Journal of Research in Personality 47 (2013) 15–21
Contents lists available at SciVerse ScienceDirect
Journal of Research in Personality
journal homepage: www.elsevier.com/locate/jrp
Posing personality: Is it possible to enact the Big Five traits in photographs?
Sointu Leikas ⇑, Markku Verkasalo, Jan-Erik Lönnqvist
Institute of Behavioural Sciences, Department of Psychology, University of Helsinki, Finland
a r t i c l e
i n f o
Article history:
Available online 7 November 2012
Keywords:
Person perception
Self-presentation
Impression management
Posing
Photographs
Personality
Big Five
a b s t r a c t
Person perception research has focused on the accuracy of observers receiving truthful target information; however, in real life people may often wish to manage the impression that they convey. We investigated whether people can ‘‘pose personality’’ in photographs. Sixty target participants posed each in 10
photographs in which they sought to express the high and low poles of the Big Five traits by means of
physical appearances. Observers (N = 401) rated targets’ personality and likability from each photograph.
The results showed that targets successfully posed as Extravert and, to lesser extent, as Introvert, Neurotic, Non-Conscientious, and Open, and that targets could not convey impressions of high and low
Agreeableness.
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
Over the last century, photography has evolved from a professional exercise into a commonplace activity accessible to everyone.
Despite becoming a part of everyday life, photographs have simultaneously maintained their position as the recorder of many of the
more exceptional moments or milestones of life. We have our wedding portraits, school yearbooks, family albums, driver’s licenses,
and passports, to name just a few. Most recently, the huge popularity of online social networks and online dating services has made it
highly likely that one will encounter photographs of a potential
romantic partner, friend, or an employee before meeting face to
face with this person. Thus, photographs are today more important
than ever, providing us with both a rich source of information
about others, as well as a means to convey information about ourselves to others. But how accurate is information about others that
is based on photographs? And, from another perspective, what
kind of images do we wish to convey to others via photographs,
and to what extent are we successful in creating such images?
To our knowledge, no previous research has investigated the extent to which specific personality impressions can be deliberately
created and conveyed by means of photographs. Previous research
on impression management has mostly focused on behavior in real
or mock job interviews and on the favorability of the impressions
created (e.g. Gilmore & Ferris, 1989; Schlenker & Weigold, 1992;
Stevens & Kristof, 1995). Such studies have provided important
information about the ways people attempt to create a positive
⇑ Corresponding author. Address: PO Box 9, Siltavuorenpenger 1 A, 00014
University of Helsinki, Finland.
E-mail address: sointu.leikas@helsinki.fi (S. Leikas).
0092-6566/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jrp.2012.10.012
impression in an interview setting, and about the success of such
attempts (for a meta-analysis, see Barrick, Shaffer, & DeGrassi,
2009). However, this research is silent on the question of how well
people are able to control more specific aspects of their public image beyond general favorability. Research asking people to explicitly convey a certain kind of image (e.g. sociable) has, with the
exception of faking studies using self-report questionnaire scores
as dependent variables (Konstabel, Aavik, & Allik, 2006; Lönnqvist,
Verkasalo, & Bezmenova, 2007), been practically nonexistent (for a
recent exception regarding the enactment of emotions, see Hall,
Gunnery, & Andrzejevski, 2011). Thus, it is somewhat unclear what
people do in order to appear, say, sociable, or open-minded, and
how well they are able to fulfill these types of impression management goals. Most relevant to the present research, it is unclear
whether personality impressions based on photographs are susceptible to impression management attempts.
Recent advances in personality perception research may shed
some light on the above presented questions. This research has
established that personality and behavioral outcomes can be somewhat accurately assessed on the basis of very little information of
the target (Ambady & Rosenthal, 1992; Borkenau & Liebler, 1992;
Borkenau, Mauer, Riemann, Spinath, & Angleitner, 2004; Carney,
Colvin, & Hall, 2007). Most pertinent to the present study, several
personality traits and behavioral outcomes can be correctly judged
even on the basis of a mere photograph of a target (e.g. Borkenau,
Brecke, Möttig, & Paelaecke, 2009; Naumann, Vazire, Rentfrow, &
Gosling, 2009; Rule & Ambady, 2011; Vazire, Naumann, Rentfrow,
& Gosling, 2008). Furthermore, person perception studies have
identified several appearance-related and behavioral cues that
are correlated with both actual personality traits and with observer-ratings of the same traits (e.g. Back, Schmukle, & Egloff,
16
S. Leikas et al. / Journal of Research in Personality 47 (2013) 15–21
2010; Borkenau & Liebler, 1992; Naumann et al., 2009). These results suggest that people’s naïve theories regarding the links between personality and behavior, and also between personality
and appearances, may be to some extent correct. If so, people could
be expected to be able to utilize this knowledge in order to create
and convey personality impressions on demand. On the other
hand, the correlational results obtained in the above mentioned
studies do not prove that observers are aware of the links between
cues and personality – the cues may be applied implicitly, without
conscious deduction from, say, smiling to high Extraversion. Furthermore, even if observers are aware of such links, they are not
necessarily able to use this knowledge to efficiently present themselves in the wanted manner.
