Per 2076

European Journal of Personality, Eur. J. Pers.
(2016)
Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/per.2076
Predicting Self-Confident Behaviour with Implicit and Explicit Self-Esteem

Measures
SASCHA KRAUSE1*, MITJA D. BACK2, BORIS EGLOFF3 and STEFAN C. SCHMUKLE1

1
University of Leipzig, Germany
2
University of Münster, Germany
3
Johannes Gutenberg-University Mainz, Germany
Abstract: The present research compared the validity of popular direct and indirect measures of self-esteem in
predicting self-confident behaviour in different social situations. In line with behavioural dual-process models, both
implicit and explicit self-esteem were hypothesized to be related to appearing self-confident to unacquainted others.
A total of 127 participants responded to the Rosenberg Self-Esteem Scale, the Multidimensional Self-Esteem Scale,
and an adjective scale for measuring explicit self-esteem (ESE). Participants’ implicit self-esteem (ISE) was assessed
with four indirect measures: the Implicit Association Test (IAT), the name-letter task (NLT), and two variants of an
affective priming task, the reaction-time affective priming task (RT-APT) and the error-based affective priming task
(EB-APT). Self-confident behaviour was observed in four different social situations: (i) self-introduction to a group;
(ii) an ostracism experience; (iii) an interview about the ostracism experience; and (iv) an interview about one’s
personal life. In general, appearing self-confident to unknown others was independently predicted by ESE and ISE.
The indirect measures of self-esteem were, as expected, not correlated, and only the self-esteem APTs—but not the
self-esteem IAT or the NLT—predicted self-confident behaviours. It is important to note that in particular the
predictive power of the self-esteem EB-APT pertained to all four criteria and was incremental to the ESE measures.
Copyright © 2016 European Association of Personality Psychology
Key words: implicit self-esteem; actual behaviour; predictive validity; Implicit Association Test; name-letter task;
affective priming task
The idea that automatically activated self-evaluations (i.e. unacquainted others. Furthermore, by employing established
implicit self-esteem, ISE) can have self-relevant behavioural self-reports of self-esteem (i.e. explicit self-esteem, ESE), we
consequences is popular in personality psychology. were additionally able to examine the incremental validity of
However, recent research has critically questioned the the ISE measures above and beyond the ESE measures.
existence and the relevance of the ISE construct mainly
because of low and inconsistent predictive validities across
studies and ISE measures (Buhrmester, Blanton, & Swann, SELF-CONFIDENT BEHAVIOUR AND
2011; Falk, Heine, Takemura, Zhang, & Hsu, 2015). A SELF-ESTEEM
conclusive evaluation of the predictive validity of ISE is
complicated by the facts that (i) only a few studies have Behaving in a self-confident manner in the presence of
indeed analysed actual behaviours as criterion measures, unacquainted others has many positive consequences for a
and (ii) with regard to behavioural prediction, ISE measures person’s social life. In zero-acquaintance situations, people
have typically been investigated in isolation from each other. who exhibit self-confidence receive more positive evalua-
In the present study, we aimed to compare the predictive tions from others and are thus more popular (Back,
power of the three most prominent classes of ISE measures Schmukle, & Egloff, 2011). Moreover, speed-dating studies
—the Implicit Association Test (IAT; Greenwald & suggest that the self-confident behaviour of flirting with
Farnham, 2000), the name-letter task (NLT; Kitayama & strangers is particularly advantageous for mate selection:
Karasawa, 1997), and the affective priming task (APT; People who flirt more with speed-dating partners are chosen
Krause, Back, Egloff, & Schmukle, 2011; Spalding & more often as potential partners (Back, Penke et al., 2011).
Hardin, 1999)—for an outcome that is (i) theoretically asso- Appearing self-confident to unacquainted others is also an
ciated with self-esteem; (ii) derived from the assumption that important predictor of job success: Sales performance, for
ISE measures should mainly predict actual behaviours; (iii) example, is associated with self-confidence (Miner, 1962).
assessed in different situations; and (iv) of high importance Furthermore, in employment interviews, a confident and
in social life: self-confident behaviour in the presence of powerful communication style is an advantage when appl-
ying for a job (Imada & Hakel, 1977; Lammers, Dubois,
*Correspondence to: Sascha Krause, Department of Psychology, University
of Leipzig, Neumarkt 9-19, 04109 Leipzig, Germany. E-mail: sascha. Rucker, & Galinski, 2013; Young & Beier, 1977). A lack
krause@uni-leipzig.de of self-confident behaviour (i.e. appearing insecure or shy
Received 12 April 2016

Copyright © 2016 European Association of Personality Psychology Revised 31 August 2016, Accepted 6 September 2016
S. Krause et al.
to unacquainted others) can have negative consequences for triggered by ISE and controlled (i.e. reflective) processes that
a person’s social life. Insecure or shy people are initially should be activated by ESE. Accordingly, observations of ac-
judged as less intelligent (Paulhus & Morgan, 1997). More- tual self-confident behaviour in the presence of unacquainted
over, shyness is inversely related to positive outcomes such others should be predicted by both ISE and ESE measures.
as dating frequency and satisfaction, number of friends, and
level of self-disclosure to friends (Jones & Briggs, 1984).
Intuitively, in new and unpredictable social situations, CANDIDATE MEASURES
appearing self-confident to others should be closely related
to self-esteem. Support for this assertion is provided by two There are a large number of measures that are used to capture
complementary perspectives on the dynamics of self-esteem: self-esteem. Whereas the selection of reliable ESE measures
First, the self-broadcasting perspective ‘implies that an is manifold, there are currently only three prominent domains
individual acts out his or her self-evaluations in observable of ISE measures that have demonstrated both reliability and
social behaviour’ (Srivastava & Beer, 2005). Thus, when validity: IATs, NLTs, and variants of the APT. Single studies
encountering strangers, people high in self-esteem behave on each one of the three classes of ISE measures have
confidently, whereas people low in self-esteem behave in provided support for their predictive and incremental
an insecure way. Second, the sociometer theory of self- validities, however, mostly for self-report outcomes (for an
esteem (Leary, Tambor, Terdal, & Downs, 1995)—which overview, see Buhrmester et al., 2011). The self-esteem
assumes that self-esteem is developed by social feedback IAT has been reported to be positively correlated with self-
processes—comes to the same conclusion: Self-confident reported criteria such as subjective well-being (Walker &
behaviour contributes to the experience of more positive Schimmack, 2008), depression (Haeffel et al., 2007),
evaluations by other people, and this fosters the development neuroticism (Robinson & Meier, 2005), the perception of
of higher self-esteem (e.g. Denissen, Penke, Schmitt, & Van being valued (Back, Krause et al., 2009), performance per-
Aken, 2008; Hutteman, Nestler, Wagner, Egloff, & Back, ceptions, and expectations after failure (Greenwald &
2015). Both the self-broadcasting and the sociometer Farnham, 2000). The NLT has shown significant correlations
perspectives are supported by findings that have shown that with various self-report criteria, including psychological
people with high self-esteem describe themselves as more well-being (Kernis, Lakey, & Heppner, 2008), life satisfac-
sociable, outgoing, and assertive, as well as less shy than tion (Kernis et al., 2008), depression (Franck, De Raedt, &
people with low self-esteem (MacDonald & Leary, 2012). De Houwer, 2007), self-concept clarity (DeHart, Pelham, &
Tennen, 2006), relationship satisfaction (LeBel & Campbell,
2009), and Internet addiction (Stieger & Burger, 2010).
A DUAL-PROCESS APPROACH TO THE RELATION Finally, two variants of self-esteem APTs—either based
BETWEEN SELF-ESTEEM AND SELF-CONFIDENT on reaction times or are based on errors made under time
BEHAVIOUR pressure—have been reported to be significantly associated
with self-report outcomes. Whereas the reaction-time APT
Traditionally, self-esteem researchers have used ESE (RT-APT) predicted differences in the self-evaluations of
measures (i.e. by asking participants to self-report how they people with different cultural backgrounds (Hetts, Sakuma,
evaluate themselves) to examine the effects of self-esteem & Pelham, 1999) and the perception of being valued (Back,
on actual self-confident behaviour in the presence of unac- Krause et al., 2009), the error-based APT (EB-APT) was sig-
quainted others. Although these deliberate evaluations of nificantly correlated with self-reported self-esteem (Wentura,
the self certainly capture important aspects of self-esteem, Kulfanek, & Greve, 2005) and performance perceptions and
they are also subject to the well-known limitations of self- expectations before and after failure (Krause, Back, Egloff,
reports, including impression management tendencies & Schmukle, 2012). Although the EB-APT was shown to
(Paulhus, 1984) and the limits of self-awareness (Greenwald be more reliable than the RT-APT (Krause et al., 2011), we
& Banaji, 1995). Moreover, self-reports of self-esteem are used both the RT-APT and the EB-APT in the present study,
not able to capture automatic self-evaluations (i.e. ISE) that as previous research has suggested that both variants of
are spontaneously activated in response to self-esteem- APTs are able to validly assess ISE.
relevant stimuli. The distinction between different represen- Much less is known about how actual behaviour corre-
tations of the self (i.e. ISE and ESE) is in accordance with sponds with these ISE measures. Table 1 provides an over-
dual-process models of personality and information process- view of studies that investigated relations between actual
ing (e.g. Back & Nestler, in press; Fazio, 1990; Gawronski & behaviours and self-esteem. As can be seen, variants of
Bodenhausen, 2006; Greenwald et al., 2002; Strack & self-esteem IATs have been shown to correlate with an ex-
Deutsch, 2004; Wilson, Lindsey, & Schooler, 2000). perimenter rating of self-esteem (Bluemke & Friese, 2012),
The distinction between implicit and explicit modes of an interviewer rating of negative affect during a mental
information processing is important not only for evaluations health interview (Robinson & Meier, 2005), an observer rat-
but also for human behaviour. According to behavioural ing of facial and body adaptors during a public speaking task
dual-process models (Back, Schmukle, & Egloff, 2009; Fazio (Rudolph, Schröder-Abé, Riketta, & Schütz, 2010), and per-
& Olson, 2003; Strack & Deutsch, 2004), people’s self- sistence and performance on scrambled letter tasks (Holland,
confident behaviour in social situations is determined by Wennekers, Bijlstra, Jongenelen, & van Knippenberg, 2009).
both automatic (i.e. spontaneous) processes that should be The NLT was significantly associated with defensive verbal
Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

