Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Journal of Personality and Social Psychology © 2009 American Psychological Association

2009, Vol. 97, No. 3, 533–548 0022-3514/09/$12.00 DOI: 10.1037/a0016229

Predicting Actual Behavior From the Explicit and Implicit


Self-Concept of Personality

Mitja D. Back Stefan C. Schmukle


Johannes Gutenberg University, Mainz Westfälische Wilhelms University, Münster

Boris Egloff
Johannes Gutenberg University, Mainz

The authors present a behavioral process model of personality that specifies explicit and implicit aspects
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of the self-concept of personality as predictors of actual behavior. An extensive behavioral study (N ⫽


This document is copyrighted by the American Psychological Association or one of its allied publishers.

130) including a variety of relevant social situations was conducted. This approach allowed reliable
measurement of more than 50 behavioral indicators. A priori assignment of indicators to the Big Five
dimensions was conducted on the basis of theory and expert ratings. In line with the authors’ model, 3
main findings were revealed: First, direct measures (questionnaires) of personality predicted actual
behavior for all Big Five dimensions. Second, indirect measures (implicit association tests) of neuroti-
cism and extraversion also predicted actual behavior. Third, the predictive validity of these indirect
measures was incremental. The authors were additionally able to show that controlling for valence did
not affect any of these results. Implications and future prospects for the study of personality and actual
behavior are discussed.

Keywords: actual behavior, personality, impulsive and reflective processes, implicit and explicit mea-
sures, indirect and direct measurement

“Sayin’ it and doin’ it are two different things.” Activation of Actual Behavior:
—Bobby Byrd, The Legendary Henry Stone Presents Bobby Byrd The Reflective–Impulsive Model
Back From the Dead
In understanding how personality may influence actual behav-
Predicting behavior is an ultimate goal of psychology. Person-
ior, it is necessary to consider the processes that determine social
ality psychology contributes to this mission by describing and
behavior in general. Over the course of the past few years, an
explaining interindividual differences in people’s behavior. In
impressive amount of research evidence has shown that individu-
pursuing this aim, however, research has almost exclusively relied
on introspective self-reports of behavior (Baumeister, Vohs, & als process information about themselves and their environment in
Funder, 2007; Funder, 2001). The influence of personality on not only an explicit (i.e., controlled or conscious) but also an
directly observable behavior— on what people actually do— has implicit (i.e., automatic or unconscious) mode (Bargh & Char-
only rarely been subjected to investigation (see e.g., Borkenau, trand, 1999; Dijksterhuis & Nordgren, 2006; Epstein, 1994; Fazio,
Mauer, Riemann, Spinath, & Angleitner, 2004; Eaton & Funder, 1990; Greenwald & Banaji, 1995; Greenwald et al., 2002; Wilson,
2003; Mehl, Gosling, & Pennebaker, 2006, for recent exceptions). Lindsley, & Schooler, 2000).
In the present study, we aimed to systematically examine the Accordingly, a recently developed prominent dual process mod-
determinants of actually observable behavior. In doing so, we el—the reflective–impulsive model (RIM; Strack & Deutsch,
applied an extensive and theory-based behavioral approach. 2004)— explains social behavior as a joint function of two distinct
albeit related kinds of processes. First, social behavior may be
understood as reasoned action: Aspects of a situation are con-
Mitja D. Back and Boris Egloff, Department of Psychology, Johannes sciously perceived and categorized, knowledge about the value and
Gutenberg University, Mainz, Germany; Stefan C. Schmukle, Department potential consequences of different behavioral options is weighted
of Psychology, Westfälische Wilhelms University, Münster, Germany. and integrated, and a decision is made that activates corresponding
This research was supported by German Research Foundation Grants behavioral schemata (reflective processes). Second, perceptual in-
EG 143/2-1 and EG 143/2-3. We would like to thank Sarah Dudenhöffer, put directly triggers the activation of schematic elements in an
Sascha Haun, Sarah Hirschmüller, Eleanor Horn, Eva Kampmann, Juliane associative network, and spread of activation automatically leads
Stopfer, and Julia Tetzner for their help with data collection. We also thank
to the activation of behavioral schemata (impulsive processes).
Steffen Nestler and Marcus Roth for comments on an earlier version of the
Impulsive processes, moreover, are influenced by two broad mo-
article.
Correspondence concerning this article should be addressed to Mitja D. tivational orientations: approach and avoidance tendencies. These
Back, Johannes Gutenberg University, Mainz, Department of Psychology, tendencies interact with the automatic perception of situational
Personality Psychology and Psychological Assessment, 55099 Mainz, Ger- aspects and generally gear behavioral activation toward approach
many. E-mail: back@uni-mainz.de or avoidance behavior (Strack & Deutsch, 2004). According to

533
534 BACK, SCHMUKLE, AND EGLOFF

these concepts, social behavior is executed whenever activation of explicit self-concept of personality can be adequately described in
a behavioral schema exceeds a certain threshold. This activation terms of a few broad personality dimensions (Saucier & Goldberg,
may be triggered by reflective and impulsive processes: Behav- 2001). The most common personality taxonomy is the five-factor
ioral schemata thus form a common pathway to actual social model (McCrae & John, 1992), which assumes that five broad and
behavior. robust factors, often referred to as the Big Five, account for a
considerable amount of covariation between self-reported person-
A Behavioral Process Model of Personality ality traits (see Ashton et al., 2004; Saucier, 2003; Saucier, Geor-
giades, Tsaousis, & Goldberg, 2005, for alternative models). These
What significance does this have with respect to the role of Big Five factors of personality are commonly known as Neuroti-
personality in determining actual behavior? How are stable indi- cism, Extraversion, Openness to Experience, Agreeableness, and
vidual differences related to reflective and impulsive precursors of Conscientiousness (Goldberg, 1990; John & Srivastava, 1999;
behavior? Figure 1 shows a personality-oriented adaptation of the McCrae & John, 1992).
RIM, the behavioral process model of personality (BPMP). Ac- In terms of the BPMP, self-reported measures of the Big Five
cording to this model, actual social behavior is the result of both can be conceptualized as reflecting differences in propositional
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

