, our find-
ings are likely attenuated relative to phenomena associated
with real-world avatar creation, exacerbating the possibility
of false negatives. To counter this limitation in future studies,
participants could be informed that they will be using their
avatars for various interpersonal tasks. Doing so would allow
researchers to examine how contextual pressures affect the
created avatars and resultant accuracy in personality percep-
tion (e.g., Vasalou & Joinson, 2009). Additional personality
data for avatar creators should also be collected in future
studies to supplement self-reports, such as personality ratings
from family members and peers (Vazire, 2006).
The avatar creators and perceivers who participated in
this sample were drawn exclusively from a university
participant pool. It is possible that the relatively young
average age of our sample means that our participants were
more familiar with both creating and assessing avatars
than the general population. Therefore, our participants
may have been more accurate in both conveying their per-
sonalities using avatars and predicting personality from
avatars than might be expected in an older, general popula-
tion sample. Replications using community samples may
be informative in establishing the boundaries of accuracy
in personality prediction from avatars.
A notable limitation of our study was that intentions to
befriend were based on self-report and relied on a single-
face valid item. To prevent participant fatigue in Phase 2, we
employed brief measures of both personality and friendship
intentions. As such, our findings with regard to friendship
intentions should be interpreted with some caution. A future
study should employ more realistic and behavior-based
measures of friendship intentions, such as indicating a will-
ingness to share one’s email address with the avatar creator
or a desire to meet with the avatar creator in person.
Furthermore, future studies should include a multi-item
measure of friendship intentions to assess feelings of liking
and desirability regarding the avatar creator.
Conclusion
Understanding impression formation in online environ-
ments is a timely and relevant undertaking, considering the
rapid explosion of online interactions in recent years. It
has even been suggested that meeting an individual online
can be more informative than meeting that individual in
person because of the wealth of information provided by
personal webpages and/or social networking profiles
(Gosling, Gaddis, & Vazire, 2007). The findings from this
study suggest that we can use virtual proxies such as avatars
to accurately infer personality information about others.
Importantly, the impressions we make on others online
may have an important impact on our real life, such as who
becomes intrigued by the possibility of our friendship.
Acknowledgment
The authors would like to thank Nick Rule for his valuable guidance
regarding earlier versions of this manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect
to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support
for the research, authorship, and/or publication of this article: The
authors acknowledge the Social Science and Humanities Research
Council (Grant 435-2012-1420) for supporting this research.
Supplemental Material
The online supplemental material is available at http://pspb.
sagepub.com/supplemental.
References
Ambady, N., Hallahan, M., & Rosenthal, R. (1995). On judging
and being judged accurately in zero-acquaintance situations.
Journal of Personality and Social Psychology, 69, 518-529.
Back, M. D., Schmukle, S. C., & Egloff, B. (2008). How extra-
verted is honey.bunny77@hotmail.de? Inferring personality
from email addresses. Journal of Research in Personality, 42,
1116-1122.
Back, M. D., Schmukle, S. C., & Egloff, B. (2010). Why are narcis-
sists so charming at first sight? Decoding the narcissism-pop-
ularity link at zero acquaintance. Journal of Personality and
Social Psychology, 98, 132-145.
Back, M. D., Schmukle, S. C., & Egloff, B. (2011). A closer look
at first sight: Social relations lens model analysis of personal-
ity and interpersonal attraction at zero acquaintance. European
Journal of Personality, 25, 225-238.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
12 Personality and Social Psychology Bulletin
Back, M. D., Stopfer, J. M., Vazire, S., Gaddis, S., Schmukle, S. C.,
Egloff, B., & Gosling, S. D. (2010). Facebook profiles reflect
actual personality, not self-idealization. Psychological Science,
21, 1-3.
Belisle, J., & Bodur, H. O. (2010). Avatars as information:
Perception of consumers based on their avatars in virtual
worlds. Psychology & Marketing, 27, 741-765.
Biesanz, J. C. (2010). The social accuracy model of interpersonal
perception: Assessing individual differences in perceptive and
expressive accuracy. Multivariate Behavioral Research, 45,
853-885.
Borkenau, P., & Liebler, A. (1992). Trait inferences: Sources of
validity at zero acquaintance. Journal of Personality and Social
Psychology, 62, 645-657.
Brunswik, E. (1956). Perception and the representative design of
psychological experiments. Berkeley: University of California
Press.
Cornetto, K. M., & Nowak, K. L. (2006). Utilizing usernames
for sex categorization in computer-mediated communication:
Examining perceptions and accuracy. CyberPsychology &
Behavior, 9, 377-387.