Questions related to accuracy, self-other agreement, and cue
validity and utilization in person perception have been amply
studied (Back, Schmukle, & Egloff, 2011; Borkenau & Liebler,
1992; Borkenau et al., 2004; Letzring, Wells, & Funder, 2006; Mehl,
Gosling, & Pennebaker, 2006; Naumann et al., 2009). Because of the
interest in accuracy and visible manifestations of personality, most
of this research has been conducted with stimulus material in
which targets express themselves in natural and spontaneous
ways. The focus of such research has thereby been on the perspective of the observer; e.g., under what conditions are observers
accurate judges of personality, and what type of cues do they utilize as basis of their personality judgments? However, as noted
above, in real life people may often have the explicit or implicit
goal to convey a particular public image – an image that may or
may not correspond with their actual personalities. Therefore, an
important next step in person perception research is to incorporate
the perspective of the target; i.e., to study person perception in settings in which targets purposefully attempt to present themselves
in certain ways. This is the goal of the present research.
In the present study, we examine the enactment of the Big Five
traits in photographs. Each of the Big Five traits can be conceived of
as a bipolar continuum, and we ask our target participants to enact
each of the 10 poles of the five traits. We have two research questions in this study. First, are targets successful in their attempts to
enact the Big Five in photographs? Second, for which traits and to
what extent are they successful? Targets are expected to be somewhat successful for all traits. Although previous person perception
research has not decisively proven that people are aware of the
links between appearances and personality, other research lines
have shown that people are able to adapt their behavior smoothly
and even automatically to environmental demands (e.g. Chartrand
& Bargh, 1999). Such abilities could be expected to generalize to
the enactment of personality traits. It is also expected that targets
are more successful in enacting Extraversion than in enacting other
traits. Several studies have shown that Extraversion is the easiest
trait to judge (Borkenau et al., 2004; Carney et al., 2007; Connelly
& Ones, 2010), the most visible trait (e.g. Borkenau et al., 2009),
and the trait for which observers use the largest amount of valid
cues (Borkenau & Liebler, 1992; Naumann et al., 2009). Thus, it is
plausible that this trait is also easiest to enact.
2. Method
2.1. Target participants and procedure
Target participants (N = 60) were recruited via e-mail invitations sent to University of Helsinki student mailing lists. Thirty women and 30 men who replied to the invitation were asked to
complete an online personality questionnaire and to recruit ‘‘two
persons who knew them well’’ to complete the same online questionnaire in an informant-report format. After completion of the
questionnaires, these 60 target participants were invited to individual photograph sessions.
Upon arrival to the photograph session, target participants were
greeted by a male researcher (second author) who led them to a
studio. Targets were asked to stand in front of a white backdrop
that was identical for all targets and conditions. First, targets posed
freely for a half-body photograph (cut from waist up). This condition is hereafter referred to as the Neutral condition. Targets were
then instructed to enact, one at a time, all 10 poles of the Big Five
traits. Similar half-body photographs were taken in each of the10
posing condition. Targets were not allowed to add, change or remove clothing, hairbands, or decorative items, to remove or add
make-up, or to groom their hair between conditions. The instructions for each trait pole were of the form ‘‘try to appear as a person
whose personality is....’’ followed by a two-adjective description of
the trait pole. The adjectives were adapted from the Finnish version (Lönnqvist, Verkasalo, & Leikas, 2008) of the Ten Item Personality Inventory (Gosling, Rentfrow, & Swann, 2003). The adjectives
were anxious, distressed for the Neurotic condition; stable, calm for
the Stable condition; extraverted, enthusiastic for the Extravert condition; reserved, quiet for the Introvert condition; intellectually curious, daydreamer for the Open condition; conventional, does not like
change for the low-Openness condition; empathic, warm for the
Agreeable condition; critical, quarrelsome for the Disagreeable condition; dependable, self-disciplined for the Conscientious condition;
and unorganized, careless for the Non-Conscientious condition. Each
instruction was visible to the targets throughout the corresponding
posing condition. Targets did not receive any other advice for posing. The order of the posing conditions was as listed above (the order was the same for all targets).
Targets were, on average, 27.0 years old (SD = 5.48, range 19–
39), and all were Caucasian. As an incentive to participate, targets
received one complementary film ticket (value 9 €) and portrait
photographs. In addition, targets were given two film tickets to
give as compensation to the two informants who had provided
peer-ratings of personality.
2.2. Target personality
Self- and peer-reports of personality for the personality criterion were gathered using the Short Five personality questionnaire
(Konstabel, Lönnqvist, Walkowitz, Konstabel, & Verkasalo, 2012).
This 60-item measure was designed to measure the five factors
and 30 facets of the Five-Factor-Model (Costa & McCrae, 1992).
Each facet is measured by both a positively and negatively keyed
item (adding up to twelve items per personality factor), and each
item is responded to on a scale from 3 (The description is completely wrong) to 3 (The description is completely right). Sample
items include: ‘‘I am often nervous, fearful, and anxious, and I worry that something might go wrong’’ for the Anxiety facet of Emotional Stability, and ‘‘I do not like to associate with people much;
I am considered a rather cold and distant person’’ (reversed) for
the Warmth facet of Extraversion.
The internal consistency of target personality ratings was assessed with Cronbach’s alphas. The alpha reliabilities of the selfratings were .84, .91, .76, .76, and .87, for Extraversion (E), Emotional Stability (ES), Openness (O), Agreeableness (A), and Conscientiousness (C), respectively. After averaging the two sets of
peer-ratings, the alpha reliabilities of the peer-ratings were .83,
.91, .86, .81, and .90, for E, ES, O, A, and C, respectively. Pearson’s
correlation coefficients between self-ratings and averaged peerratings were r = .74, .61, .50, .47, .55, for E, ES, O, A, and C, respectively. Self-ratings and averaged peer-ratings were averaged to
form criteria scores of targets’ personality characteristics.