impulsive and controlled processes, as described in the RIM representations of the self that result from the typical functioning
(Strack & Deutsch, 2004). Personality may be understood as the of reflective processes (consciously perceiving and categorizing
result of the typical functioning (across time and multiple situa- the situation, considering alternative options, deciding what to do,
tions) of both kinds of processes. On the one hand, the typical initiating an action) in situations that provoke meaningful differ-
operation of reflective processes— how people typically perceive ences in how people act. With respect to the Big Five, these
and categorize situations, which behavioral options they prefer, situations could, for instance, be described as stressful situations in
and how they deliberately realize these preferences— condense which people differentially behave anxiously or do not (neuroti-
into propositional representations of the self (e.g., “I am a person cism); situations that allow wide latitude for being expressive or
who likes to have fun with others”). Measures of these aspects of not and for seeking contact or not (extraversion); situations that are
the self in turn predict how behavior (e.g., telling a joke) in a given unusual or that offer wide latitude for either displaying innovative
situation (e.g., arriving at a party) is voluntarily controlled. Indi- or knowledgeable responses or not displaying them (openness);
vidual differences in propositional representations of the self may situations in which the interests or concerns of others may be at
be termed the explicit self-concept of personality. This is generally least as salient as one’s own and there is wide latitude for either
measured with direct measures of personality (e.g., a personality amplifying or minimizing interpersonal conflict (agreeableness);
questionnaire for the measurement of extraversion).1 and finally situations in which there is wide latitude for either
On the other hand, the typical functioning of impulsive processes— exercising effortful self-regulation or not doing so (conscientious-
how situational cues are automatically processed and what kinds of ness).2
actions are automatically performed—should lead to chronic as- Research has shown that self-reported personality differences
sociative links between associative network elements. Associative have meaningful consequences for individual, interpersonal, and
representations of the self (e.g., “me”–“cheerful”) can result from social–institutional life outcomes (Ozer & Benet-Martinez, 2006).
repeated activation of the self, as a concept in an associative Personality as measured by self-reports also expresses itself in
network, together with a pattern of impulsive behavioral activation people’s natural and virtual environments (Back, Schmukle, &
that is characterized by perceptual features of the situation (e.g., Egloff, 2008; Gosling, Ko, Mannarelli, & Morris, 2002; Vazire &
many others), an impulsive motivational tendency (e.g., approach- Gosling, 2004). Moreover, explicit personality characteristics cor-
ing), and a resulting spontaneous behavior (e.g., smiling) that in relate with self-reported daily behaviors (Grucza & Goldberg,
turn activates the trait concept (e.g., cheerful). The more often an 2007; Hong, Paunonen, & Slade, 2008; Paunonen, 2003). Extra-
individual executes such a course of action, the stronger her/his version, for example, predicts a self-reported estimation of the
association between the self and the respective trait concept will average number of parties attended per month (Paunonen, 2003).
be. Measures of associative representations of the self thus predict In contrast, surprisingly little is known about the influence of
the automatic execution of behavior in a given situation. Individual explicit personality differences on actually observable behavior
differences in associative representations of the self may be re- (Baumeister et al., 2007; Funder, 2001). In fact, in many cases, the
ferred to as the implicit self-concept of personality (Asendorpf,
Banse, & Mücke, 2002). This can be measured with indirect tests
1
of personality, for example, an implicit association test (IAT) for In line with suggestions made by Anastasi (1968) and De Houwer
the measurement of extraversion. Understanding and predicting (2006), we use the terms direct and indirect when referring to measurement
actual behavior therefore requires consideration of both the ex- procedures (e.g., a questionnaire for the measurement of extraversion and
plicit and the implicit self-concept of personality. an implicit association test for the measurement of extraversion). In con-
trast, we use the terms explicit and implicit when referring to outcomes of
these measures (e.g., explicit and implicit extraversion scores) and the
Explicit Self-Concept of Personality and the underlying constructs (e.g., the explicit self-concept of extraversion and the
Prediction of Behavior implicit self-concept of extraversion).
2
The description of features of situations that typically provoke Big
The most straightforward way to gain information about an Five–related behaviors does not imply that situations generally need to be
individual’s personality is to simply ask him/her what kind of classified in terms of the Big Five structure (for detailed analyses concern-
person s/he is. Consequently, self-reports are typically used to ing basic domains of situation variables see Saucier, Bel-Bahar, & Fernan-
measure personality. At a superordinate level, the structure of this dez, 2007; ten Berge & De Raad, 1999; Yang, Read, & Miller, 2006).
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 535
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 1. The behavioral process model of personality, a personality-oriented adaptation of the reflective–
impulsive model (Strack & Deutsch, 2004). Terms inside the dashed rectangle refer to characteristics of the
actual situation (state), and terms outside the dashed rectangle to stable characteristics of the person (trait).
Multiple dashed rectangles indicate multiple situations. IAT ⫽ implicit association test.