DOI: 10.1002/per
Predicting self-confident behaviour with self-esteem
Table 1. Relations between actual behaviours and self-esteem measures
Reference Actual behaviours N Nobserver IAT NLT APT ESE
Bluemke and Friese (2012) Experimenter rating of self-esteem 144 1 .24** — — .20*
Robinson and Meier (2005) Interviewer rating of negative affect 54 1 .40* — — .23
Rudolph et al. (2010) Facial and body adaptors 49 1 .29* — — .08
Rudolph et al. (2010) Illustrators, control of posture and speech 49 1 .15 — — .35*
Holland et al. (2009) Persistence in scrambled letter tasks 42 — .54* — — —
Holland et al. (2009) Performance in scrambled letter tasks 32 — .80** — — —
Kernis et al. (2008) Defensive verbal behaviour 101 1 — .56** — .26**
Spalding and Hardin (1999) Interviewer rating of nonverbal anxiety 64 1 — — .39** .14
Vandromme et al. (2011) Gaze avoidance 24 — — — .55** .30
Vandromme et al. (2011) Gaze duration 23 — — — .42* .01
Note: N = number of participants in the study. Nobserver = number of observers of behaviour. IAT = Implicit Association Test; NLT = name-letter task;
APT = affective priming task; ESE = explicit self-esteem measures.
*p < 05.
**p < .01.
behaviour during an interview (Kernis et al., 2008). reports of self-esteem are—as ISE measures are as well—
Reaction-time based variants of self-esteem APTs have also able to predict actual behaviours in different social contexts.
been shown to correlate with actual behaviours, specifically Taken together, there is little research on the prediction of
with an interviewer rating of nonverbal anxiety during an actual self-esteem-related behaviour with ISE and ESE
emotional health interview (Spalding & Hardin, 1999), gaze measures. Moreover, as can be seen in Table 1, there is a lack
avoidance, and the time spent looking at the experimenter of knowledge about how different measures of ISE and ESE
during instructions (Vandromme, Hermans, & Spruyt, 2011). simultaneously predict identical actual behaviours. In all
Undoubtedly, the most widely used measure of ESE is identified studies, only one measure of ISE and one measure
the Rosenberg Self-esteem Scale (RSES; Rosenberg, 1965). of ESE were applied. Whereas this poses no problem for ESE
The RSES assesses global self-esteem by asking participants measures, as they are highly correlated with each other (e.g.
to respond to five positive and five negative statements about Zeigler-Hill, 2010), for ISE measures, researchers have yet to
the self. Other ESE measures include, for example, the Self- determine the extent to which the different but uncorrelated
Rating Scale (Fleming & Courtney, 1984) or adjective scales assessment procedures (see, e.g. Bosson, Swann, &
(e.g. Brown & Kobayashi, 2002). Previous studies have re- Pennebaker, 2000; Krause et al., 2011; Rudolph,
peatedly confirmed the predictive validity of ESE measures Schröder-Abé, Schütz, Gregg, & Sedikides, 2008) are able
(particularly the RSES) for several consequential outcome to predict the same behavioural outcomes. Previous studies
variables including subjective well-being (e.g. Schimmack that have applied more than one ISE measure to compare
& Diener, 2003), depression (e.g. Murrell, Meeks, & Walker, their ability to predict the same criteria have used self-re-
1991), physical health (e.g. DeLongis, Folkman, & Lazarus, port-based variables as outcomes (e.g. Back, Krause et al.,
1988), and relationship satisfaction (Erol & Orth, 2013). 2009; Bosson et al., 2000; Falk et al., 2015; Verplanken,
As is common in personality and social psychology, re- Friborg, Wang, Trafimow, & Woolf, 2007).
search with ESE measures is mainly focused on self-report Furthermore, Table 1 shows that previous research on
outcomes (for criticisms of this practice, see, e.g. Baumeister, ISE and its relation to actual behaviour is further limited by
Vohs, & Funder, 2007; Furr, 2009). Therefore, actual beha- small sample sizes or small numbers of behavioural
vioural studies are rare. In the last column of Table 1, we observers. Specifically, only two studies had more than 100
added the predictive validities of various ESE measures that participants (Bluemke & Friese, 2012; Kernis et al., 2008),
were used in the above-mentioned ISE research. As can be and no study had more than one observer who had rated
seen, self-reports of self-esteem have been reported to be sig- the behaviour of all participants (see the fourth column of
nificantly associated with an experimenter rating of self- Table 1). Notably, the question of incremental validity (i.e.
esteem (row 2; Bluemke & Friese, 2012); observer ratings whether ISE measures predict behavioural outcomes above
of illustrators, the ability to control posture, and the ability and beyond ESE measures) needs to be addressed because
to control speech during a public speaking task (row 5; future self-esteem research will benefit only from ISE
Rudolph et al., 2010); and defensive verbal behaviour during measures that are able to explain substantial amounts of var-
an interview (row 8; Kernis et al., 2008). Notably, the effect iance in behavioural outcomes above and beyond standard
sizes for the interviewer rating of negative affect during a self-esteem self-reports. Please note that the two previous
mental health interview (row 3; Robinson & Meier, 2005) studies that found both significant ESE and significant ISE
and avoiding the gaze of the experimenter during instructions correlations with the same behavioural outcomes (see
(row 10; Vandromme et al., 2011) were at least as high as Table 1, Bluemke & Friese, 2012; Kernis et al., 2008) did
those reported above but not significant (probably because not explicitly address the question of incremental validity.
of small sample sizes). Altogether, previous research that Last but not least, the question of whether both kinds of
employed ESE measures has provided evidence that self- self-esteem predict actual behaviours consistently across

DOI: 10.1002/per
S. Krause et al.
different social contexts is still unresolved. As can be seen in in the study, including a more detailed description of the
Table 1, previous research that predicted behaviour with ISE procedure and additional measures. Moreover, the data and
measures focused on only one social situation. Because we analysis scripts of our study can be downloaded from
aimed to investigate the robustness of ISE-to-behaviour https://osf.io/fqc4a/.
effects, not only did we include multiple measures of ISE,
but we also created different social situations in which actual
behaviour was obtained. As an outcome measure, we used a Participants
criterion that is theoretically associated with self-esteem: One hundred thirty-one unacquainted1 students from diffe-
self-confident behaviour in the presence of unacquainted rent fields of study participated in exchange for research
others. participation credit or monetary compensation (20€). The
first group of four participants served as a training group to
(i) test the study procedure; (ii) acquaint the experimenter
THE PRESENT RESEARCH with the study procedure; (iii) obtain video examples of
self-confident behaviours for the observer trainings; and
For the present study, we designed a real-life context invol- (iv) optimize the study procedure after interviewing the four
ving first-time meetings; that is, group members did not training group members about their experiences in the study.
know each other before the study. To examine whether ISE The data from this training group were excluded from the
and ESE would affect self-confident behaviour in different analyses. The average age of the remaining 127 participants
authentic social situations, we designed four realistic con- (70 women) was 24.1 years (SD = 3.4). Participants belonged
texts: Participants were first asked to introduce themselves to 32 groups, each consisting of four members.2
to a group of strangers. Second, in order to explore partici-
pants’ levels of self-confidence when experiencing social
exclusion by other group members, participants played the Procedure
ostracism game ‘Cyberball’ (Williams, Cheung, & Choi, Before group sessions, participants completed three online
2000). Third, participants were individually interviewed measures of ESE in a fixed order. After filling out the well-
about their ostracism experiences by an unacquainted established RSES, participants responded to the German
experimenter. Fourth, participants had to speak with the version of Fleming and Courtney’s (1984) Self-Rating Scale
interviewer about positive and negative aspects of their (Multidimensional Self-Esteem Scale; MSES; Schütz &
personal lives (cf. Spalding & Hardin, 1999). For each of Sellin, 2006) and an adjective scale (AS) containing the
the four different outcomes, we obtained trained observer target words used in the ISE measures. In addition to these
ratings of actual behaviour, thus allowing us to investigate self-report measures, participants had to fill out several other
the degree to which participants behaved self-confidently in online questionnaires (for more information on these ques-
the eyes of uninvolved others. tionnaires, see the SOM).
Dual-process models of social information processing After arriving at the laboratory, participants were
represent social behaviour as not driven by purely automatic photographed in a standardized position with a neutral facial
or purely controlled processes but rather as the result of both expression to create face photos for relevant tasks. Next,
kinds of processes (cf. Back et al., 2009; Fazio & Olson, participants were asked to stand in front of the group and
2003; Strack & Deutsch, 2004). On the basis of such models, briefly introduce themselves. Afterwards, participants were
we expected that observer ratings of self-confident behaviour seated in separate cubicles to complete four ISE measures
in the presence of unacquainted others would be affected by in a fixed order (i.e. the IAT, two APTs (i.e. the RT-APT
both the ISE and ESE of targets. Thus, both types of self- and the EB-APT), and the NLT, respectively) on a personal
esteem were expected to demonstrate predictive validity for computer. Attribute or target adjectives for the different
the four different outcomes (see Table 1 for similar results reaction-based ISE measures were selected according to their
involving other self-esteem-related behaviours). Moreover, valences from German norm tables (Hager & Hasselhorn,
because ISE and ESE measures capture different aspects of 1994). All ISE measures were administered on personal
self-esteem (e.g. Back, Krause et al., 2009; Bosson et al., computers using Inquisit software (2006). Instead of a
2000; Greenwald & Farnham, 2000; Spalding & Hardin, standard computer keyboard, participants used external
1999), using both types of measures together was expected response pads to complete the reaction-time-based ISE
to improve the prediction of self-confident behaviour (i.e. in- measures. This technical detail was applied to reduce error
cremental validity of ISE above and beyond ESE). variance in the recorded latencies because key presses on a
standard computer keyboard are buffered in the keyboard
hardware before they are transmitted to the computer (Voss,
METHOD Leonhart, & Stahl, 2007).
In this section, we provide an overview of the procedures and 1

After completing the ISE measures, each participant had to indicate how
measures that are relevant to the main aim of the study (i.e. well she/he was acquainted with each group member before the study (6-
predicting self-confident behaviour with ISE and ESE point Likert scale; 1 = not a bit, 6 = very well; M = 1.05; SD = 0.22).
2
Please note that in one session, an experimenter had to replace a missing
measures). The supplementary online materials (SOM) pro- participant. Excluding this session’s three participants did not change the
vide a complete account of all variables that were assessed pattern of results.