For comparison purposes, targets also rated their Big Five personality traits with the same five single items that were used to ob-
S. Leikas et al. / Journal of Research in Personality 47 (2013) 15–21
tain observer-ratings (see below). Correlations between these single-item scores and the criterion scores were .69, .74, .43, .57, and
.69 for E, ES, O, A, and C, respectively.
2.3. Observer participants and procedure
Observer participants (N = 401; 343 women) were students
from other Finnish universities (the Universities of Eastern Finland,
Jyväskylä, Tampere, and Turku) than that which the targets were
enrolled at (the University of Helsinki). Observers were, on average, 25.58 years old (SD = 7.42; range: 18–58). They were recruited
from courses on psychology research methods. Students on these
courses were asked to participate in an online study on social perception and were provided with the internet address of the study.
In that address, a link was provided from which observers had access to the photograph rating task.
Each observer rated 11 photographs, each depicting a different
target and a different posing condition. Each photograph (and,
thus, each target in each pose) was evaluated by 6–8 observers
(M = 6.68). Observers viewed the photographs one at a time, and
rated each target with bipolar single items measuring the Big Five
traits. The Big Five single items were constructed using the above
presented two-adjective descriptions that were given as instructions to the targets. In essence, for each trait, the descriptions of
the two poles were merged into one single bipolar item. The five
bipolar items, each rated on a seven-point scales anchored at both
ends with the two-adjective descriptions, thus constituted an
abbreviation of the 10 unipolar items of the Finnish language version of the Ten Item Personality Inventory (Gosling et al., 2003;
Lönnqvist et al., 2008). We used bipolar items because we wanted
to separately evaluate success in enacting the two poles of each
trait. Furthermore, in order to control for demand and acquiescence effects, the items were so keyed that higher scores meant
higher desirability for two out of five traits (Emotional Stability
and Agreeableness), and lower desirability for three out of five
traits (Extraversion, Openness, and Conscientiousness; i.e., higher
scores on these scales meant lower Extraversion, Openness, and
Conscientiousness). Ratings on the last three were reverse-scored
prior to analyses – in all analyses, high scores reflect the more
desirable pole of the trait.
In addition to the Big Five, observers evaluated the likability of
the targets with a single item ‘‘how much do you think you would
like this person’’; the observers rated this item on a seven-point
scale ranging from 1 (little) to 7 (very much). All rating scales appeared simultaneously below the photograph. Observers rated
the target on all six scales and scrolled down on the screen to move
to the next photograph. The single rater ICCs (1,1) were .42, .24,
.38, .00, .55, and .27, and the average raters ICCs (1, 6.68) were
.84, .75, .86, .18, .92, and .81 for ratings of Extraversion, Emotional
Stability, Openness, Agreeableness, Conscientiousness, and Likability, respectively. Note that there was practically no consensus for
ratings of Agreeableness; we return to this issue in Discussion.
2.4. Attractiveness ratings
As physical attractiveness is related to actual Extraversion and
Agreeableness (Meier, Robinson, Carter, & Hinsz, 2010), and to observer-ratings of several personality traits (e.g. Dion, Berscheid, &
Walster, 1972) we measured and controlled for the effects of
attractiveness. Ratings of targets’ attractiveness were based on
their free pose photographs. The photographs were presented,
one target at a time, in an online questionnaire, along with the
question ‘‘how attractive do you find this person’’ and a sevenpoint scale ranging from 1 (not at all attractive) to 5 (highly attractive). In order to ensure that judges would be unfamiliar with targets, judges were recruited amongst the authors’ overseas
17
acquaintances. A total of 13 judges rated the attractiveness of each
of the 60 targets. The mean of the attractiveness ratings was 2.93
(SD = 0.89). Interjudge reliability was high; the single rater ICC
(2, 1) and average rater ICC (2, 13) were .39 and .89, respectively
(Cronbach’s a = .89).
3. Results
Table 1 shows the means and standard deviations of the observer-ratings, and Table 2 shows the intercorrelations between target personality, target attractiveness, and observer-ratings based
on the neutral (free pose) photograph. As shown in Table 2, some
accuracy, as indexed by a significant correlation between personality criteria scores and observer-ratings based on the neutral photograph, was found for all traits except for Conscientiousness.
To examine targets’ success in expressing the Big Five trait
poles, first, a series of regressions was conducted with the SPSS
Mixed Models procedure using REML estimation. Observer-rated
trait scores were, each in turn, regressed on the 11-level (5
traits 2 poles + Neutral) posing condition variable, entered as a
fixed effect. Observations were nested within targets and the intercept was included, allowing the ratings to represent the distinctive
evaluation of a given target apart from normative ratings across
targets (i.e., the mean rating across targets was controlled for).