difference between observed and self-reported behavior is not even the behavior item would probably correlate almost perfectly—
considered. Research on the predictive validity of self-reported clearly an exaggeration of how much actual nervous behavior is
personality is often conducted in line with a “just ask them” predicted by self-reported neuroticism.
maxim: The behavioral criterion (e.g., for neuroticism) is mea- To fully understand how personality influences behavior, one
sured by asking the person how s/he behaves (e.g., “How often do must therefore observe the actual behavior of participants. In
you act nervously when confronted with a stressful situation?”). recent years, a number of important personality-oriented behav-
Such an approach is problematic for at least two reasons. First, ioral approaches have been developed and adopted. For example,
self-reported behavior (asking people what they do) and actual Funder, Furr, and Colvin (2000) developed the Riverside Behav-
behavior (observing what people do) are neither conceptually nor ioral Q-Sort (RBQ), a detailed scheme for the coding of social
empirically identical (e.g., Todd, Penke, Fasolo, & Lenton, 2007; behaviors in one-to-one social interactions. Borkenau and col-
West & Brown, 1975). People’s reported behavior differs from leagues (2004) created a variety of social situations designed to
their actual behavior due to introspective limits and response provoke personality-related behaviors, and Pennebaker and col-
tendencies (Greenwald & Banaji, 1995; Nisbett & Wilson, 1977). leagues developed a technique for measuring linguistic aspects of
For instance, people do not know exactly how they feel or behave behavior, the Linguistic Inquiry and Word Count (LIWC; Penne-
in a stressful situation. Moreover, they differ in the extent to which baker, Chung, Ireland, Gonzales, & Booth, 2007; Pennebaker,
they want to be seen as vulnerable and accordingly adjust their Francis, & Booth, 2001). Using these and similar behavioral ap-
self-reports of behavior. While the self-concept of one’s own proaches, studies on the accuracy of personality judgments, on
behavior is most certainly an interesting issue for investigation, it nonverbal communication, and on word use have successfully
cannot be equated with behavior itself (Gosling, John, Craik, & demonstrated that the direct measurement of some personality
Robins, 1998; Vazire & Mehl, 2008). Second, when exclusively traits correlates with actually observed behaviors (e.g., Berry &
counting on self-reports in investigations of personality– behavior Hansen, 2000; Borkenau & Liebler, 1992; Creed & Funder, 1997;
relations, there is often considerable item overlap between predic- DePaulo & Friedman, 1998; Eaton & Funder, 2003; Fast &
tor and criterion, as a result of which empirical findings become Funder, 2008; Levesque & Kenny, 1993; Mehl et al., 2006; Pen-
rather trivial. For instance, questionnaires for the measurement of nebaker & King, 1999; Riggio & Riggio, 2002).
neuroticism may also contain an item asking about nervous reac- On the basis of these results and the conceptual descriptions of
tions in stressful situations. In this case, both the personality and the Big Five (John & Srivastava, 1999; McCrae & John, 1992), a
536 BACK, SCHMUKLE, AND EGLOFF

number of behavioral correlates of self-reported personality can be thus be complemented by indirect tests of personality that measure
expected. Neuroticism should, for instance, predict nonverbal and implicit aspects of the self.
verbal uncertainty, especially in stressful situations. There has Among the first attempts to assess the implicit self-concept of
been some evidence to suggest that facets of self-reported neurot- personality were projective measures such as the Thematic Apper-
icism do indeed correlate with more restricted, awkward, negative, ception Test (TAT; Murray, 1943). Over the last few decades,
and nervous social behavior (Asendorpf, 1988; Creed & Funder, renewed attention to implicit measures has also led to the devel-
1997; Pennebaker & King, 1999; Pilkonis, 1977; Riggio & Riggio, opment of a wide range of new procedures for the indirect mea-
2002). Extraversion would be expected to manifest itself in situ- surement of personality (Banse & Greenwald, 2007; Bosson,
ations that allow wide latitude for being expressive (or not) and for Swann, & Pennebaker, 2000; De Houwer, 2006; Nosek, Green-
seeking contact (or not). Prior studies have revealed that self- wald, & Banaji, 2007). One of the most prominent indirect mea-
reported extraverts have a more powerful voice (Scherer, 1978) sures is the implicit association test (IAT; Greenwald, McGhee, &
and behave in a more expressive and sociable way (Borkenau & Schwartz, 1998). The IAT measures the strength of associations
Liebler, 1992; Eaton & Funder, 2003; Levesque & Kenny, 1993; between concepts by comparing response times in two combined
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Mehl et al., 2006; Riggio & Riggio, 2002). Openness should be discrimination tasks. Participants are required to sort stimuli rep-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