DOI: 10.1002/per
Next, participants were separately escorted to individual Formann, 2012). The likeability ratings of a participant’s
rooms. Here, participants first played the ostracism game first initial served as a measure of ISE because a recent
Cyberball (Williams et al., 2000) on a personal computer. meta-analysis found that individuals’ ratings of their first
Thereafter, the experimenter entered the room and initial were more closely related to personal self-esteem
interviewed each participant about the ostracism experience. than individuals’ ratings of their last initial (Stieger et al.,
Then, participants were asked to tell the interviewer about 2012).
positive and negative aspects of their personality. Finally, The first initial preference score was computed by appl-
participants were debriefed, thanked, and given course credit. ying the ipsatized double-correction algorithm (Baccus,
To examine whether ISE and ESE would affect self- Baldwin, & Packer, 2004), which is recommended for both
confident behaviour, we employed four different realistic so- theoretical and empirical reasons (LeBel & Gawronski,
cial contexts: (i) self-introduction to a group; (ii) the ostra- 2009). Separately for the first and second name-letter rating
cism experience; (iii) the interview about the ostracism rounds, the average of a participant’s ratings of all non-initial
experience; and (iv) the interview about participants’ letters was subtracted from the participant’s rating of his or
personal lives. her first initial. Next, normative letter baselines—which were
computed by averaging the ipsatized letter ratings for indi-
viduals whose first initial did not include the letter—were
Implicit self-esteem (ISE) measures
subtracted from this difference. The mean of the first and
Implicit association test (IAT) second rounds of the first initial preference score served as
The self-esteem IAT is a computerized categorization task a measure of ISE.
that measures automatic associations of self-relevant and
non-self-relevant words by presenting pleasant and unpleas-
ant words. Five self-related stimuli (I, me, my, own, self) Error based affective priming task (EB-APT)
and other-related stimuli (other, their, those, you, your) were The self-esteem EB-APT is a computerized categorization
presented, as well as five attribute stimuli representing posi- task that quantifies the degree to which exposure to self-
tive (beautiful, cheerful, honest, liked, sincere) and negative related versus other-related primes facilitates the subsequent
(arrogant, cowardly, mean, ugly, vicious) categories. The categorization of words with unambiguous valences into a
underlying assumption of the IAT is that if two category con- positive and negative category, respectively. Because they
cepts are highly associated (e.g. self-concept and positivity in experience positive affect following self-related stimuli,
the case of a person with high self-esteem), the sorting task people with high self-esteem are thought to be faster (or
will be easier (i.e. faster) when the two associated category show more correct reactions) when a positive stimulus is pre-
concepts share the same response key (i.e. self-positive vs. ceded by a self-related prime and slower (or show more false
other-negative) than when they share different response keys reactions) when a negative stimulus is preceded by a self-
(i.e. self-negative vs. other-positive). related prime, compared with people with low self-esteem.
The IAT procedure followed the five-block structure de- Building on research showing that using pictures of faces
scribed by Nosek, Greenwald, and Banaji (2005). In the as primes in an APT provides a reliable and valid measure of
fourth block, participants had to perform twice as many ISE (e.g. Krause et al., 2011, 2012), pictures of participants’
single-categorization trials (40) as in the first two practice faces (500 × 500 pixels) were presented as self-related
blocks to reduce undesirable order effects from the combined primes. Pictures of the faces of unacquainted persons served
judgement blocks (Nosek et al., 2005). The critical Blocks 3 as other-related primes. During the task, participants were
and 5 consisted of 80 trials each. Stimulus presentation alter- asked to identify unambiguously valenced target adjectives
nated between the target and the attribute dimensions. The as pleasant (beautiful, cheerful, honest, liked, sincere) or
critical block order was held constant (i.e. the Self + Pleasant unpleasant (arrogant, cowardly, mean, ugly, vicious) as
block was followed by the Self + Unpleasant block) to min- quickly and accurately as possible.
imize the effect of procedural variations on the measurement On the basis of evidence that the analysis of errors—in-
of individual differences in self-esteem. The self-esteem IAT stead of reaction times—substantially improves the reliabil-
scores were computed with an improved scoring algorithm ity and validity of self-esteem APT effects (Krause et al.,
(D1 measure) described by Greenwald, Nosek, and Banaji 2011, 2012; Wentura et al., 2005), we used the response-
(2003). deadline technique (Degner, Wentura, Gniewosz, & Noack,
2007) to obtain a sufficient number of errors in the priming
Name-letter task (NLT) task. Each judgement trial began with the presentation of a
The NLT is based on the well-documented finding that prime, which remained on the screen for 67 ms and was
people prefer the letters of their own names, especially their immediately replaced by a blank screen for 33 ms. Then the
initials, over the remaining letters of the alphabet (Nuttin, target word was presented for a maximum of 550 ms. If the
1985, 1987). Participants evaluated how much they liked participant did not respond within this time span, the target
each letter of the alphabet on response scales ranging from word was replaced by a red exclamation mark. The length
1 (I dislike this letter very much) to 7 (I like this letter very of the response deadline was automatically adjusted accor-
much). To capture reliable name-letter effects, participants ding to the participant’s average accuracy and speed in the
had to rate the likeability of each letter twice (duplicate preceding critical block (cf. Degner, 2009; Draine &
administration; Rudolph et al., 2008; Stieger, Voracek, & Greenwald, 1998).

DOI: 10.1002/per
S. Krause et al.
An initial practice block was followed by five critical Adjective scale (AS)
blocks of 60 trials each (each target word was paired once Next, participants rated the 10 attribute or target adjectives
with the self-related and once with an other-related prime). from the IAT and the APT on a scale ranging from 1 (not
Critical trials were presented in a fixed order to optimize at all) to 6 (very much). The instructions read: ‘Indicate the
the reliability of individual differences (Banse, 2001). Self- extent to which the following adjectives apply to you.’ The
esteem priming scores were calculated by excluding the trials AS score was computed as the mean rating for the five adjec-
in which the response was given after the deadline (Degner tives describing the positive pole and the five reverse-scored
et al., 2007). For each participant, a positivity index for the adjectives describing the negative pole.
self-related prime was calculated by subtracting the average
error rate for pleasant targets from the average error rate for
Social situations
unpleasant targets that followed the same prime. Likewise,
a positivity index for other-related primes was computed to Self-introduction
control for a preference for positive versus negative target Participants were asked to stand in front of the group and
adjectives in the analysis. To yield a single ISE priming ef- briefly introduce themselves by providing information about
fect, the positivity index for the self-related prime was their first name, age, place of origin, field of study, semester,
subtracted from the positivity index for other-related primes interests, and leisure-time activities.
(cf. Degner et al., 2007).
Because of technical problems, one participant did not Ostracism game
complete the self-esteem EB-APT (the Inquisit programme For each participant, a natural experience of being excluded
crashed at the end of the fourth critical block). This subject’s from a social group was created by using the ball-tossing
data were discarded. game Cyberball (Williams et al., 2000). During the game,
participants were led to believe that they were playing an
Reaction time affective priming task (RT-APT) interactive ball-tossing game with the other group members,
Before completing the EB-APT, participants were but in fact, the other players were simulated by the computer.
administered a standard reaction-time-based priming task Before the game began, participants were explicitly asked to
(RT-APT) as an additional ISE measure. The RT-APT proce- visualize the game situation, themselves, and the other
dure was identical to the above-mentioned EB-APT, but no players to make the passing behaviour more realistic.
response-deadline technique was used (for a detailed descrip- Throughout the game, each participant saw his or her face
tion of the reaction time priming measure, see Krause et al., in the middle of the bottom field on the screen; the other
2011). Thus, after the target word was presented, partici- three players’ faces were depicted in the middle left, middle
pants’ reaction times to the unambiguously valenced top, and middle right fields on the screen. When the animated
adjectives were measured and analysed. The examination of ball was tossed to the participant, she or he had to click on
self-esteem RT-APT effects followed the improved scoring one of the other three pictures to pass the ball to the selected
algorithm presented in Krause et al. (2011); that is, reaction group member.
times were trimmed and error penalties were given. After The game was programmed such that each participant
completing the RT-APT, participants had a 5-min break experienced the same degree of ostracism. In order to make
before they started with the above-described EB-APT. the ostracism appear natural, we created a successive (rather
than abrupt) shift from social inclusion to social exclusion:
During the first 16 throws, participants received one fourth
Explicit self-esteem (ESE) measures of the passes (perfect social inclusion). Over the next 24
throws, participants received one eighth of the passes; that
Rosenberg self-esteem scale (RSES) is, each participant was allowed only three throws, whereas
Participants responded to the German adaptation (von each confederate made seven throws (first level of social
Collani & Herzberg, 2003) of the RSES. The RSES was exclusion). Over the next 33 throws, participants received
employed as a direct measure of people’s conscious feelings one eleventh of the passes; that is, each participant was given
of global self-worth. The RSES score was obtained by aver- only three throws, whereas each confederate made 10 throws
aging the 10 RSES items, each measured on a 4-point Likert (second level of social exclusion). During the last 32 throws,
scale ranging from 1 (strongly disagree) to 4 (strongly participants were never given the ball again by the other
agree). group members (perfect social exclusion).
To check whether or not participants had experienced
Multidimensional self-esteem scale (MSES) social exclusion during the game, we measured participants’
After responding to the RSES, participants filled out mood before and after the game with five questions that
the MSES. The MSES (Schütz & Sellin, 2006)—a asked participants to indicate how they felt at that moment
German-specific adaption (i.e. not a one-to-one translation) (Williams et al., 2000). The anchors for the five items, rang-
of Fleming and Courtney’s (1984) original Self-Rating ing from 1 (negative mood) to 6 (positive mood), were
Scale—was applied as an additional direct measure of accepted–rejected, bad–good, happy–sad, tense–relaxed,
people’s conscious feelings about their self-esteem. The total and pleased–enraged (αbefore = .77; αafter = .88). Participants’
score was obtained from 32 items, each measured on a actual behaviour during the game was recorded by a web
7-point Likert scale ranging from 1 (never) to 7 (always). cam that was fixed above the screen.