Condition was a significant predictor of observer-ratings for all
traits (Fs 6.16–39.87, all ps < .001) except Agreeableness (F = 0.65,
p = .77). Turning to our main research questions, we first examined,
one trait at a time, how successful targets were in enacting the two
trait poles (e.g., were targets rated as more extraverted in the
Extravert condition than in the Introvert condition?). Mean-level
differences between the high and low pole conditions were statistically significant (p < .05 or lower) and in the expected direction
for all traits except Agreeableness. That is, targets were evaluated
as more extraverted in the Extravert than in the Introvert condition
(Ms. 5.50 vs. 3.03, Cohen’s d = 2.32); as more emotionally stable
when posing as Stable than when posing as Neurotic (Ms. 4.53
vs. 3.48, d = 0.91); as more open in the Open than in the low-Openness condition (Ms. 5.04 vs. 4.03, d = 1.14), and as more conscientious in the Conscientious than in the Non-Conscientious
condition (Ms. 4.92 vs. 4.22, d = 0.75). However, targets were seen
as equally agreeable in the Agreeable and Disagreeable conditions
(Ms. 4.28 vs. 4.21, respectively, ns., d = 0.05).1
We also examined whether targets were rated by observers as
closer to the enacted trait pole in the posing conditions than in
the Neutral condition. The results showed that targets were rated
as more extraverted in the Extravert condition (d = 1.02), and less
extraverted in the Introvert condition (d = 1.00) than in the Neutral
condition. Targets were seen as less emotionally stable in the Neurotic than in the Neutral condition (d = 0.97), but observer-ratings
of Emotional Stability did not differ between the Stable and Neutral
conditions. Targets were rated as higher on Openness in the Open
1
We re-conducted the above described Mixed Models analyses controlling for
several possibly confounding variables. While controlling for the corresponding
personality criterion score, target attractiveness, and observer-rated likability, the
results remained identical (all condition effects p < .001). The actual personality trait
score also significantly predicted observer-ratings of this trait for Extraversion
(b = .18), Emotional Stability (b = .11), and Openness (b = .13). Target attractiveness
(based on independent ratings of the Neutral condition photo) predicted observerrated Extraversion (F = 16.40, p < .001, b = .28), Emotional Stability (F = 28.90, p < .001,
b = .28), and Openness (F = 4.45, p < .05, b = .16). Further, attractiveness predicted
observer-rated Likability (F = 25.79, p < .001, b = .31). We also checked whether target
sex or sex condition interaction affected the results. Target sex had a main effect on
observer-rated Emotional Stability (F = 7.85, p < .01) and Openness (F = 5.62, p < .05).
Men were rated higher on both traits (Emotional Stability: Ms = 4.43 vs. 4.08;
Openness: Ms = 4.43 vs. 4.22). No interactions between sex and condition were
detected. All main effects of condition remained significant (p < .001) when controlling for target sex and the sex condition interaction.
18
S. Leikas et al. / Journal of Research in Personality 47 (2013) 15–21
Table 1
Observer-Ratings by Posing Condition.
Trait
E
Condition
Neutral
Extravert
Introvert
Stable
Neurotic
Open
Low-Openness
Agreeable
Disagreeable
Conscientious
Non-Consc.
4.08
5.50
3.03
3.81
3.14
4.13
3.49
4.34
3.75
3.83
4.58
ES
(1.07)
(1.23)
(0.86)
(1.07)
(0.99)
(1.03)
(1.01)
(1.24)
(1.13)
(1.23)
(1.50)
4.68
5.03
3.66
4.53
3.48
4.59
3.96
4.73
3.72
4.31
4.49
O
(0.95)
(0.71)
(0.87)
(0.82)
(0.99)
(0.82)
(0.91)
(0.69)
(0.85)
(0.86)
(1.09)
4.45
4.79
4.08
4.41
4.11
5.04
4.03
4.69
3.87
4.13
4.71
A
(0.84)
(0.61)
(0.93)
(0.91)
(0.85)
(0.86)
(0.85)
(0.75)
(0.74)
(0.89)
(0.93)
C
4.46
4.27
4.43
4.47
4.28
4.41
4.36
4.28
4.21
4.20
4.35
(0.89)
(1.06)
(0.87)
(0.80)
(0.93)
(0.91)
(0.84)
(0.96)
(1.05)
(0.90)
(0.95)
4.73
4.50
4.62
4.77
4.45
4.71
4.86
4.74
4.65
4.92
4.22
LIKE
(0.88)
(0.92)
(0.83)
(0.84)
(0.93)
(0.82)
(0.94)
(0.82)
(0.86)
(0.88)
(0.93)
4.32
4.79
3.83
4.17
3.64
4.40
3.93
4.47
3.57
3.99
4.32
(0.83)
(0.58)
(0.66)
(0.72)
(0.68)
(0.70)
(0.77)
(0.76)
(0.77)
(0.77)
(0.83)
Note: All ratings were made on a scale from 1 to 7. Neutral = condition in which the targets posed freely. Non-Consc. = Non-Conscientious. E = Extraversion. ES = Emotional
Stability. O = Openness. A = Agreeableness. C = Conscientiousness. LIKE = Likability.
Table 2
Intercorrelations between target personality criteria scores, attractiveness, and observer-rated personality and likability in the Neutral condition.
1. E criterion
2. ES criterion
3. O criterion
4. A criterion
5. C criterion
6. E observer
7. ES observer
8. O observer
9. A observer
10. C observer
11. LIKE observer
12. Attractiveness
1.
2.
3.
4.
5.
1
.51***
.37**
.19
.17
.31*
.29*
.05
.04
.03
.24
.36**
1
.19
.43**
.52***
.14
.32*
.18
.04
.02
.14
.20
1
.14
.02
.15
.19
.36**
.24
.05
.17
.21
1
.39**
.08
.17
.02
.25*
.10
.17
.13
1
.09
.02
.05
.02
.13
.01
.02
6.
7.
8.
9.
10.
11.
1
.06
.01
.05
1
.35**
.28*
1
.55***
1
.72***
.40**
.06
.12
.59***
.52***
1
.48***
.10
.30*
.76***
.57***
1
.01
.05
.42**
.28*
Note: Observer-ratings are from the Neutral condition; i.e., the condition in which targets posed freely. Accuracy correlations (i.e., correlations between a personality criterion
and the corresponding observer-rating) are boldfaced. E = Extraversion. ES = Emotional Stability. O = Openness. A = Agreeableness. C = Conscientiousness. LIKE = likability.