reflected in intellectual competence, creative and original ideas, resenting four concepts by using only two responses, each assigned
and a willingness to encounter unusual situations. However, little to two of the four concepts. The underlying assumption of the IAT
is known about the actual influence of openness on behavior. Open is that if two concepts are highly associated, the sorting task will
individuals seem, for example, to know more about current events be easier (i.e., faster) when the two associated concepts share the
(e.g., Beier & Ackermann, 2001). In a groundbreaking behavioral same response key than when they share different response keys.
study on interpersonal perception, Borkenau and colleagues (2004) The IAT can be applied to assess aspects of the implicit self-
were able to demonstrate that openness to experience is most concept of personality by combining the categorization of items
easily observed in situations requiring creative and unusual behav- into the categories me and others with the classification of items
ior. For agreeable individuals, helping others and not being ag- into two opposing personality trait categories.3 For example, the
gressive should be typical. Accordingly, some studies have shown IAT has been used to measure the implicit self-concept of extra-
that self-reported agreeable individuals behave in a more attentive version (Mierke & Klauer, 2003; Schmukle & Egloff, 2005). Here,
and friendly manner (Berry & Hansen, 2000; Borkenau & Liebler, categorization into the categories me and others was combined
1992; Mehl et al., 2006). Finally, conscientious persons would be with a classification of items into extraversion and introversion
expected to behave in a self-organized way and to abide by categories. By measuring the relative ease of categorizing me and
common arrangements. Some behavioral studies have found that extraversion items versus others and introversion items, as com-
conscientiousness fosters punctual and more formal behavior pared with me and introversion items versus others and extraver-
(Back, Schmukle, & Egloff, 2006; Borkenau & Liebler, 1992). sion items, the IAT effect thus represents an indicator of the
In sum, for each of the five dimensions, there is some evidence implicit self-concept of extraversion.
that the explicit self-concept of personality predicts actual and In contrast to other implicit measures of personality (Bosson et
socially relevant behavior. Nonetheless, given the theoretical im- al., 2000; Schmukle, 2005), personality IATs routinely demon-
portance of personality in the study of behavior, this body of strate good reliability (Back, Schmukle, & Egloff, 2005; Egloff,
evidence remains surprisingly small. Moreover, the few behavioral Schwerdtfeger, & Schmukle, 2005; Nosek et al., 2007), which is a
studies conducted have investigated only specific personality prerequisite for measuring individual differences. Moreover,
traits, have used single or few behavioral criteria, and have failed Schmukle, Back, and Egloff (2008) recently successfully applied
to specify a priori what kinds of behavior should theoretically be the IAT for the indirect measurement of the Big Five (see also
predicted by what kinds of personality traits. More research on the Grumm & Von Collani, 2007; Steffens & Schulze-König, 2006).
behavioral consequences of self-reported personality is evidently As is also the case for direct measures of personality, however,
needed (Baumeister et al., 2007; Funder, 2001). Specifically, more the key issue when it comes to evaluating the utility of indirect
systematic studies—including a multitude of trait-relevant situa- personality measures is predictive validity. An indirect measure of
tions and behavioral indicators that, moreover, are theoretically personality valuably complements direct personality measures
related to the Big Five dimensions and sufficiently aggregated over (i.e., questionnaires), given that it predicts relevant behavior even
observers, time, or situations—are desirable (Funder, 1999; Ken- when self-reported personality is controlled for. Consequently, in
rick & Funder, 1988). indirectly measuring personality, “the focus should be placed on
whether implicit measures predict relevant behaviour over and
above explicit measures” (Perugini & Banse, 2007, p. 260).
Implicit Self-Concept of Personality and the The predictive validity of indirect tests of personality has most
Prediction of Behavior clearly been demonstrated in the context of shyness and anxious-
ness. A shyness IAT predicted global ratings of shyness in con-
Actual behavior is triggered by both reflective processes, which
versation with an unfamiliar opposite-gender confederate of
are influenced by individual differences in propositional represen-
above-average physical attractiveness, as well as for more specific
tations of the self (the explicit self-concept of personality), and
impulsive processes, which are subject to the influence of various
associative representations of the self (the implicit self-concept of 3
This procedure applies to personality IATs. In contrast, attitude IATs
personality). For a more complete understanding of how and why combine the categorization of items into the categories good and bad with
people’s social behavior differs, self-reported personality must the classification of items into two opposing attitude object categories.
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 537

indicators of shyness, including gaze aversion (Asendorpf et al., Overview of the Present Study
2002). IATs indirectly measuring anxiousness predicted global
ratings of anxiousness in stressful speech situations (Egloff & The present study, the Mainz Observation of Behavior Study,
Schmukle, 2002; Schnabel, Banse, & Asendorpf, 2006), more investigated the predictive validity of explicit and implicit mea-
specific indicators of nervousness such as dysfluency of speech sures of personality. Applying an extensive behavioral approach,
(Egloff & Schmukle, 2002), and performance decrements follow- we aimed to determine how well explicit and implicit aspects of
ing failure (Egloff & Schmukle, 2002). Moreover, these IATs also the self-concept of personality predict people’s actual behavior.
showed incremental validity, that is, they predicted actual behavior We defined behavioral aspects of the Big Five on the basis of a
above and beyond explicit measures of personality (e.g., Asen- systematic investigation of theoretical and empirical approaches to
dorpf et al., 2002; Egloff & Schmukle, 2002; Schnabel et al., personality and social behavior and subsequently assigned a mul-
2006). titude of concrete actual behaviors to each of the five dimensions.
Banse and Fischer (2002) found some evidence for the predic- For the observation of these behaviors, we compiled a diverse set
tive validity of an aggressiveness IAT in the context of sports: The of relevant social situations (see Table 1), including speaking
IAT predicted the number of matches with penalties in a group of situations, interpersonal interactions, creativity tasks, helping sit-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

semiprofessional ice hockey players. In examining global ratings uations, performance tasks, body-oriented tasks, and unstructured
of angry and aggressive behavior upon being unjustly accused, situations. This resulted in more than 1 hr’s worth of videotaped
Steffens and Schulze-König (2006) found a significant correlation behavior for each participant. Several trained expert raters objec-
with an agreeableness IAT, whereas Schnabel and colleagues tively derived or rated predefined behavioral criteria. By aggre-
(2006) failed to find a correlation with an angriness IAT. More- gating across expert raters, more than 50 reliable behavioral cri-
over, Steffens and Schulze-König reported a correlation between a teria were assessed (see Table 2). A meaningful behavioral
conscientiousness IAT and one indicator of conscientious behavior criterion was subsequently adopted for each of the Big Five
(the number of errors made during a self-paced concentration test). dimensions by aggregating across a multitude of actually observed
While compelling evidence for the predictive validity of attitude behaviors that were theoretically related to the respective dimen-
IATs exists (see Greenwald, Poehlman, Uhlmann, & Banaji, in sion.
press, for a recent meta-analysis), it remains unclear exactly what These behavioral aggregates were then used to test the predic-
kinds of personality IATs routinely predict actual behavior. We tive validity of our direct and indirect measures of personality. To
aimed to systematically investigate, on the basis of a behavioral allow for a direct and fair comparison of the predictive validity of
process model, the predictive validity of various IATs designed to both kinds of measures, we applied the same descriptive person-
measure the implicit self-concept of personality. Most importantly, ality adjectives in both a Big Five questionnaire and a Big Five
we aimed to establish what kinds of personality IATs incremen- IAT. In line with our behavioral process model of personality, we
tally predict actual behavior above and beyond direct measures of formulated three hypotheses. First, explicit measures of personal-
personality. ity would predict the corresponding behavioral aggregates. Sec-