DOI: 10.1002/per
Ostracism interview We used this molar-level behavioural approach rather

After the Cyberball game ended, the experimenter entered than observing microlevel behaviours (e.g. smiling, gestur-
the room and interviewed the participant about the ostracism ing, eyebrow flashes, etc.) because macrolevel behavioural
experience. The interview contained six questions (e.g. ‘Do measures are better suited to capture the psychological mean-
you think you were ostracized?’) and was standardized; that ing behind specific acts (Sherman, Nave, & Funder, 2009). A
is, each interviewer had to ask the questions in the specified smile, for example, can be caused by different motives (e.g.
order (the complete list of questions is included in the amusement, sarcasm, irritation), and thus, observers are only
Appendix of the SOM). able to judge whether a smile is an indicator of self-confident
behaviour when they include the context of the specific act.
Furthermore, some people with high self-confidence might
Personal interview smile when meeting strangers, whereas others may speak
After the participant had answered the last question from the with a loud voice. Thus, a self-confident behaviour cannot
ostracism interview, the experimenter changed the focus of be reduced to specific microlevel behaviours. Accordingly,
the conversation toward positive and negative aspects of microlevel behaviours show less cross-situational consis-
the participant’s personal life (‘Now that we have talked tency than molar-level behaviours (Funder & Colvin, 1991).
about your ostracism experience, I want to ask you about To ensure that the behavioural assessments of partici-
personal aspects of your life’). This part of the interview pants’ self-confidence were reliable and valid, each observer
was comprised of eight questions (e.g. ‘If you look back on had previously completed an observer training along with the
your life, do successes or failures predominate?’) that had other two observers in the specific rating group. As men-
to be asked in a specified order. At the end of the conversa- tioned previously in the Participants section, the first four
tion, the experimenter confronted participants with five study participants served as a training group that provided
self-threatening situations. Participants were asked to video examples for this observer training. At the beginning
imagine the situation and then indicate how they would cope of the training, each of the three observers in a rating group
with it (the complete list of questions is included in the independently evaluated the self-confidence of one training-
Appendix of the SOM). group participant. Thereafter, the three group members
discussed their individual ratings (i.e. each of them explained
her/his observer rating). After exchanging their ideas, the
Ratings of self-confident behaviours three observers collectively evaluated (i.e. by discussing their
thoughts) the self-confident behaviours of two additional
In each of the four social situations, participants’ behaviour training-group participants. The aim of this procedure was
was recorded by a video camera that was positioned in front to ensure that the observers would arrive at similar evaluation
of them. After the end of the study, independent groups of criteria. Finally, each observer independently evaluated the
three observers evaluated the degree to which each partici- self-confidence of the last training-group participant. The
pant appeared to be self-confident in each situation. In observer training was successful because group members
contrast to previous research in which behaviour was pre- reported no disagreements in their evaluations of this last
dicted with ISE (see Table 1), we aimed to reduce observer participant.
bias in the self-confidence ratings by using more than one After the end of the training session, each observer was
observer for each situation. Moreover, to avoid dependency given a rating sheet on which all of the participants were pre-
in the observer ratings, we aimed to recruit different sented in different random orders. Essentially, the observers
observers for each social outcome.3 were asked to independently rate each participant’s beha-
An observer rating of self-confident behaviour was made viour on the extent to which it showed self-confidence. After
on a 6-point Likert scale ranging from 1 (not at all) to 6 (very finishing the behavioural assessments, the ratings of the three
much). In order to ensure that the observers had a common observers in a group were averaged into a self-confident
understanding of what is meant by ‘self-confident beha- behaviour score for each individual participant in each of
viour,’ each rating sheet contained written definitions of be- the four social situations.
havioural anchors of self-confident versus insecure
behaviour. The following anchors were used: insecure
appearance versus self-confident appearance, insecure/low Power analysis
voice versus clear/strong voice, excitement versus calmness, Two analyses were computed to determine the statistical
tense posture versus straight posture. Although an observer power of our analyses. According to a meta-analysis of the
was asked to consider the behavioural anchors for a rating, results of Table 1, the ISE measures predicted actual beha-
she or he was explicitly instructed to make a global judge- viours with an average effect size of r = .42. With an alpha
ment on the basis of her or his holistic impression of how level of .05 and a sample size of 127, the power to detect this
self-confidently a person acted (for a similar approach, see effect with a two-sided significance test was .999. A compre-
Spalding & Hardin, 1999). hensive meta-analysis of the predictive validity of the IAT
found an average criterion correlation of .274 (Greenwald,
3
In sum, 10 psychology students (seven women) participated in the study as Poehlman, Uhlmann, & Banaji, 2009). Similarly, in a com-
observers of self-confident behaviors. Eight of these students rated one situ-
ation, and two of these students rated two situations. Accordingly, each of prehensive meta-analysis, Cameron, Brown-Iannuzzi, and
the four situations was evaluated by a different team of observers. Payne (2012) found that priming tasks showed an average

DOI: 10.1002/per
S. Krause et al.
correlation with behavioural measures of .28. Under the as- correlated with two of three ESE measures (rRSES = .24,
sumption that the ISE measures in our study would be corre- p < .01; rMSES = .30, p < .01). Given the fact that ISE and
lated with the behavioural outcomes with an effect size of ESE measures usually do not significantly correlate at all,
r = .28, the power to detect this effect was .90 in our study. this finding provides first evidence for the ability of the
EB-APT to validly assess self-esteem.
Criterion measures
RESULTS
The results of the Cyberball game suggest that, on average,
participants experienced social exclusion: Compared with
Descriptive statistics
their pregame mood (M = 4.24, SD = 0.79), participants
Self-esteem measures reported a significantly less positive mood after completing
The descriptive statistics and intercorrelations for the self- the Cyberball game (M = 3.13, SD = 0.99), t(126) = 8.20,
esteem measures can be found in Table 2. As can be seen p < .01, d = 1.24. The difference score computed between
in column 4, all seven self-esteem measures exhibited a pre- and postgame mood was significantly correlated with
statistically significant positivity bias. People’s general ten- ESE (rRSES = .38, p < .01; rMSES = .51, p < .01; rAS = .35,
dency to have a positive self-view (Bosson et al., 2000) p < .01), indicating that participants who self-reported high
was thus reflected by each self-esteem measure. With the levels of pregame self-esteem reported a smaller decrease
exception of the RT-APT, the reliabilities of the self-esteem in mood after social exclusion than people who self-reported
measures were higher than .70 and thus satisfactory. This is low levels of pregame self-esteem. The mood difference
particularly remarkable for the NLT and the EB-APT score was not correlated with any of the ISE measures
because, with the exception of the IAT, implicit measures (rIAT = .07, p = .44; rNLT = .04, p = .65; rRT-APT = .03,
tend to be chronically unreliable (e.g. Bosson et al., 2000; p = .73; rEB-APT = .07, p = .41).
LeBel & Paunonen, 2011). The descriptive statistics and intercorrelations for the four
As expected, all three ESE measures were significantly self-confident behaviours are shown in Table 3. To estimate
correlated (mean r = .60). The different types of ISE mea- observer agreement for each behavioural measure, we calcu-
sures (i.e. IAT, NLT, APT), by contrast, were not signifi- lated intraclass correlation coefficients (ICCs [2,k]). The
cantly correlated (mean r = .08) and thus demonstrated their reliabilities of the observer ratings varied across situations:
independence from each other—a typical finding in ISE Whereas each observer showed a good level of agreement
research (e.g. Bosson et al., 2000; Krause et al., 2011; between the two interview situations (with ICCs > .70), the
Rudolph et al., 2008). The average r of .12 between the direct interobserver reliability was relatively low albeit satisfactory
and indirect self-esteem measures replicated the finding that (with ICCs of around .60) in the self-introduction and game
the two kinds of measures do not share a substantial amount situations. All four behavioural criterion measures were pos-
of variance (for meta-analyses on the APT, IAT, and NLT, itively related, indicating some consistency in individual
see Cameron et al., 2012; Greenwald et al., 2009; Krizan & differences in self-confident behaviour across situations
Suls, 2008). However, regarding the convergence between (α = .76). Therefore, we additionally created a composite
ISE and ESE measures, a closer look on the intercorrelations score by averaging across the four criterion measures (see
reveals that the self-esteem EB-APT was significantly last line and last column of Table 3).
Table 2. Descriptive statistics and intercorrelations for self-esteem measures
Descriptives Intercorrelations
Self-esteem
measures M SD d Rel MSES AS IAT NLT RT-APT EB-APT
Explicit self-esteem
RSES 3.12 0.53 1.18** .88 .70** .52** .02 .04 .09 .24**
MSES 4.55 0.86 0.64** .93 .59** .13 .14 .07 .30**
AS 4.58 0.50 2.02** .74 — .13 .15 .01 .14
Implicit self-esteem
IAT 0.62 0.31 2.02** .85 — .14 .01 .03
NLT 1.10 1.38 0.80** .78 — .06 .15
RT-APT 0.06 0.09 0.66** .64 — .50**
EB-APT 0.12 0.22 0.56** .73 —
Note: N = 127. RSES = Rosenberg Self-Esteem Scale; MSES = Multidimensional Self-Esteem Scale; AS = adjective scale; IAT = self-esteem Implicit Associa-
tion Test; NLT = name-letter task; RT-APT = reaction-time affective priming task; EB-APT = error-based affective priming task. M = mean; SD = standard devi-
ation; d = Cohen’s d represents the effect size of the positivity bias tested against the neutral value for each measure (2.5, 4.0, and 3.5 indicate neutral self-esteem
on the RSES, MSES, and AS; 0 indicates neutral self-esteem on the IAT, NLT, RT-APT, and the EB-APT); Rel = Cronbach’s alpha based on the items from the
RSES, MSES, and AS; test–retest reliability for the first initial from the NLT; Spearman–Brown corrected split-half reliability—estimated by separately applying
the scoring algorithm to two mutually exclusive subsets of critical task trials—for the IAT, the RT-APT, and the EB-APT.
*p < 05.
**p < .01.