Observer = observer-rating.
*
p < .05.
**
p < .01.
***
p < .001.
than in the Neutral condition (d = 0.62), and lower on Openness in
the low-Openness than in the Neutral condition (d = 0.45). Finally,
targets were rated as less Conscientious in the Non-Conscientious
condition than in the Neutral condition (d = 0.62), but not more
Conscientious in the Conscientious condition than in the Neutral
condition. Observer-ratings of Agreeableness did not differ between the Neutral condition and the corresponding posing
conditions.
The above results showed that targets were generally successful
in enacting both poles of the Big Five traits, with the exception of
Agreeableness. When instructed to express one pole of a trait, targets were actually seen as closer to this pole than to the opposite
pole. This was true for all posing conditions except the ones in
which targets were instructed to portray high or low Agreeableness (in both of these conditions, the targets were rated as similarly agreeable). Furthermore, in six out of the 10 posing
conditions, targets were seen as closer to the enacted pole than
in the Neutral condition. But how distinct were the impressions
that targets conveyed? For all of the Big Five traits, one pole is
more socially desirable than the other. It is thus possible that targets were only able to convey an indistinctive pleasant or positive
impression when instructed to enact the desirable pole of any trait,
and an indistinctive unpleasant impression when enacting the
undesirable pole.
To evaluate whether targets could express the trait poles distinctively, we asked three further questions. First, were targets,
when expressing a specific pole of a trait, rated as closer to this
pole than in any other posing condition? For example, were targets
rated as more emotionally stable in the Stable condition than in
any other condition? To answer this question, we returned to the
results of the above described analyses, but this time compared,
one trait at time, trait ratings obtained in the condition that corresponded to the trait with trait ratings in the remaining eight posing
conditions (e.g., Extraversion ratings in the Extravert condition
were compared with the Extraversion ratings in the Stable, Neurotic, Open, Low-Openness, Agreeable, Disagreeable, Conscientious, and Non-Conscientious conditions).
On an absolute level, in six out of the 10 conditions, ratings of
the corresponding trait were closer to the enacted pole than in
any other posing condition. However, only in the Extravert condition were targets rated as statistically significantly (p < .05) closer
to the enacted pole (i.e., as higher in Extraversion) than in any
other condition. Regarding the other conditions, targets were
rather successful in the Introvert condition (ratings of trait Extraversion were lower in this condition than in seven of the other conditions), Neurotic condition (ratings of trait Emotional Stability
were lower in this condition than in seven of the other conditions),
Open condition (ratings of Openness were higher in this condition
than in six of the other conditions), and Non-Conscientious (ratings
of Conscientiousness were lower in this condition than in seven of
the other conditions) conditions.
The second question that we asked in order to evaluate whether
targets could express the trait poles distinctively was whether the
instructions to portray the high vs. low poles of a given trait
S. Leikas et al. / Journal of Research in Personality 47 (2013) 15–21
affected the corresponding trait rating more than they affected ratings of other traits. For instance, was the difference in trait ratings
between the Open and low-Openness conditions larger for Openness ratings than for ratings of the other traits? To examine this,
we computed, for all five traits, effect sizes (Cohen’s ds) between
the mean ratings obtained when enacting the high and low poles
of each of the five traits. Comparing ds showed that between the
Extravert and Introvert conditions, ratings of Extraversion
(d = 2.31) differed more than ratings of other traits, although ratings of Emotional Stability (d = 1.29), and Openness (d = 0.84) also
differed strongly, unlike the ratings of Agreeableness (d = 0.13) and
Conscientiousness (d = 0.12). Between the Stable vs. Neurotic conditions, ratings of Emotional Stability (d = 0.91) differed more than
ratings of other traits (ds = 0.58, 0.30, 0.19, 0.36 for differences in
ratings of Extraversion, Openness, Agreeableness, and Conscientiousness, respectively). Between the Open vs. Low-Openness conditions, ratings of Openness (d = 1.14) differed more than the
ratings of other traits (ds 0.55, 0.04, 0.65, and 0.16 for ratings of
Extraversion, Agreeableness, Emotional Stability, and Conscientiousness, respectively). Between the Agreeable and Disagreeable
conditions, ratings of Emotional Stability (d = 1.08) and Openness
(d = 0.85) differed most strongly, followed by ratings of Extraversion (d = 0.47). The differences in ratings of Conscientiousness
(d = 0.11) and Agreeableness (d = 0.05) were very small. Finally, between Conscientious vs. Non-Conscientious conditions, ratings of
Conscientiousness (d = 0.75), Extraversion (d = 0.60), and Openness
(d = 0.61) differed more than ratings of Emotional Stability
(d = 0.18) or Agreeableness (d = 0.10). In sum, with the exception
of Agreeableness, the high vs. low pole condition effect always affected the corresponding trait most strongly, although this effect
was most pronounced for Extraversion, and not very pronounced
for Conscientiousness.
The third question we asked in order to investigate how distinctively targets could create specific impressions was whether the
targets were rated as closer to the trait pole that they were instructed to express than could have be expected based on their
self-rated personality. For instance, were the targets seen as more
extraverted in the Extravert condition than they actually were
according to their self-ratings? To answer to this question, we
compared single-item self-ratings on a given trait with observerrating of the same trait in the conditions in which targets were
posing the two ends of that trait. (e.g., targets’ self-rated extraversion was compared with observer-ratings of extraversion in the
Extravert and Introvert conditions). Single-items were used instead
of the personality criterion scores as the latter were obtained with
a different personality measure and mean levels were hence not
comparable with the observer-ratings.