Table 1
Experimental Situations

Situation Description Average duration

Reception The experimenter welcomed the participant and noted time of arrival. 1 min
Small talk The experimenter opened with standardized small talk: “What is your name?” “What are you 2 min
studying?” “What’s it like?” The experimenter additionally asked whether the participant’s
cell phone was switched off.
Big Five IATs See text. 30 min
Photographs Pictures of the participants were taken in a standardized standing position. 2 min
Self-introduction Participants were seated in front of the camera and requested to introduce themselves: “Please 4 min
tell us about yourself, about your hobbies, what you are interested in, and so on.”
Vision of the future “Please describe what your life and the world in general will be like in 15 years.” 3 min
Concentration test Administration of a standardized concentration test (FAIR; Moosbrugger & Oehlschläger, 1996). 7 min
Brick pantomime “Please think about multiple uses of a brick and present your ideas—one by one—in pantomime 3 min
form.”
Short story Participants were instructed to write a short story in which the words air crash, firework, 7 min
chambermaid, middle ages, and supermarket were included.
Knowledge test The knowledge test consisted of 19 general knowledge questions in a multiple choice format 10 min
and 7 open-ended questions pertaining to more current topics.
Waiting situation Participants were left alone because the experimenter “had to see to something outside.” 2 min
Helping situation Just as the participant was about to leave, the experimenter asked if s/he could “help with the 5 min
preparation of another experiment.” The experimenter handed the participant some materials
(e.g., paper, pencils) and asked her/him to distribute them among five tables in another room.
The number of materials distributed to each table was recorded.
Sending back a questionnaire Participants were given a questionnaire along with instructions to fill it out and send it back in 1 week
exactly 1 week. The time of arrival of the questionnaire was recorded.

Note. FAIR ⫽ Frankfurt Attention Inventory.


538 BACK, SCHMUKLE, AND EGLOFF

Table 2
Behavioral Criteria for Each of the Big Five Personality Dimensions

Items Typical behavioral aspects Behavioral criteria ␣ Mexp1 Mexp2

Neuroticism
Anxious, nervous, fearful, Nonverbal nervousness, verbal uncertainty, Global behavior SI (rating) .83 5.00 0.81
uncertain, afraid, calm, negative self-view mentioned Global behavior vision of the future (rating) .89 5.00 0.81
relaxed, restful, at Global transcript SI (rating) .86 5.00 0.66
ease, balanced Gaze aversion SI (counting) .98 4.38 1.38
Tense body posture SI (rating) .74 4.38 0.75
Tense leg posture SI (rating) .88 4.25 0.72
Silence during SI (counting) .96 3.25 2.19
Reassuring whether cell phone is switched off
(yes/no) obj. 2.63 1.59
Reassuring questions in helping situation
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(counting) obj. 3.50 2.09


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Dysfluency of speech SI (filler words, LIWC;


Pennebaker et al., 2007, 2001) obj. 4.50 1.19
Negations SI (LIWC) obj. 3.75 0.88
Words related to anxiety and depression SI
(LIWC) obj. 4.50 0.66

Extraversion
Sociable, talkative, active, Expressive nonverbal and verbal behavior; Global behavior SI (rating) .83 4.88 1.09
impulsive, outgoing, impulsive behavior; social contact Global behavior vision of the future (rating) .89 4.88 1.06
shy, reticent, passive, sought; showy appearance Global transcript SI (rating) .88 5.00 0.88
deliberate, reserved Expressivity of facial expression SI (rating) .73 4.88 0.91
Loudness of voice SI (rating) .90 4.63 1.06
Number of words SI (LIWC) obj. 3.88 1.50
False alarms in concentration test (counting) obj. 1.75 1.44
Own questions during small talk (counting) .86 3.38 1.88
Second-person pronouns SI (LIWC) obj. 4.00 1.38
Other references SI (LIWC) obj. 3.88 1.31
Stylish dress (rating) .76 4.25 1.31
Flashy dress (rating) .66 4.75 1.47

Openness
Imaginative, civilized, Intellectual competence; creative and Global short story (rating) .67 4.75 0.78
well-educated, original ideas; openness to unusual Global transcript SI (rating) .89 4.75 0.72
interested, gifted, situations General knowledge (test) .70 4.50 0.75
unimaginative, Actual knowledge (test) .62 4.63 0.91
primitive, uneducated, Verbal eloquence SI (rating) .76 4.00 1.38
indifferent, limited Original and unusual brick categories (counting) .96 4.88 0.88
Number of words in short story (LIWC) obj. 4.13 1.25
Open answers in small-talk situation (rating) .90 4.25 2.00

Agreeableness
Trusting, well-meaning, Helpfulness; friendly and trustful Global behavior in helping situation (rating) .85 4.75 1.19
friendly, helpful, good- nonverbal and verbal behavior; Global transcript SI (rating) .80 4.88 0.53
natured, obstinate, compliant behavior; nonaggressiveness; Quality of help (counting) obj. 2.88 1.13
quarrelsome, hostile, social and selfless orientation mentioned Friendly voice in helping situation (rating) .93 4.63 0.69
hard-hearted, resentful Friendly voice in small talk (rating) .73 4.63 0.75
Attentive body posture in small talk (rating) .63 3.75 1.78
Checking out room in waiting situation (rating) .90 3.88 2.03
Aggressive–destructive brick categories
(counting) .98 4.25 1.31
Number of swear words SI (LIWC) obj. 4.00 1.53
Relative frequency of other- versus self-words
SI (LIWC) obj. 3.13 1.31
Words related to social processes SI (LIWC) obj. 3.63 1.34
Words related to family SI (LIWC) obj. 3.38 1.09
(table continues)
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 539

Table 2 (continued)