Dion, K. K., Berscheid, E., & Walster, E. (1972). What is beauti-
ful is good. Journal of Personality and Social Psychology, 24,
285-290.
Dunn, R. A., & Guadagno, R. E. (2012). My avatar and me—
Gender and personality predictors of avatar-self discrepancy.
Computers in Human Behavior, 28, 97-106.
Eaton, L. G., & Funder, D. C. (2003). The creation and conse-
quences of the social world: An interactional analysis of extra-
version. European Journal of Personality, 17, 375-395.
Funder, D. C. (1999). Personality judgment: A realistic approach
to person perception. San Diego, CA: Academic Press.
Funder, D. C., & Dobroth, K. M. (1987). Differences between
traits: Properties associated with interjudge agreement. Journal
of Personality and Social Psychology, 52, 409-418.
Funder, D. C., & Sneed, C. D. (1993). Behavioral manifestations of
personality: An ecological approach to judgmental accuracy.
Journal of Personality and Social Psychology, 64, 479-490.
Furr, M. R. (2008). A framework for profile similarity: Integrating
similarity, normativeness, and distinctiveness. Journal of
Personality, 76, 1267-1315.
Gill, A. J., Oberlander, J., & Austin, E. (2006). Rating e-mail
personality at zero acquaintance. Personality and Individual
Differences, 40, 497-507.
Gillath, O., Bahns, A. J., Ge, F., & Crandall, C. S. (2012). Shoes as a
source of first impressions. Journal of Research in Personality,
46, 423-430.
Gosling, S. D., Gaddis, S., & Vazire, S. (2007, March). Personality
impressions based on Facebook profiles. Paper at the
International AAAI Conference on Weblogs and Social Media,
Boulder, CO.
Gosling, S. D., Ko, S. J., Mannarelli, T., & Morris, M. E. (2002). A
room with a cue: Personality judgments based on offices and
bedrooms. Journal of Personality and Social Psychology, 82,
379-398.
Heisler, J. M., & Crabill, S. L. (2006). Who are “stinkybug” and
“Packerfan4”? Email pseudonyms and participants’ percep-
tions of demography, productivity, and personality. Journal of
Computer-Mediated Communication, 12, 114-135.
Hoffner, C. (2008). Parasocial and online social relationships. In
S. L. Calvert & B. J. Wilson (Eds.), The handbook of children,
media, and development (pp. 309-333). Malden, MA: Wiley-
Blackwell.
Holzwarth, M., Janiszewski, C., & Neumann, M. M. (2006). The
influence of avatars on online consumer shopping behavior.
Journal of Marketing, 70, 19-36.
Human, L. J., & Biesanz, J. C. (2011). Target adjustment and self-
other agreement: Utilizing trait observability to disentangle
judgeability and self-knowledge. Journal of Personality and
Social Psychology, 101, 202-216.
Human, L. J., & Biesanz, J. C. (2013). Targeting the good target: An
integrative review of the characteristics and consequences of
being accurately perceived. Personality and Social Psychology
Review, 17, 248-272.
John, O. P., Donahue, E. M., & Kentle, R. L. (1991). The Big Five
Inventory-Versions 4a and 54. Berkeley, CA: University of
California, Berkeley, Institute of Personality and Social Research.
John, O. P., & Robins, R. W. (1993). Determinants of interjudge
agreement on personality traits: The big five domains, observ-
ability, evaluativeness, and the unique perspective of the self.
Journal of Personality, 61, 521-551.
John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy:
History, measurement, and theoretical perspectives. In L. A.
Pervin & O. P. John (Eds.), Handbook of personality: Theory
and research (2nd ed., pp. 102-138). New York, NY: Guilford.
Jugert, P., Noack, P., & Rutland, A. (2011). Friendship prefer-
ences among German and Turkish preadolescents. Child
Development, 82, 812-829.
Kafai, Y. B., Fields, D. A., & Cook, M. S. (2010). Your second
selves: Player-designed avatars. Games and Culture, 5, 23-42.
Küfner, A. C. P., Nestler, S., & Back, M. D. (2013). The two
pathways to being an (un-)popular narcissist. Journal of
Personality, 81, 184-195.
McAndrew, F. T., & De Jonge, C. R. (2011). Electronic person per-
ception: What do we infer about people from the style of their
e-mail messages? Social Psychological & Personality Science,
2, 403-407.
Nass, C., & Moon, Y. (2000). Machines and mindlessness: Social
responses to computers. Journal of Social Issues, 56, 81-103.
Naumann, L. P., Vazire, S., Rentfrow, P. J., & Gosling, S. D.
(2009). Personality judgments based on physical appearance.
Personality and Social Psychology Bulletin, 35, 1661-1671.