Repeated-measures ANOVAs showed that in the Extravert condition, targets were seen as higher in Extraversion than could have
been expected based on their self-reports (F = 63.10, p < .001, Ms.
5.50 vs. 3.33). In the Introvert condition, observer-ratings did not
differ from self-ratings (F = 1.51, p = .22, Ms = 3.03 vs. 3.33). In
the Stable condition, observer-ratings did not differ from self-ratings (F = 0.31, p = .58, Ms = 4.53 vs. 4.42). In the Neurotic condition,
targets were seen as lower in Emotional Stability than their self-reports suggested (F = 14.62, p < .001, Ms. 3.48 vs. 4.42). In the Open
condition, targets were seen as less open than their self-ratings
suggested (F = 9.18, p < .01, Ms = 5.04 vs. 5.62; this may have been
due to the fact that the single-item self-ratings of Openness were,
on average, very high), and the same was true of the low-Openness
condition (F = 68.23, p < .001, Ms = 4.03 vs. 5.62). In both the Agreeable condition and Disagreeable conditions, targets were seen as
less agreeable than their self-ratings suggested (F = 4.78, p < .05,
Ms = 4.28 vs. 4.75 and F = 4.47, p < .05, Ms = 4.21 vs. 4.75, respectively). Finally, observer-ratings of conscientiousness did not differ
from target self-ratings in the Conscientious condition (F = 0.11,
19
p = 0.75, Ms = 4.92 vs. 4.85). However, in the non-Conscientious
condition, targets were seen as lower in Conscientiousness than
they reported being (F = 7.01, p < .05, Ms = 4.22 vs. 4.85). Thus, in
the Extravert, Neurotic, low-Openness, Disagreeable, and Non-Conscientious conditions, targets were seen as closer to the corresponding trait pole than their single-item self-ratings suggested.
To sum up the analyses regarding how distinctively the trait
poles could be portrayed, targets were clearly able to convey a distinct expression of high Extraversion. Targets were also able to create somewhat distinct impressions of Introversion, Neuroticism,
high Openness, and low Conscientiousness. However, distinctively
expressing low Openness, high Conscientiousness, and Emotional
Stability, proved difficult. Finally, according to all criteria, targets
were unable to convey impressions of being either Agreeable or
Disagreeable. Observers tended to confound ratings of Extraversion, Emotional Stability, Openness, and Likability, whereas ratings
of Agreeableness and Conscientiousness were distinct from these
and from each other.
4. Discussion
Physical appearance plays an important role in first impressions, and first impressions have far-reaching consequences for later judgments of a person. Thus, the extent to which people are
able to control and adjust the impressions they convey via their
physical appearance is of both applied and theoretical significance.
The present study examined the extent to which such control is
possible in the context of photographs. With everyone a Google
search away, first impressions of potentially important others are
increasingly likely to be based on photographs. The results suggested that it is possible to control the impressions of personality
in photographs. However, success in such impression management
depends on the particular trait in question, and also on the particular pole of the trait.
The results unambiguously suggest that people are very capable
of creating a distinct impression of high Extraversion in a photograph – regardless of their actual standing on this trait. Participants
were also moderately successful at expressing low Extraversion,
low Conscientiousness, low Emotional Stability, and high Openness
through physical appearances. Expressing high Emotional Stability,
low Openness, and high Conscientiousness distinctively was difficult, although targets were, in the corresponding conditions, seen
as closer to these trait poles than to the opposite trait poles. Finally,
creating impressions of high or low Agreeableness through appearances proved an unachievable task.
4.1. Posing personality: some poles of some traits are easier to pose
than others
That high Extraversion emerged as the clearly easiest trait pole
to enact is consistent with previous, more observer-focused, person perception research that has repeatedly shown that Extraversion is the most visible trait (Borkenau & Liebler, 1992; Connelly
& Ones, 2010), the easiest to judge (Beer & Watson, 2008; Carney
et al., 2007), and that it can be correctly evaluated on the basis
of a mere photograph (Borkenau et al., 2009; Naumann et al.,
2009). Our results complement previous literature by showing that
Extraversion (at least its high pole) is also the easiest trait for the
target to enact, on demand, via physical appearance. This may be
because physical cues related to high Extraversion are highly visible, and could thus be very familiar and recognizable to most
people.
To a lesser extent, the above considerations also apply to high
Openness that was, perhaps somewhat surprisingly, quite successfully enacted in the present study. Naumann et al. (2009) showed
20
S. Leikas et al. / Journal of Research in Personality 47 (2013) 15–21
that targets scoring high on Openness often look away from the
camera in photographs, and that looking away also predicts observer-ratings of Openness. Thus, it is possible that people are either
consciously or intuitively aware of some physical indicators of high
Openness, and use these indicators to enact and judge Openness,
either automatically or in a controlled manner. It should be
acknowledged, though, that targets reported very high Openness
levels, and observer-ratings of Openness in the Open condition
did not differ from targets’ self-ratings. Thus, it seems possible that
the targets’ success in posing as Open may be at least partly due to
targets’ actually high Openness levels – nevertheless, targets were
rated as higher in Openness in the Open condition than in the Neutral condition.