Items Typical behavioral aspects Behavioral criteria ␣ Mexp1 Mexp2

Conscientiousness
Meticulous, reliable, neat, Common arrangements adhered to; Global transcript SI (rating) .89 4.88 0.50
fussy, thorough, linguistic correctness; erroneous Minutes too late in attending experiment
careless, unreliable, behavior avoided; formal dress and (counting) obj. 5.00 1.47
chaotic, frivolous, appearance Lateness in sending back questionnaire
erratic (counting) obj. 4.88 0.59
Number of errors in short story (counting) .89 4.50 0.38
Understandability in small talk (rating) .79 3.75 1.50
Number of errors in concentration test
(counting) obj. 4.38 0.38
Slouching body posture SI (rating) .93 3.88 1.25
Formal dress (rating) .79 4.38 0.66
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Note. Items and behavioral criteria that indicate a low value on the respective personality dimension appear in italics. Mexp1 ⫽ mean of the expert ratings
for the respective personality dimension (possible range: 0 –5); Mexp2 ⫽ mean of the expert ratings for all other personality dimensions (possible range:
0 –5); SI ⫽ self-introduction; obj. ⫽ objective behavioral criteria; LIWC ⫽ Linguistic Inquiry and Word Count.

ond, implicit measures would also predict actual behavior. Third, to the Big Five dimensions of personality. Participants’ behav-
because explicit and implicit measures of personality are thought ior was videotaped throughout the course of the experiment.
to reflect the typical functioning of different kinds of processes
that activate behavior via a common pathway, implicit predictive Measures
validities would prove incremental when controlling for explicit
Personality IAT. The personality IAT—which consisted of
measures.
five subtests, each designed to measure one of the Big Five
Additionally, we aimed to establish whether the potential
personality dimensions—was administered on a personal com-
effects of direct and indirect measures on behavior are specif-
puter with the Inquisit program (Draine, 2001). In the follow-
ically related to the self-concept of personality as opposed to
ing, we present the extraversion subtest as an example of the
self-esteem. Since a positive–negative valence polarity is in-
procedure adopted in conducting the IAT subtests. In this
herent to most self-concept IATs, it has been recommended that
subtest, stimuli from the categories me (me, my, own, I, self)
self-concept IATs should “show construct validity for their
and others (they, your, them, you, others) as well as items from
specific semantic content as distinct from a general positive or
the categories extraversion (sociable, talkative, active, impul-
negative evaluation” (Schnabel, Asendorpf, & Greenwald,
sive, outgoing) and introversion (shy, reticent, passive, delib-
2008, p. 521). For this reason, we aimed to test whether pre-
erate, reserved) were presented. The IAT extraversion subtest
dictive validities of the Big Five IAT are influenced by con-
comprises five blocks. In the first block, comprising 20 trials
trolling for valence.
(each item was presented twice), participants practiced discrim-
inating items belonging to the category me from those belong-
Method ing to the category others (target discrimination). In Block 2,
the same procedure was conducted with respect to attribute
Participants
discrimination; participants were required to sort items into
Participants were 130 students (77 women, 53 men) of the extraversion and introversion categories. In Block 4, the
Johannes Gutenberg University in Mainz, Germany, recruited switched key assignment was practiced (20 trials). The critical
mainly from introductory psychology courses and through flyers in Blocks 3 and 5 each consisted of 60 trials. In these trials,
several university buildings. They participated in exchange for participants categorized items into two combined categories,
research participation credit or monetary compensation (20 €). each including the attribute and the target concept assigned to
Their average age was 22.43 years (SD ⫽ 3.98). the same key.
Participants were informed that they would be required to
Procedure make a series of category judgments. In each trial, a stimulus
word was displayed in the center of a computer screen. Cate-
After consenting to participate, participants were given a gory labels were displayed on the left- and right-hand sides of
questionnaire for the measurement of self-reported personality, the window. Participants used the letter A on the left-hand side
which they completed at home. A few days later, they partici- of the keyboard and the number 5 from the numeric keypad on
pated in a behavioral study lasting about 70 – 80 min that the right-hand side for their responses. They were further in-
included the indirect personality measures. The order and structed: “Please try to be as accurate though also as quick as
length of the experimental situations of this session are de- possible. If your selection is incorrect, you will see a red X. To
scribed in Table 1. These situations were designed to provoke a continue to the next judgment, you must make the correct
variety of socially relevant behaviors and behavioral outcomes selection.” Participants were told to keep their index fingers on
(see Behavioral criteria below) that are theoretically assignable the A and 5 keys throughout the experiment in order to facilitate
540 BACK, SCHMUKLE, AND EGLOFF

a fast response. Intertrial intervals were set to 150 ms. The mean of the five extraversion and the five reverse-scored introver-
computer recorded elapsed time between the start of each sion ratings).
stimulus presentation and the correct response. Mean latencies Valence measures. To control for the potential influence of
and error rates were displayed after each block. valence, we included two pure measures of valence: an indirect
IAT data were treated with an improved scoring algorithm and a direct measure. A self-esteem IAT was performed as an
(so-called D1 measure) as described by Greenwald, Nosek, and indirect measure of valence directly after completion of the Big
Banaji (2003): (a) trials with latencies greater than 10,000 ms were Five IATs. In this IAT, categorization of items into the cate-
eliminated; (b) error trials were included in the analysis by using gories me (me, my, own, I, self) and others (they, your, them,
the latency between stimulus presentation and correct response you, others) was combined with categorization of stimuli into
(built-in error penalty); (c) the mean latency for the critical trials of the categories like (nice, likeable, pleasant, good, positive) and
the me ⫹ extraversion block was subtracted from the mean latency dislike (annoying, dislikeable, repulsive, bad, negative). In-
for the critical trials of the me ⫹ introversion block; (d) the IAT structions, procedure, and computation of the IAT effect were
effect was computed by dividing this difference by the individual- analogous to those for the personality IATs. The same like and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

respondent reaction time standard deviation. Thus, the higher the dislike items were included in the personality questionnaire and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