Nowak, K. L., Hamilton, M. A., & Hammond, C. C. (2009). The
effect of image features of judgments of homophily, credibil-
ity, and intention to use as avatars in future interactions. Media
Psychology, 12, 50-76.
Nowak, K. L., & Rauh, C. (2006). The influence of the avatar on the
online perceptions of anthropomorphism, androgyny, credibil-
ity, homophily, and attraction. Journal of Computer-Mediated
Communication, 11, 153-178.
Nowak, K. L., & Rauh, C. (2008). Choose your “buddy icon” care-
fully: The influence of avatar androgyny, anthropomorphism,
and credibility in online interactions. Computers in Human
Behavior, 24, 1473-1493.
Okdie, B. M., Guadagno, R. E., Bernieri, F. J., Geers, A. L., &
Mclarney-Vesotski, A. R. (2011). Getting to know you: Face-
to-face versus online interactions. Computers in Human
Behavior, 27, 153-159.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
Fong and Mar 13
Park, A. E., & Henley, T. B. (2007). Personality and fantasy
game character preferences. Imagination, Cognition, and
Personality, 27, 37-46.
Rammstedt, B., & John, O. P. (2007). Measuring personality in
one minute or less: A 10-item short version of the Big Five
Inventory in English and German. Journal of Research in
Personality, 41, 203-212.
Rouse, S. V., & Haas, H. A. (2003). Exploring the accuracies and
inaccuracies of personality perception following Internet-
mediated communication. Journal of Research in Personality,
37, 446-467.
Selfhout, M., Burk, W., Branje, S., Denissen, J., van Aken, M.,
& Meeus, W. (2010). Emerging late adolescent friendship
networks and Big Five personality traits: A social network
approach. Journal of Personality, 78, 509-538.
Spence, J. T. (1993). Gender-related traits and gender ideology:
Evidence for a multifactorial theory. Journal of Personality
and Social Psychology, 64, 624-635.
Steinkuehler, C. A., & Williams, D. (2006). Where everybody
knows your (screen) name: Online games as “third places.”
Journal of Computer Mediated Communication, 11, 885-909.
Stopfer, J. M., Egloff, B., Nestler, S., & Back, M. D. (2013). Being
popular in online social networks: How agentic, communal,
and creativity traits relate to judgments of status and liking.
Journal of Research in Personality, 47, 592-598.
Taylor, L. D. (2011). Avatars and emotional engagement in asyn-
chronous online communication. Cyberpsychology, Behavior,
and Social Networking, 4, 207-212.
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., Banziger, T., Goldie, P., & Pitt, J. (2008).
Avatars in social media: Balancing accuracy, playfulness
and embodied messages. International Journal of Human-
Computer Studies, 66, 801-811.
Vazire, S. (2006). Informant reports: A cheap, fast, and easy method
for personality assessment. Journal of Research in Personality,
40, 472-481.
Vazire, S., & Gosling, S. (2004). E-perceptions: Personality impres-
sions based on personal websites. Journal of Personality and
Social Psychology, 87, 123-132.
Wall, H. J., Taylor, P. J., Dixon, J., Conchie, S. M., & Ellis, D.
A. (2013). Rich contexts do not always enrich the accuracy
of personality judgments. Journal of Experimental Social
Psychology, 49, 1190-1195.
Walther, J. B., Loh, T., & Granka, L. (2005). Let me count the
ways: The interchange of verbal and nonverbal cues in com-
puter-mediated and face-to-face affinity. Journal of Language
and Social Psychology, 24, 36-65.
Williams, D., Kennedy, T. L. M., & Moore, R. (2011). Behind the
avatar: The patterns, practices, and functions of role playing in
MMOs. Games and Culture, 6, 171-200.
Wohlrab, S., Fink, B., Kappeler, P. M., & Brewer, G. (2009).
Differences in personality attributions toward tattooed and
nontattoed virtual human characters. Journal of Individual
Differences, 30, 1-5.
Yee, N., Ducheneaut, N., Nelson, L., & Likarish, P. (2011).
Introverted elves and conscientious gnomes: The expression of
personality in World of Warcraft. In D. Tan (Ed.), Proceedings
of CHI 2011 (pp. 753-762). New York, NY: ACM Press.
Yee, N., Harris, H., Jabon, M., & Bailenson, J. N. (2011). The
expression of personality in virtual worlds. Social Psychology
and Personality Science, 2, 5-12.
Yoon, G., & Vargas, P. T. (2014). Know thy avatar: The unintended
effect of virtual-self representation on behavior. Psychological
Science, 25, 1043-1045.
at UNIV TORONTO on January 9, 2015psp.sagepub.comDownloaded from