4.2. Is it impossible to enact Agreeableness?
Targets were rated as equally high in Agreeableness in all conditions, including the Disagreeable condition. Furthermore, observer-ratings of Agreeableness showed practically zero consensus,
suggesting that each observer made Agreeableness judgments idiosyncratically. Why? Targets did look different across posing conditions, as evidenced by targets’ success in posing some of the
traits. Despite this, the profile of the Agreeableness ratings remained flat over all posing conditions, did not resemble any of
the other trait profiles, and observer consensus was low for these
ratings. We could think of two possible explanations to this somewhat peculiar result. First, it seems possible targets misunderstood
the instructions to enact Agreeableness. Observer-ratings of Agreeableness and Likability were unrelated across conditions (average r
across posing conditions was .09), and this seems to support this
methodological explanation of the Agreeableness results. However,
ratings of Likability and Agreeableness were uncorrelated also for
the Neutral condition photograph (r = .01, ns.) – observers simply
did not combine Likability and Agreeableness.
The second possible explanation for the Agreeableness profile is
that observers were unable or unwilling to judge Agreeableness on
the basis of the cues available in photographs. At first glance, this
idea may appear at odds with results according to which people
are eager to judge others’ Agreeableness, even on the basis of as little information as Facebook profile photographs (Ames & Bianchi,
2008). However, photographs used in such social media profiles
may be much more informative regarding personality than the
portrait-style photographs used in the present study. Facebook
profile photographs (Ames & Bianchi, 2008) may contain a wide
variety of personality cues apart from physical appearance, such
as the background of the photograph, framing and cropping (e.g.
close-up vs. full-body photo), the location the photograph is taken
(e.g. outdoors vs. indoors), props, and the size of the target in the
photograph. By contrast, our targets all posed in waist-up photographs in front of a standardized background. It may be that
observers felt that they cannot evaluate strangers’ Agreeableness
on the basis of so little information.
The low consensus of Agreeableness ratings suggests that
observers did not base their ratings of Agreeableness on target
characteristics. This is consistent with previous research that has
shown that social projection (a particular type of perceiver effect
in person perception; Kenny, 1994) is stronger for Agreeableness
than for other traits (Beer & Watson, 2008; Wood, Harms, & Vazire,
2010). This means that observers tend to use their own levels of
Agreeableness level as proxies for targets’ Agreeableness. Such projection could also explain the present results: if observers projected their own level of Agreeableness into their ratings of
target Agreeableness, low consensus would be expected.
In sum, the results regarding Agreeableness suggest that
although people are highly interested in others’ Agreeableness
(Ames & Bianchi, 2008), they may be cautious in making strong
inferences of Agreeableness when information is very sparse. From
an applied perspective, this can be regarded as fortunate, because
it suggests that it is difficult to convey a false image of high Agreeableness, at least by means of physical appearances. However, it
should be noted that the photographs utilized in the present study
had a standardized form; it may be easier to fake Agreeableness in
photographs for which one can freely choose content and form.
How this could be done is an interesting question for future
research.
4.3. Limitations
The results presented here are the first to examine the purposeful creation of personality impressions by means of photographs,
and were based on a single study with only 60 targets. Thus, our
above offered interpretations of the results must be viewed as preliminary. Furthermore, targets were not allowed to change clothing, hairdo, or make-up between conditions. However, these
types of cues are indicative of personality. For instance, neatness
of dress is a valid and utilized physical cue for Conscientiousness
(Naumann et al., 2009), and in real-life settings people will probably try to dress neat when attempting to appear high in Conscientiousness (e.g., in a job interview). Thus, allowing for a change of
clothes between conditions would have provided a more realistic
impression creation setting (however, this would have happened
at the cost of introducing more ambiguity regarding unknown
and unobserved changes between conditions).
Target participants received somewhat limited instructions for
the posing (i.e., only two adjectives per posing condition). This
was due to time constraints; a comprehensive discussion of the
meaning and nature of personality traits would have prolonged
the photography sessions considerably. However, it is possible that
different targets understood the short instructions in different
ways, leading to different goals within a posing condition. In future
research, it would be desirable to instruct participants more extensively regarding the impressions that they are asked to create.
As a further limitation, the order of conditions in which the participants were photographed was kept constant. This may have affected the enactment of the trait poles expressed in the last (high
Conscientiousness, low Conscientiousness) or first (Emotional Stability, Neuroticism) conditions, through fatigue or initial nervousness, respectively.
4.4. Conclusions and future directions
People often wish to adjust and control the image that they convey to others, but this tendency has received relatively little attention in the literature. Given the hugely increased possibilities for
connecting with others online, the present findings according to
which people are, in portrait photographs, able to ‘‘fake’’ some of
their personality characteristics (e.g., high Extraversion, high
Openness), but not others (e.g., high Agreeableness) are important.
A relevant future research topic is how people create personality impressions in photographs. That is, what physical cues do targets display in order to pose different traits? Furthermore, is the
enactment of traits automatic or controlled? For instance, are the
targets aware of using particular cues? In a more applied vein, it
would be informative to investigate posing in photographs that
the targets are allowed to construct more freely. For instance, target participants could independently take the photographs using
similar instructions as those given in the present study. Such research setting would give a more comprehensive account of the
ways in which to convey personality impressions via photographs,
as well as a more realistic estimate of how successfully people can
use photographs to manage or manipulate the image that they create of themselves in social networking sites. Another interesting
S. Leikas et al. / Journal of Research in Personality 47 (2013) 15–21
topic would be the study of impression creation during behavioral
episodes. For instance, is it possible for a person low in Openness to
convey and image of high Openness in a dyadic discussion?
Finally, a potentially interesting avenue of research concerns
individual differences in both the ability to create specific personality impressions and the ability to see through such attempts. In
our view, the above described research lines would substantially
complement the person perception literature by bringing in the
target’s perspective.