IAT effect, the more extraverted the estimated implicit self- were rated on a scale ranging from 0 (not at all) to 5 (very
concept of a given participant. much). Again, participants were instructed to “please indicate
A total of five such subtests were presented. Since it was the extent to which the following attributes apply to you.” A
necessary to practice discriminating only the items me and direct measure of valence was computed as the mean rating for
others in the first IAT subtest administered, the personality IAT the five positive stimuli ratings and the five reverse-scored
consisted of a total of 21 blocks (5 blocks for the first subtest, negative stimuli.
4 blocks for each of the following four subtests). The IAT Behavioral criteria. For each of the five personality dimen-
subtests differed with respect to the attributes that were to be sions, a number of behavioral criteria were defined a priori. In
discriminated. In addition to the above-described discrimination predefining criteria, we referred to conceptual descriptions of
of extraversion versus introversion, we employed anxiety ver- the Big Five dimensions (Goldberg, 1990; John & Srivastava,
sus calmness; agreeableness versus disagreeableness; conscien- 1999; McCrae & John, 1992) as well as to prior research on
tiousness versus carelessness; and openness versus narrow- behavioral personality correlates (e.g., Asendorpf et al., 2002;
mindedness. In selecting these category labels, we refrained Borkenau & Liebler, 1992; Borkenau et al., 2004; Funder et al.,
from using technical terms of academic psychology, such as 2000; Mehl et al., 2006). In Table 2, each of the Big Five
neuroticism or emotional stability, in order to ensure compre- dimensions (column 1) is presented together with theoretically
hensibility for all participants. Stimuli selection was based on defined behavioral aspects (column 2) and the behavioral cri-
prototypical items for each category chosen from a list of 823 teria actually adopted (column 3). A number of behavioral
adjectives assigned to the Big Five personality dimensions criteria were objectively measured by counting behavioral oc-
(Ostendorf, 1994). Whenever feasible, we avoided the use of currences (e.g., lateness of attendance) or by applying the
negated adjectives common to many adjective lists (Goldberg, German adaptation (Wolf et al., 2008) of the Linguistic Inquiry
1992; Ostendorf, 1994) in order to rule out alternative response and Word Count (LIWC; Pennebaker et al., 2007, 2001; e.g.,
strategies, which may be facilitated by such adjectives when number of negations in the self-introduction). Other criteria
completing the IAT (e.g., sorting all stimuli that begin with were counted (e.g., number of errors in the short story) or rated
“un-” into one category). All stimuli used in the Big Five IAT (e.g., expressivity of facial expression) by two observers in
can be found in column 1 of Table 2. The presentation sequence order to ensure sufficient reliability. As can be seen in column
of the five IAT subtests, as well as the order of critical blocks, 4 of Table 2, these behavioral criteria proved reasonably reli-
was counterbalanced across participants. For the purpose of able, with a mean coefficient alpha of .83.
maintaining variation in experimental demands and minimizing In order to empirically validate whether theoretical assign-
the utilization of response strategies, we set breaks between ment of behavioral criteria to one of the Big Five dimensions
IAT subtests. During these breaks, participants were presented was warranted, we had eight independent experts rate each of
with TAT and Rorschach pictures and instructed to write down the 52 behavioral criteria with regard to their degree of belong-
their ideas concerning the situation (TAT: breaks 1 and 3) and ing to each of the Big Five dimensions. Ratings were performed
the inkblot (Rorschach: breaks 2 and 4) displayed. Participants on a scale ranging from 0 (not prototypical at all for this
wrote for 5 min per situation (TAT) and 3 min per inkblot dimension) to 5 (very prototypical for this dimension). As Table
(Rorschach). Their written ideas were not subjected to analysis. 2 shows, behavioral criteria were seen as prototypical for the
Self-reported personality. Participants explicitly rated the 50 dimension to which they had been assigned (column 5) and less
IAT attribute stimuli (see column 1 of Table 2) on a scale ranging prototypical for the remaining dimensions (column 6). More-
from 0 (not at all) to 5 (very much). They were instructed to over, expert ratings were highly consensual, with alpha coeffi-
“please indicate the extent to which the following attributes apply cients ranging from .91 for extraversion to .96 for agreeable-
to you.” Scores for each of the five personality dimensions (ex- ness. Finally, we aggregated across behavioral criteria in order
traversion, neuroticism, agreeableness, conscientiousness, and to obtain a single behavioral validation criterion for each Big
openness) were computed as the mean rating for the five stimuli Five dimension. To obtain reliability indexes for these criteria,
describing the positive pole of the respective dimension and the we created two parallel behavioral aggregate measures for each
five reverse-scored stimuli describing the negative pole (e.g., the personality dimension and computed reliabilities, which were
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 541

.71 (neuroticism), .88 (extraversion), .73 (openness), .62 Table 4


(agreeableness), and .50 (conscientiousness).4 Intercorrelations of Explicit and Implicit Personality Measures

Dimension 1 2 3 4 5
Results
ⴱⴱ
1. Neuroticism .25 ⫺.01 ⫺.03 ⫺.04 ⫺.21ⴱ
Descriptive Statistics 2. Extraversion ⫺.15† .31ⴱⴱⴱ .27ⴱⴱ .08 ⫺.03
3. Openness ⫺.13 .31ⴱⴱⴱ .00 .30ⴱⴱ .22ⴱ
Means, standard deviations, and reliabilities of the personality 4. Agreeableness ⫺.34ⴱⴱⴱ .14 .18ⴱ ⫺.02 .12
predictors can be found in Table 3. Internal consistencies proved 5. Conscientiousness ⫺.10 ⫺.09 .08 .24ⴱⴱ .01
satisfactory for all measures, with a mean coefficient alpha of .81
for explicit measures and .76 for implicit measures. Intercorrela- Note. Intercorrelations between implicit measures are shown above the
tions between personality measures were comparatively low, with diagonal, and intercorrelations between explicit measures are shown below
it. Explicit⫺implicit correlations (italicized) are shown on the diagonal.
a mean absolute intercorrelation of .18 for explicit measures and †
p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
.13 for implicit measures (see Table 4).5 Explicit–implicit corre-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

lations (explicit–implicit consistencies) were present for only neu- plicit personality. In addition, implicit agreeableness proved to be
This document is copyrighted by the American Psychological Association or one of its allied publishers.