DOI: 10.1002/per
Table 3. Descriptive statistics and intercorrelations for self-confident behaviours
Descriptives Intercorrelations
Self-confident
behaviours M SD ICC Ostracism game Ostracism interview Personal interview Composite score
Self-introduction 4.02 0.71 .59 .40** .53** .40** .73**

Ostracism game 4.13 0.92 .61 — .46** .28** .71**
Ostracism interview 4.14 0.91 .71 — .60** .85**
Personal interview 4.06 0.92 .81 — .76**
Composite score 4.08 0.66 — —
Note: N = 127. M = mean; SD = standard deviation; ICC = intraclass correlation (2,k).

**p < .01.
Predictive validities In order to examine the ability of the self-esteem

EB-APT to incrementally predict each of the behavioural
Table 4 shows the correlations between the self-esteem
outcomes above and beyond ESE, we computed hierarchi-
measures and the observer ratings of self-confident beha-
cal regressions with the observer ratings of the self-
viours. As can be seen in the first three columns, the ESE
introduction, ostracism game, ostracism interview, personal
measures (at least marginally) significantly predicted each
interview, and the composite score as separate criteria.6 In
of the criterion variables. There was only one nonsignificant
each of the five regressions, the RSES, the MSES, and
but positive correlation between self-confident behaviour
the AS were entered as separate predictors in Step 1, and
during the self-introduction and the RSES (r = .14, p = .11).
the self-esteem priming score was entered in Step 2. As
Remarkably, the MSES showed substantially higher predic-
can be seen in Table 5, the self-esteem EB-APT at least
tive validities than the established RSES.
marginally significantly predicted each of the five validity
For the ISE measures, a similar pattern of significant pre-
criteria above and beyond ESE, even when all three direct
dictive validities was obtained only for the error-based prim-
self-esteem measures are included in the analysis. The
ing procedure. That is, whereas the self-esteem EB-APT
highest amounts of variance that were additionally
significantly predicted each outcome variable,4 the self-
explained by the self-esteem EB-APT were found for the
esteem IAT and the NLT were not significantly correlated
personal interview situation and the composite score of
with any of the self-confident behaviours. Only two predic-
observed self-confident behaviour (see the last two rows of
tions using the NLT almost reached significance, one for
Table 5).
observer-rated self-confident behaviour during the ostracism
The last column of Table 5 shows the amount of variance
game (r = .17, p = .06) and one for the composite behaviour
explained by both types of self-esteem for each validity
score (r = .15, p = .097). Notably, although both kinds of
criterion. Thus, using the EB-APT in addition to standard
self-esteem priming measures were substantially correlated
self-reports of self-esteem provided a better explanation of
(r = .50, p < .01; see Table 2), the predictive validities were
self-confident behaviour in social settings.
substantially higher for the EB-APT than for the RT-APT,
In addition to the incremental validation strategy, we used
and were consistently significant only for the EB-APT (see
commonality analyses (Nimon, Lewis, Kane, & Haynes,
the last column of Table 4).
2008) to disentangle the amount of criterion variance attribut-
As can be seen in Table 4, a comparison of the sizes of
able to both kinds of self-esteem measures (common vari-
the correlation coefficients between the seven self-esteem
ance) from explained variance unique to ESE measures and
measures and the five behavioural criteria showed that the
explained variance unique to the self-esteem EB-APT. To
EB-APT demonstrated predictive validities that were
create a single index of ESE, we first standardized and then
descriptively stronger than for five of the other single mea-
averaged the scores on the RSES, the MSES, and the AS
sures (including two established ESE measures, the RSES
(α = .82). As can be seen in Figure ure 1, consistently across
and the AS). Only the MSES demonstrated stronger predic-
all four situations, about one quarter of the variance was
tive validities than the error-based priming measure, but the
shared between the composite of ESE and the priming mea-
differences were not significant (Z ≤ 1.74, p ≥ .08).5
sure (25.7% for the composite behaviour score), but both
kinds of measures also had unique variance. The amount of
unique variance for the self-esteem EB-APT was about one
4
A reviewer correctly pointed out that this phrase was incorrect for one of the
outcomes. Self-confident behavior in the self-introduction situation was
assessed before participants completed the self-esteem EB-APT. Thus, care- 6
We additionally computed multilevel mixed effects regression analyses re-
fully worded, the priming measure showed ‘retrodictive’ validity in this case. ported in Table 5 with the group as a random factor to account for possible
In the remainder of the manuscript, we use the term predictive validity of the effects of grouping. However, in all five mixed effects regression analyses,
self-esteem EB-APT for readability reasons and because the priming mea- the random group factor was not significant as shown by a likelihood ratio
sure indeed significantly predicted the remaining three behavioral outcomes. test comparing the mixed effects regression model with the standard linear
5
We found only two significant sex differences in the prediction of observer regression model (LR test statistics < .53, ps > .23). In addition, the ICCs
ratings of self-confident behaviors with the ESE and ISE measures (see Ta- were smaller than .06 for all five analyses, also indicating that group effects
ble S1 in the SOM). Regarding the predictive validities of the interactions were negligible for all five dependent variables. Accordingly, the fixed ef-
between ESE and ISE measures for the outcome variables, there was no con- fects in the mixed effects analyses did not differ substantially from the ef-
sistent pattern of significant effects (see Table S2 in the SOM). fects obtained from the ordinary least squares regressions.

DOI: 10.1002/per
S. Krause et al.
Table 4. Predictive validities of the explicit and implicit self-esteem measures
Explicit self-esteem Implicit self-esteem

Self-confident
behaviours RSES MSES AS IAT NLT RT-APT EB-APT
Self-introduction .14 .29** .25** .03 .13 .05 .23*

Ostracism game .17† .24** .17† .02 .17† .18* .24**
Ostracism interview .22* .46** .26** .14 .08 .09 .27**
Personal interview .36** .48** .31** .04 .08 .15† .36**
Composite score .30** .49** .33** .08 .15† .16† .36**
Note: N = 127. RSES = Rosenberg Self-Esteem Scale; MSES = Multidimensional Self-Esteem Scale; AS = adjective scale; IAT = Implicit Association Test;
NLT = name-letter task; RT-APT = reaction-time affective priming task; EB-APT = error-based affective priming task.
†
p < .10.
*p < .05.
**p < .01.
Table 5. Incremental validity of the self-esteem EB-APT above and beyond the ESE measures
Hierarchical regression
Step 1 Step 2
EB-APT
ESEM ESEM + EB-APT
Self-confident behaviours R2 β ΔR2 R2
Self-introduction .11** .16 .02† .13**

Ostracism game .07* .17 .03† .10*
Ostracism interview .25** .15 .02† .27**
Personal interview .23** .24 .05** .29**
Composite score .26** .23 .05** .31**
Note: N = 127. ESEM = Explicit self-esteem measures; EB-APT = error-based affective priming task.
†
p < .10.
*p < .05.
**p < .01.
Figure 1. Commonality analyses for the self-esteem EB-APT and the composite score of the three explicit self-esteem measures (ESEM). 100% refers to the
mean total amount of explained variance in a behavioural outcome. Unique EB-APT variance refers to variance uniquely explained by the error-based priming
measure. Unique ESEM variance refers to variance uniquely explained by the composite of the three ESE measures (i.e. RSES, MSES, and AS). Common EB-
APT + ESEM variance refers to explained variance that is common to the priming and composite of ESE scores.
quarter (23.4% for the composite behaviour score), about one Additional results
half of the variance in the composite behaviour score was
uniquely explained by ESE measures (50.9 %). It is important In the SOM, we provide a detailed overview of additional
to note that both analytic strategies (i.e. incremental validity variables and results that were obtained in our self-esteem
and commonality analyses) showed that the self-esteem EB- study. Some of the presented findings should be highlighted
APT substantially added to the prediction of self-confident because they might be important for future research on
behaviour above and beyond ESE measures. behaviour prediction and self-confidence. First, for the