Acknowledgments
Sointu Leikas, Jan-Erik Lönnqvist, and Markku Verkasalo, Institute of Behavioural Sciences, University of Helsinki, Finland. This
research was supported by the Academy of Finland (Grant Number
127641).
References
Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive behaviors as predictors
of interpersonal consequences: A meta-analysis. Psychological Bulletin, 111,
256–274.
Ames, D. R., & Bianchi, E. C. (2008). The agreeableness asymmetry in first
impressions: Perceiver’s impulse to (mis)judge agreeableness and how it is
moderated by power. Personality and Social Psychology Bulletin, 34, 1719–1736.
Back, M. D., Schmukle, S. C., & Egloff, B. (2010). Why are narcissists so charming at
first sight? Decoding the narcissism-popularity 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 personality and interpersonal attraction at zero
acquaintance. European Journal of Personality, 25, 225–238.
Barrick, M. R., Shaffer, J. A., & DeGrassi, S. W. (2009). What you see may not be what
you get: Relationships among self-presentation tactics and ratings of interview
and job performance. Journal of Applied Psychology, 94, 1394–1411.
Beer, A., & Watson, D. (2008). The effects of information and exposure on self-other
agreement. Journal of Research in Personality, 44, 38–45.
Borkenau, P., Brecke, S., Möttig, C., & Paelaecke, M. (2009). Extraversion is accurately
perceived after a 50-ms exposure to a face. Journal of Research in Personality, 43,
703–706.
Borkenau, P., & Liebler, A. (1992). Trait inferences: Sources of validity at zero
acquaintance. Journal of Personality and Social Psychology, 62, 645–657.
Borkenau, P., Mauer, N., Riemann, R., Spinath, F. M., & Angleitner, A. (2004). Thin
slices of behavior as cues of personality and intelligence. Journal of Personality
and Social Psychology, 86, 599–614.
Carney, D. R., Colvin, C. R., & Hall, J. A. (2007). A thin slice perspective on the
accuracy of first impressions. Journal of Research in Personality, 41, 1054–1072.
Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The perceptionbehavior link and social interaction. Journal of Personality and Social Psychology,
76, 893–910.
21
Connelly, B. S., & Ones, D. S. (2010). An other perspective on personality: Metaanalytic integration of observers’ accuracy and predictive validity. Psychological
Bulletin, 136, 1092–1122.
Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and
NEO Five-Factor Inventory (NEO-FFI). Odessa, FL: Psychological Assessment
Resources.
Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of
Personality and Social Psychology, 24, 285–290.
Gilmore, D. C., & Ferris, G. R. (1989). The effects of applicant impression
management tactics on interviewer judgments. Journal of Management, 15,
557–564.
Gosling, S. D., Rentfrow, P. J., & Swann, W. B. Jr., (2003). A very brief measure of the
Big-Five personality domains. Journal of Research in Personality, 37, 504–528.
Hall, J. A., Gunnery, S. D., & Andrzejevski, S. A. (2011). Nonverbal emotion displays,
communication modality, and the judgment of personality. Journal of Research
in Personality, 45, 77–83.
Kenny, D. A. (1994). Interpersonal perception. New York: Guilford.
Konstabel, K., Aavik, T., & Allik, J. (2006). Social desirability and consensual validity
of personality traits. European Journal of Personality, 20, 549–566.
Konstabel, K., Lönnqvist, J.-E., Walkowitz, G., Konstabel, K., & Verkasalo, M. (2012).
The ‘Short Five’ (S5): Measuring personality traits using comprehensive single
items. European Journal of Personality, 26, 13–29.
Letzring, T. D., Wells, S. M., & Funder, D. C. (2006). Information quantity and quality
affect the realistic accuracy of personality judgment. Journal of Personality and
Social Psychology, 91, 111–123.
Lönnqvist, J. E., Verkasalo, M., & Bezmenova, I. (2007). Agentic and communal bias
in socially desirable responding. European Journal of Personality, 21, 853–868.
Lönnqvist, J.-E., Verkasalo, M., & Leikas, S. (2008). Viiden suuren
persoonallisuusfaktorin 10, 60, ja300 osion julkiset mittarit [10, 60, and 300
item public domain measures of the Big Five]. Psykologia, 5, 328–341.
Mehl, M. R., Gosling, S. D., & Pennebaker, J. W. (2006). Personality in its natural
habitat: Manifestations and implicit folk theories of personality in daily life.
Journal of Personality and Social Psychology, 90, 862–877.
Meier, B. P., Robinson, M. D., Carter, M. S., & Hinsz, V. B. (2010). Are sociable people
more beautiful? A zero-acquaintance analysis of agreeableness, extraversion,
and attractiveness. Journal of Research in Personality, 44, 293–296.
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.
Rule, N. O., & Ambady, N. (2011). Judgments of power from college yearbook photos
and later career success. Social Psychological and Personality Science, 2, 154–158.
Schlenker, B. R., & Weigold, M. F. (1992). Interpersonal processes involving
impression regulation and management. Annual Review of Psychology, 43,
133–168.
Stevens, C. K., & Kristof, A. L. (1995). Making the right impression: A field study of
applicant impression management during job interviews. Journal of Applied
Psychology, 80, 587–606.
Vazire, S., Naumann, L. P., Rentfrow, P. J., & Gosling, S. D. (2008). Portrait of a
narcissist: Manifestations of narcissism in physical appearance. Journal of
Research in Personality, 42, 1439–1447.
Wood, D., Harms, P., & Vazire, S. (2010). Perceiver effects as projective tests: What
your perceptions of others say about you. Journal of Personality and Social
Psychology, 99, 174–190.