DOI: 10.1002/per
ratings of self-confident behaviours reported above, the in self-esteem expect to be valued in new social situations
observers watched participants’ videotapes that presented (Back, Krause et al., 2009), but they also behave in a more
both the visual and auditory information from the respective self-confident manner than people with low self-esteem do.
recording. In addition to these full information video-based In line with the assumption of dual-process models that
judgements, we also obtained observer ratings of self- many social-behavioural outcomes are not driven by purely
confident behaviours by presenting videos with either (i) automatic or purely controlled processes but are rather the
only the visual information (i.e. no sound) or (ii) only the result of both kinds of processes (cf. Back et al., 2009; Fazio
auditory information (i.e. no picture). In short, we found a & Olson, 2003; Strack & Deutsch, 2004), we found that for
consistent pattern that the predictive validities of the ESE each criterion variable, both the ISE and ESE measures
measures and the self-esteem EB-APT were highest, when contributed to the prediction of self-confident behaviour.
observers had all of the information about self-confident Future research might extend our approach and investigate
behaviour, that is, both the auditory and visual cues (for a the effects of ISE and ESE on self-confident behaviour in
detailed description of the results, see the section ‘5. other social contexts (e.g. first dates) or with respect to
Additional Observer Ratings of Self-Confident Behaviour’ socially important outcomes (e.g. mating success, sales
in the SOM). success).
Second, in addition to above-described observer ratings
of participants’ self-confidence, we obtained further
The roles of ESE and ISE measures in predicting
behaviour-based ratings. Amongst others, in the two
self-confident behaviour
ostracism situations (i.e. ostracism game and ostracism
interview), observers rated the degree to which each partici- The three ESE measures of our study (i.e. the RSES, the
pant experienced ostracism. In particular in the ostracism in- MSES, and the AS) were highly significantly correlated,
terview situation, we found that the two well-established and each of them demonstrated validity in predicting
ESE measures (i.e. the RSES and the MSES) and the self- observed self-confident behaviour across different social
esteem EB-APT were significantly correlated with observer situations. The positive results for the criterion validity of
ratings of perceived ostracism. This finding offers further ev- the ESE measures provide evidence that not only are self-
idence for the predictive validities of the RSES, MSES and reports of self-esteem useful for predicting other self-report
the self-esteem EB-APT with respect to actual behaviour. It outcomes (for criticisms of this practice, see, e.g. Baumeister
is important to note that the priming measure again demon- et al., 2007; Furr, 2009), but they are also useful for
strated its value in predicting observer ratings of actual be- predicting actual human behaviour. Furthermore, the results
haviour (for a detailed description of the results, see the of the convergent and criterion validities confirmed the
section ‘6. Behaviour Ratings besides Self-Confidence’ in assumption that different self-report measures of self-esteem
the SOM). validly tap the same construct: ESE (e.g. Bosson et al., 2000;
Falk et al., 2015).
The convergent and criterion validity results were less
DISCUSSION consistent for the three types of ISE measures. These
measures were not correlated, indicating that the IAT, the
Predicting individual differences in people’s actual beha- NLT, and the priming measures (i.e. the EB-APT and the
viour is an important goal of personality psychology. In RT-APT) capture different constructs or at least different
pursuing this aim, dual-process models of social information aspects of ISE as suggested by Back et al. (2009). It is inter-
processing propose that newly developed indirect measures esting that the self-esteem EB-APT correlates more highly
be used because they capture automatically activated evalua- with two established ESE measures (RSES and MSES) than
tions that people are unable or unwilling to report in direct the other ISE measures do. Given the fact that ISE and ESE
measures (e.g. questionnaires) and are especially relevant measures usually do not significantly correlate at all (e.g.
for behavioural outcomes. On the basis of this assumption, Bosson et al., 2000; Krause et al., 2011), this finding
we examined and compared the validities of ISE and ESE provides further evidence for the EB-APT’s ability to validly
measures in predicting self-confident behaviour in the pres- assess self-esteem. Moreover, the lower ISE-ESE correla-
ence of unacquainted others. tions of the IAT, NLT, and RT-APT might point to the fact
that these three ISE measures are less valid than the EB-APT.
This conclusion is supported by the fact that only the self-
Self-esteem and the prediction of self-confident behaviour
esteem EB-APT was consistently significantly correlated
In four different social situations, we found significant with all four self-confident behaviours and also incrementally
correlations between self-evaluations and observer-rated predicted each of the four outcomes beyond self-reported
self-confidence. Thus, compared with people with low self- self-esteem. Notably, the correlation and commonality analy-
esteem, the behaviour of people with high self-esteem was ses showed that the self-esteem EB-APT was, compared with
more self-confident during (i) a self-introduction to a group; the established ESE measures, the second strongest single
(ii) an experience of being ostracized by unacquainted group predictor of the behavioural outcomes. Consequently, in
members; (iii) an interview about the ostracism experience; our study, the self-esteem EB-APT was the only indirect
and (iv) an interview about their personal life. Taken measure that reliably captured a component of automatically
together, according to our findings, not only do people high activated self-evaluation that was consequential for

DOI: 10.1002/per
S. Krause et al.
subsequent self-esteem-related behaviours and was thus a immediately) activated. This activation then facilitates the re-
valid indicator of ISE. With regard to previous results, our sponse to affectively congruent target words. That is, when
study replicated findings that reflected the self-esteem APT’s ISE is high, the presence of a self-relevant prime (e.g. a face
ability to predict actual behaviours above and beyond ESE picture) leads immediately to a positive evaluation and thus
measures (Spalding & Hardin, 1999; Vandromme et al., to fast or correct reactions to unequivocally positive target
2011). adjectives. For people low in ISE, by contrast, the same
The low and nonsignificant criterion validity coefficients prime leads to an automatically activated negative evaluation
for the self-esteem IAT, the NLT, and the self-esteem and thus to slow or incorrect reactions to the same target
RT-APT were not expected. This is particularly surprising adjectives.
for the IAT because previous studies had found that the Second, because ISE measures were designed to assess
self-esteem IAT could predict behavioural outcomes (see automatically activated evaluations of self-relevant stimuli
Table 1). For the NLT, the obtained results were (somewhat) compared with other-relevant stimuli, the material has to be
less surprising because there is currently only one study that unambiguous. Here, presenting face pictures seems to be
reported a direct effect of name-letter liking on defensive the most effective way because they are clearly identifiable
verbal behaviour during an interview (Kernis et al., 2008). as self- versus other-relevant stimuli. Pronouns or name let-
For the RT-APT, the relatively low reliability might explain ters, as presented in earlier self-esteem APTs (e.g. Bosson
the lower criterion validity coefficients compared to the et al., 2000; Wentura et al., 2005), do not have this quality
EB-APT. of clarity because they do not guarantee that a self- or
In sum, in our comparison of the predictive and incre- other-relevant evaluation will be automatically activated.
mental validities of established ISE measures, we found that Furthermore, using digitalized photos as primes is in line
only one measure, the self-esteem EB-APT, was significantly with the recommendations of Fazio, Jackson, Dunton, and
related to self-confident behaviour. It is interesting that these Williams (1995), who stated ‘…our laboratory has typically
results are in line with recently published articles that have observed stronger facilitation effects than in our earlier work
criticized the lack of validity of the self-esteem IAT, the using words as primes’ (p. 1014). Recently published studies
NLT, and the self-esteem RT-APT (Buhrmester et al., that have used face-picture affective priming as measures of
2011; Falk et al., 2015). Regarding the predictive power of ISE and implicit partner evaluation have provided evidence
the self-esteem EB-APT, it is noteworthy that its ability to for the usefulness of this task feature (Krause et al., 2011,
predict self-confident behaviour was consistently found in 2012; McNulty, Olson, Meltzer, & Shaffer, 2013; McNulty,
not just one situation but in four different social contexts Baker, & Olson, 2014).
(and was thus replicated within one study). Nevertheless, Third, we replicated our laboratory’s previous results,
although these validity results are promising for the self- which showed that an error-based priming procedure is a
esteem priming measure, they are based on only one sample. valuable tool for assessing ISE with the affective priming
Accordingly, future research should aim to replicate our task (Krause et al., 2012). In the present study, we used the
findings, also using multiple measures of ISE and multiple more user-friendly adaptive response-deadline technique
behavioural outcomes. Such replications are needed to pro- (Degner et al., 2007), which simplifies the established
vide robust evidence for one optimistic interpretation of the response-window technique (Draine & Greenwald, 1998).
present results— that the self-esteem EB-APT indeed has Again, we were able to show that an error-based self-esteem
the potential to overcome the weak and inconsistent APT can be used to (i) assess the expected positivity bias that
validities found for ISE measures (Buhrmester et al., 2011). reflects people’s general tendency to have a positive implicit
self-evaluation (Bosson et al., 2000); (ii) measure ISE with a
reliability coefficient that is comparable to values typically
The self-esteem EB-APT: a valuable addition to the
reported for the established IAT method (Krause et al.,
self-esteem-assessment toolbox
2011); and (iii) incrementally predict multiple self-esteem-
Previous research has already shown that improved self- relevant outcomes beyond self-reports of self-esteem (Krause
esteem APTs are comparatively reliable (Krause et al., et al., 2012). In contrast, the standard reaction-time-based
2011) and able to incrementally predict self-esteem-relevant priming procedure of our study (i.e. the RT-APT) showed
outcomes (Back, Krause et al., 2009; Krause et al., 2012). lower reliabilities and validities in comparison with the
In the present research, in particular the self-esteem EB- response-deadline priming measure. This is in line with pre-
APT shared the self-esteem IAT’s and the NLT’s satisfactory vious results of our lab (Krause et al., 2011, 2012). We
reliabilities, but it was the only ISE measure with predictive therefore recommend using the self-esteem EB-APT when
validity. Although this finding is certainly in need of a thor- it comes to investigating the effects of ISE with the affective
ough replication, we want to highlight three features of the priming task.
self-esteem EB-APT that we regard as important for why Based on the combination of the theoretical foundation of
our variant of the priming technique might add to the under- affective priming effects, the clarity of self- and other-
standing of self-esteem and its behavioural consequences. relevant stimuli, and the error-generating technique, our
First, there is a well-established theoretical explanation results suggest that the self-esteem EB-APT is well-suited
for how affective priming effects arise (cf. Fazio, 2001): for the reliable and valid assessment of ISE. Future research
The presentation of an attitude object (e.g. a self-relevant on the consequences of ISE might consider additionally
stimulus) leads to an evaluation that is automatically (i.e. making use of the self-esteem EB-APT, particularly when it

DOI: 10.1002/per
comes to predicting actual behaviours (the corresponding ISE measures as well as multiple actual behaviours as
Inquisit script and manual can be downloaded from https:// criterion variables are sorely needed.
osf.io/fqc4a/). Taken together, the current study’s predictive and
incremental validity results provide evidence that the self-
esteem EB-APT is a valuable complement to self-reports of
Limitations self-esteem, in particular when self-confident behaviours in
the presence of unacquainted others are under investigation.
Some limitations of the present study should be noted: First,
we obtained the self-reports of ESE a few days before we
assessed ISE and self-confidence. The time interval between
the online measurement of ESE and the lab sessions was on DECLARATION OF CONFLICTS OF INTEREST
average 5.12 days (SD = 4.82 days). Therefore, it is possible
that situational influences (e.g. daily mood) could have led The authors declare no potential conflicts of interest with
to lower predictive validities of the ESE measures compared respect to the research, authorship, or publication of this
with a design in which ESE was measured at the same time article.
as the ISE and criterion measures.
Second, because of the fixed order of the tasks and
situations, we could not rule out the possibility that the ISE FUNDING
measures’ positivity effects or predictive validities were
influenced by this specific order. We did not counterbalance The authors disclose receipt of the following financial sup-
the sequence of the tasks in order to maximize the reliability port for the research, authorship, or publication of this article:
of individual differences in the measurement of implicit and Preparation of this manuscript was supported by Grant
explicit self-esteem. At least our results did not suggest sub- SCHM 1646/2-3 from the German Research Foundation.
stantial order effects because the effect sizes and reliabilities
of the popular ISE measures that were presented (IAT, NLT,
RT-APT, and EB-APT) showed values in the range that is REFERENCES
typically reported for these measures (Bosson et al., 2000;
Rudolph et al., 2008; Krause et al., 2011). Asendorpf, J. B., Banse, R., & Mücke, D. (2002). Double dissocia-
Third, although we aggregated behaviours across tion between implicit and explicit personality self-concept: The
case of shy behavior. Journal of Personality and Social Psychol-
observers and situations, we employed only one global ogy, 83, 380–393.
behaviourally anchored self-confidence item. Although this Baccus, J. R., Baldwin, M. W., & Packer, D. J. (2004). Increasing
molar-level assessment seems to be an appropriate approach implicit self-esteem through classical conditioning. Psychologi-
for capturing behavioural self-confidence on a psychological Science, 15, 498–502.
cally meaningful level (see Funder & Colvin, 1991), future Back, M. D., Krause, S., Hirschmüller, S., Stopfer, J., Egloff, B., &
Schmukle, S. C. (2009). Unraveling the three faces of self-
research is well-advised to additionally assess more specific esteem: A new information-processing sociometer perspective.
microlevel behaviours (e.g. specific gestures, postures, and Journal of Research in Personality, 43, 933–937.
linguistic indicators). In doing so, in addition to testing the Back, M. D., & Nestler, S. (in press). Dual process approaches to
additive prediction pattern revealed here (see Perugini, personality. In R. Deutsch, B. Gawronski, & W. Hofmann
Richetin, & Zogmaister, 2010), researchers might also try (Eds.), Reflective and impulsive determinants of human behavior.
New York: Psychology Press.
to test whether ESE measures are more predictive of Back, M. D., Penke, L., Schmukle, S. C., Sachse, K., Borkenau,
controlled behaviours and ISE measures are more predictive P., & Asendorpf, J. B. (2011). Why mate choices are not as
of impulsive behaviours (i.e. double-dissociation pattern; e.g. reciprocal as we assume: The role of personality, flirting and
Asendorpf, Banse, & Mücke, 2002; Egloff & Schmukle, physical attractiveness. European Journal of Personality, 25,
2002). 120–132.
Back, M. D., Schmukle, S. C., & Egloff, B. (2009). Predicting
actual behavior from the explicit and implicit self-concept of
personality. Journal of Personality and Social Psychology, 97,
Conclusions 533–548.
Back, M. D., Schmukle, S. C., & Egloff, B. (2011). A closer look at
Behaving in a self-confident manner in the presence of first sight: Social relations lens model analysis of personality and
unacquainted others is a fundamental and consequential interpersonal attraction at zero acquaintance. European Journal
psychological phenomenon. In the present study, we were of Personality, 25, 225–238.
able to show that ISE and ESE play crucial roles in Banse, R. (2001). Affective priming with liked and disliked
persons: Prime visibility determines congruency and
explaining individual differences in self-confident behaviour incongruency effects. Cognition and Emotion, 15, 501–520.
in different social situations. It is important to note that in our Baumeister, R. F., Vohs, K. D., & Funder, D. C. (2007).
comparison of the ability of the three types of popular indi- Psychology as the science of self-reports and finger movements.
rect self-esteem measures to predict self-confident behaviour, Whatever happened to actual behavior? Perspectives on
only the self-esteem EB-APT was found to be a valid Psychological Science, 2, 396–403.
Bluemke, M., & Friese, M. (2012). On the validity of idio-
measure of ISE. Of course, these results do not call the pre- graphic and generic self-concept Implicit Association Tests:
dictive validity of the self-esteem IAT and the NLT into A core-concept model. European Journal of Personality, 26,
question in general. More studies with multiple established 515–528.

DOI: 10.1002/per
S. Krause et al.
Bosson, J. K., Swann, W. B., & Pennebaker, J. W. (2000). Stalking Furr, R. M. (2009). Personality psychology as a truly behavioural
the perfect measure of implicit self-esteem: The blind men and science. European Journal of Personality, 23, 369–401.
the elephant revisited. Journal of Personality and Social Psychol- Gawronski, B., & Bodenhausen, G. V. (2006). Associative and
ogy, 79, 631–643. propositional processes in evaluation: An integrative review of
Brown, J. D., & Kobayashi, C. (2002). Self-enhancement in Japan implicit and explicit attitude change. Psychological Bulletin,
and America. Asian Journal of Social Psychology, 5, 145–168. 132, 692–731.
Buhrmester, M. D., Blanton, H., & Swann, W. B. Jr. (2011). Greenwald, A. G., & Banaji, M. R. (1995). Implicit social
Implicit self-esteem: Nature, measurement, and a new way cognition: Attitudes, self-esteem, and stereotypes. Psychological
forward. Journal of Personality and Social Psychology, 100, Review, 102, 4–27.
365–385. Greenwald, A. G., Banaji, M. R., Rudman, L. A., Farnham, S. D.,
Cameron, C. D., Brown-Iannuzzi, J. L., & Payne, B. K. (2012). Nosek, B. A., & Mellott, D. S. (2002). A unified theory of
Sequential priming measures of implicit social cognition: A implicit attitudes, stereotypes, self-esteem, and self-concept.
meta-analysis of associations with behavior and explicit attitudes. Psychological Review, 109, 3–25.
Personality and Social Psychology Review, 16, 330–350. Greenwald, A. G., & Farnham, S. D. (2000). Using the Implicit
Degner, J. (2009). On the (un-)controllability of affective priming: Association Test to measure self-esteem and self-concept.
Strategic manipulation is feasible but can possibly be prevented. Journal of Personality and Social Psychology, 79,
Cognition & Emotion, 23, 327–354. 1022–1038.
Degner, J., Wentura, D., Gniewosz, B., & Noack, P. (2007). Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Under-
Hostility-related prejudice against Turks in adolescents: standing and using the Implicit Association Test: I. An improved
Masked affective priming allows for a differentiation of auto- scoring algorithm. Journal of Personality and Social Psychology,
matic prejudice. Basic and Applied Social Psychology, 29, 85, 197–216.
245–256. Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji,
DeHart, T., Pelham, B. W., & Tennen, H. (2006). What lies M. R. (2009). Understanding and using the Implicit Association
beneath: Parenting style and implicit self-esteem. Journal of Test: III. Meta-analysis of predictive validity. Journal of Person-
Experimental Social Psychology, 42, 1–17. ality and Social Psychology, 97, 17–41.
DeLongis, A., Folkman, S., & Lazarus, R. S. (1988). The impact of Haeffel, G. J., Abramson, L. Y., Brazy, P. C., Shah, J. Y.,
daily stress on health and mood: Psychological and social re- Teachman, B. A., & Nosek, B. A. (2007). Explicit and implicit
sources as mediators. Journal of Personality and Social Psychol- cognition: A preliminary test of a dual-process theory of cogni-
ogy, 54, 486–496. tive vulnerability to depression. Behaviour Research and Ther-
Denissen, J. J. A., Penke, L., Schmitt, D. P., & Van Aken, M. A. G. apy, 45, 1155–1167.
(2008). Self-esteem reactions to social interactions: Evidence for Hager, W., & Hasselhorn, M. (1994). Handbuch deutschsprachiger
sociometer mechanisms across days, people, and nations. Journal Wortnormen [Handbook of German word norms]. Göttingen,
of Personality and Social Psychology, 95, 181–196. Germany: Hofgrefe.
Draine, S. C., & Greenwald, A. G. (1998). Replicable unconscious Hetts, J. J., Sakuma, M., & Pelham, B. W. (1999). Two roads to
semantic priming. Journal of Experimental Psychology. General, positive regard: Implicit and explicit self-evaluation and culture.
127, 286–303. Journal of Experimental Social Psychology, 35, 512–559.
Egloff, B., & Schmukle, S. C. (2002). Predictive validity of an Holland, R. W., Wennekers, A. M., Bijlstra, G., Jongenelen, M. M.,
Implicit Association Test for assessing anxiety. Journal of & van Knippenberg, A. (2009). Self-symbols as implicit motiva-
Personality and Social Psychology, 83, 1441–1455. tors. Social Cognition, 27, 579–600.
Erol, R. Y., & Orth, U. (2013). Actor and partner effects of self- Hutteman, R., Nestler, S., Wagner, J., Egloff, B., & Back, M. D.
esteem on relationship satisfaction and the mediating role of se- (2015). Wherever I may roam: Processes of self-esteem develop-
cure attachment between the partners. Journal of Research in ment from adolescence to emerging adulthood in the context of
Personality, 47, 26–35. international student exchange. Journal of Personality and Social
Falk, C. F., Heine, S. J., Takemura, K., Zhang, C. X. J., & Hsu, Psychology, 108, 767–783.
C.-W. (2015). Are implicit self-esteem measures valid for Imada, A. S., & Hakel, M. D. (1977). Influence of nonverbal
assessing individual and cultural differences? Journal of communication and rater proximity on impressions and decisions
Personality, 83, 56–68. in simulated employment interviews. Journal of Applied Psychol-
Fazio, R. H. (1990). Multiple processes by which attitudes guide ogy, 62, 295–300.
behavior: The MODE model as an integrative framework. In M. Inquisit 2.0.60616 [Computer Software]. (2006). Seattle, WA:
P. Zanna (Ed.), Advances in experimental social psychology Millisecond Software LLC.
(pp. 75–109), 23. San Diego, CA: Academic Press. Jones, W. H., & Briggs, S. R. (1984). The self-other discrepancy in
Fazio, R. H. (2001). On the automatic activation of associated eval- social shyness. In R. Schwarzer (Ed.), The self in anxiety, stress,
uations: An overview. Cognition and Emotion, 15, 115–141. and depression (pp. 93–107). Amsterdam: North Holland.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. Kernis, M. H., Lakey, C. E., & Heppner, W. L. (2008). Secure
(1995). Variability in automatic activation as an unobtrusive mea- versus fragile high self-esteem as a predictor of verbal defensive-
sure of racial attitudes: A bona fide pipeline? Journal of Person- ness: Converging findings across three different markers. Journal
ality and Social Psychology, 69, 1013–1027. of Personality, 76, 477–512.
Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social Kitayama, S., & Karasawa, M. (1997). Implicit self-esteem in
cognition research: Their meaning and use. Annual Review of Japan: Name letters and birthday numbers. Personality and
Psychology, 54, 297–327. Social Psychology Bulletin, 23, 736–742.
Fleming, J. S., & Courtney, B. E. (1984). The dimensionality of Krause, S., Back, M. D., Egloff, B., & Schmukle, S. C. (2011).
self-esteem II: Hierarchical facet model for revised measurement Reliability of implicit self-esteem measures revisited. European
scales. Journal of Personality and Social Psychology, 46, Journal of Personality, 25, 239–251.
404–421. Krause, S., Back, M. D., Egloff, B., & Schmukle, S. C. (2012). A
Franck, E., De Raedt, R., & De Houwer, J. (2007). Implicit but not new reliable and valid tool for measuring implicit self-esteem:
explicit self-esteem predicts future depressive symptomatology. The response-window affective priming task. European Journal
Behaviour Research and Therapy, 45, 2448–2455. of Psychological Assessment, 28, 87–94.
Funder, D. C., & Colvin, C. R. (1991). Explorations in behavioral Krizan, Z., & Suls, J. (2008). Are implicit and explicit measures of
consistency: Properties of persons, situations, and behaviors. self-esteem related? A meta-analysis for the name-letter test.
Journal of Personality and Social Psychology, 60, 773–794. Personality and Individual Differences, 44, 521–531.

DOI: 10.1002/per
Lammers, J., Dubois, D., Rucker, D. D., & Galinski, A. D. (2013). observed or spontaneous behavior, but not self-reported or
Power gets the job: Priming power improves interview outcomes. controlled behavior. Journal of Psychology, 218, 12–19.
Journal of Experimental Social Psychology, 49, 776–779. Rudolph, A., Schröder-Abé, M., Schütz, A., Gregg, A. P., &
Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Sedikides, C. (2008). Through a glass, less darkly? Reassessing
Self-esteem as an interpersonal monitor: The sociometer convergent and discriminant validity in measures of implicit
hypothesis. Journal of Personality and Social Psychology, 68, self-esteem. European Journal of Psychological Assessment,
518–530. 24, 273–281.
LeBel, E. P., & Campbell, L. (2009). Implicit partner affect, rela- Schimmack, U., & Diener, E. (2003). Predictive validity of explicit
tionship satisfaction, and the prediction of romantic breakup. and implicit self-esteem for subjective well-being. Journal of
Journal of Experimental Social Psychology, 45, 1291–1294. Research in Personality, 37, 100–106.
LeBel, E. P., & Gawronski, B. (2009). How to find what’s in a Schütz, A., & Sellin, I. (2006). Die Multidimensionale
name: Scrutinizing the optimality of five scoring algorithms for Selbstwertskala (MSWS) [The multidimensional self-esteem
the name-letter task. European Journal of Personality, 23, scale]. Göttingen, Germany: Hofgrefe.
85–106. Sherman, R. A., Nave, C. S., & Funder, D. C. (2009). The apparent
LeBel, E. P., & Paunonen, S. V. (2011). Sexy but often unreliable: objectivity of behavior is illusory. European Journal of
Impact of unreliability on the replicability of experimental Personality, 23, 430–433.
findings involving implicit measures. Personality and Social Spalding, L. R., & Hardin, C. D. (1999). Unconscious unease and
Psychology Bulletin, 37, 570–583. self-handicapping: Behavioral consequences of individual
MacDonald, G., & Leary, M. R. (2012). Individual differences in differences in implicit and ESE. Psychological Science, 10,
self-esteem. In M. R. Leary, & J. P. Tangney (Eds.), 535–539.
Handbook of self and identity (pp. 354–377). New York: Srivastava, S., & Beer, J. S. (2005). How self-evaluations relate to
Guildford Press. being liked by others: Integrating sociometer and attachment
McNulty, J. K., Baker, L. R., & Olson, M. A. (2014). Implicit self- perspectives. Journal of Personality and Social Psychology, 89,
evaluations predict changes in implicit partner evaluations. 966–977.
Psychological Science, 25, 1649–1657. Stieger, S., & Burger, C. (2010). Implicit and ESE in the context of
McNulty, J. K., Olson, M. A., Meltzer, A. L., & Shaffer, M. J. Internet addiction. CyberPsychology, Behavior, and Social
(2013). Though they may be unaware, newlyweds implicitly Networking, 13, 681–688.
know whether their marriage will be satisfying. Science, 342, Stieger, S., Voracek, M., & Formann, A. K. (2012). How to admin-
1119–1120. ister the initial preference task. European Journal of Personality,
Miner, J. B. (1962). Personality and ability factors in sales perfor- 26, 63–78.
mance. Journal of Applied Psychology, 46, 6–13. Strack, F., & Deutsch, R. (2004). Reflective and impulsive determi-
Murrell, S. A., Meeks, S., & Walker, J. (1991). Protective functions nants of social behavior. Personality and Social Psychology
of health and self-esteem against depression in older adults facing Review, 8, 220–247.
illness or bereavement. Psychology and Aging, 6, 352–360. Vandromme, H., Hermans, D., & Spruyt, A. (2011). Indirectly
Nimon, K., Lewis, M., Kane, R., & Haynes, R. M. (2008). An R measured self-esteem predicts gaze avoidance. Self and Identity,
package to compute commonality coefficients in the multiple 10, 32–43.
regression case: An introduction to the package and a practical Verplanken, B., Friborg, O., Wang, C. E., Trafimow, D., & Woolf,
example. Behavior Research Methods, 40, 457–466. K. (2007). Mental habits: Metacognitive reflection on negative
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2005). Under- self-thinking. Journal of Personality and Social Psychology, 92,
standing and using the Implicit Association Test: II. Method 526–541.
variables and construct validity. Personality and Social Psychol- von Collani, G., & Herzberg, P. Y. (2003). Eine revidierte Fassung
ogy Bulletin, 31, 166–180. der deutschsprachigen Skala zum Selbstwertgefühl von
Nuttin, J. M. Jr. (1985). Narcissism beyond Gestalt and awareness: Rosenberg. Zeitschrift für Differentielle und Diagnostische
The name letter effect. European Journal of Social Psychology, Psychologie, 24, 3–7.
15, 353–361. Voss, A., Leonhart, R., & Stahl, C. (2007). How to make your own
Nuttin, J. M. Jr. (1987). Affective consequences of mere ownership: response boxes: A step-by-step guide for the construction of reli-
The name letter effect in twelve European languages. European able and inexpensive parallel-port response pads from computer
Journal of Social Psychology, 17, 381–402. mice. Behavior Research Methods, 39, 797–801.
Paulhus, D. L. (1984). Two component-models of socially desirable Walker, S. S., & Schimmack, U. (2008). Validity of a happiness
responding. Journal of Personality and Social Psychology, 46, Implicit Association Test as a measure of subjective well-being.
598–609. Journal of Research in Personality, 42, 490–497.
Paulhus, D. L., & Morgan, K. L. (1997). Perceptions of intelligence Wentura, D., Kulfanek, M., & Greve, W. (2005). Masked affective
in leaderless groups: The dynamic effects of shyness and priming by name letters: Evidence for a correspondence of
acquaintance. Journal of Personality and Social Psychology, explicit and implicit self-esteem. Journal of Experimental Social
72, 581–591. Psychology, 41, 654–663.
Perugini, M., Richetin, J., & Zogmaister, C. (2010). Prediction of Williams, K. D., Cheung, C. K. T., & Choi, W. (2000).
behavior. In B. Gawronski, & B. K. Payne (Eds.), Handbook of Cyberostracism: Effects of being ignored over the internet.
implicit social cognition: Measurement, theory, and applications Journal of Personality and Social Psychology, 79, 748–762.
(pp. 255–277). New York: Guilford. Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of
Robinson, M. D., & Meier, B. P. (2005). Rotten to the core: dual attitudes. Psychological Review, 107, 101–126.
Neuroticism and implicit evaluations of the self. Self and Identity, Young, D. M., & Beier, E. G. (1977). Role of applicant nonverbal
4, 361–372. communication in employment interview. Journal of
Rosenberg, M. (1965). Society and the adolescent self-image. Employment Counseling, 14, 154–165.
Princeton: University Press. Zeigler-Hill, V. (2010). The interpersonal nature of self-esteem: Do
Rudolph, A., Schröder-Abé, M., Riketta, M., & Schütz, A. (2010). different measures of self-esteem possess similar interpersonal
Easier when done than said! Implicit self-esteem predicts content? Journal of Research in Personality, 44, 22–30.

DOI: 10.1002/per

Per 2076

Uploaded by

Copyright:

Available Formats

Per 2076

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Per 2076

Uploaded by

Copyright:

Available Formats

European Journal of Personality, Eur. J. Pers.

Predicting Self-Conﬁdent Behaviour with Implicit and Explicit Self-Esteem

SASCHA KRAUSE1*, MITJA D. BACK2, BORIS EGLOFF3 and STEFAN C. SCHMUKLE1

Received 12 April 2016

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Table 1. Relations between actual behaviours and self-esteem measures

Reference Actual behaviours N Nobserver IAT NLT APT ESE

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

In this section, we provide an overview of the procedures and 1

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Ostracism interview We used this molar-level behavioural approach rather

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Table 2. Descriptive statistics and intercorrelations for self-esteem measures

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Table 3. Descriptive statistics and intercorrelations for self-conﬁdent behaviours

Self-introduction 4.02 0.71 .59 .40** .53** .40** .73**

Note: N = 127. M = mean; SD = standard deviation; ICC = intraclass correlation (2,k).

Predictive validities In order to examine the ability of the self-esteem

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Table 4. Predictive validities of the explicit and implicit self-esteem measures

Explicit self-esteem Implicit self-esteem

Self-introduction .14 .29** .25** .03 .13 .05 .23*

Self-introduction .11** .16 .02† .13**

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

Copyright © 2016 European Association of Personality Psychology Eur. J. Pers. (2016)

You might also like

Self-introduction 4.02 0.71 .59 .40 .53 .40 .73

Self-introduction .14 .29 .25 .03 .13 .05 .23*

Self-introduction .11 .16 .02† .13