roticism and extraversion (see diagonal of Table 4). a marginally significant predictor of agreeable behavior when
controlling for explicit agreeableness.
Main Analyses
Supplementary Analyses: A Question of Valence?
To examine the predictive validity of personality measures, we
computed correlations between each measure and the correspond- As outlined above, the predictive validity of self-concept IATs
ing theoretically derived behavioral validation criterion. Predictive may potentially be influenced by a valence confound (e.g., Banse
validities of explicit measures are presented in column 1 of & Greenwald, 2007; Schnabel et al., 2008). Specifically, valence
Table 5. All correlations were highly significant and considerably may have masked or produced the validity coefficients found for
higher than r ⫽ .30. An individual’s explicit self-concept of our Big Five IATs. Internal consistencies of both valence measures
personality thus predicts his/her actual behavior. This applied to all were satisfactory, with ␣ ⫽ .81 for explicit valence (M ⫽ 3.74,
Big Five dimensions. We subsequently correlated implicit person- SD ⫽ 0.55) and ␣ ⫽ .74 for implicit valence (M ⫽ 0.38, SD ⫽
ality measures and behavioral aggregates (see Table 5, column 2). 0.28). Explicit and implicit valence measures were uncorrelated
While no significant correlations were found for openness, agree- (r ⫽ –.01, ns). To check for the potentially confounding influence
ableness, or conscientiousness, implicit neuroticism and implicit of valence, we first computed correlations between valence and
extraversion predicted actual behavior. Absolute discriminant va- personality (see Table 6). As can be seen in column 2, there were
lidities averaged .11 for explicit and .09 for implicit measures, small correlations between implicit personality and the effect of
ranging from .01 to .26 for explicit measures and from .00 to .19 the self-evaluation IAT (mean r ⫽ .20). The valence confound
for implicit measures. Only two discriminant validity coefficients was, however, more pronounced for explicit personality (mean r ⫽
exceeded .20. Discriminant validities were thus considerably lower .40; see column 1). Second, we computed partial correlations to
than convergent validities (see Appendix). determine whether validity coefficients were changed by control-
Next, we analyzed whether implicit personality measures ling for valence. As shown in Table 7, predictive validities of both
showed incremental predictive validity. To this end, we computed direct and indirect measures of personality were not affected by
partial correlations between implicit personality and actual behav- controlling for valence.
ior for each of the five dimensions, in each case controlling for the
corresponding explicit personality measure. As can be seen in Discussion
column 3 of Table 5, implicit neuroticism and implicit extraversion
In the present study, we investigated the predictive validity of
significantly predicted actual behavior when controlling for ex-
both direct and indirect measures of personality. Using a system-

Table 3 4
In light of the broadness of the personality constructs and the limited
Descriptive Statistics of Explicit and Implicit Personality
number of multiply determined, single behavioral indicators, we regarded
Predictors these numbers as satisfactory. Additionally, we would like to point out that
when analyzing behavioral measures of personality, it is necessary to
Explicit Implicit
consider the individual specificity of behavioral responses (see Asendorpf,
Dimension M SD ␣ M SD ␣ 1988, for a detailed discussion). Many behavioral indicators of a trait (e.g.,
extraversion) that can be identified at the group level (e.g., gesturing,
Neuroticism 2.22 0.75 .84 ⫺0.31 0.28 .70 speaking in a loud voice) are not interchangeable at the individual level.
Extraversion 3.24 0.79 .87 0.02 0.38 .85 Some extraverts speak loudly but do not gesture, while others primarily
Openness 3.88 0.47 .73 0.32 0.30 .77 react with gestures. This unavoidably leads to low internal consistency of
Agreeableness 3.76 0.54 .75 0.40 0.29 .71 the behavioral aggregate. In a study on the behavioral assessment of
Conscientiousness 3.05 0.79 .85 0.28 0.26 .75 shyness, Asendorpf (1988) was, however, able to show that an aggregate
of behavioral measures with low intercorrelations can nevertheless be a
Note. Descriptives for explicit personality predictors are based on ques-
tionnaire data with a possible range from 0 to 5. Descriptives for implicit valid indicator of personality.
5
personality predictors are based on IAT data, which were treated with the Throughout the article, mean correlations were computed with Fisher’s
D1 measure. r-to-z formula.
542 BACK, SCHMUKLE, AND EGLOFF

Table 5 Table 7
Predicting Behavioral Aggregates From Explicit and Implicit Predicting Behavioral Aggregates From Explicit and Implicit
Personality Measures Personality Measures While Controlling for Valence

Implicit (controlled Implicit (controlled


Dimension Explicit Implicit for explicit) Dimension Explicit Implicit for explicit)

Neuroticism .36ⴱⴱⴱ .35ⴱⴱⴱ .29ⴱⴱ Neuroticism .26ⴱⴱ .38ⴱⴱⴱ .32ⴱⴱ


Extraversion .38ⴱⴱⴱ .31ⴱⴱⴱ .22ⴱ Extraversion .35ⴱⴱⴱ .34ⴱⴱⴱ .25ⴱⴱ
Openness .31ⴱⴱⴱ ⫺.14 ⫺.14 Openness .32ⴱⴱⴱ ⫺.09 ⫺.10
Agreeableness .35ⴱⴱⴱ .13 .15† Agreeableness .29ⴱⴱ .13 .14
Conscientiousness .30ⴱⴱⴱ ⫺.08 ⫺.09 Conscientiousness .27ⴱⴱ ⫺.08 ⫺.08
ⴱ ⴱⴱ ⴱⴱⴱ ⴱⴱ ⴱⴱⴱ

p ⬍ .10. p ⬍ .05. p ⬍ .01. p ⬍ .001. p ⬍ .01. p ⬍ .001.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

atic and extensive behavioral approach, we found the following


This document is copyrighted by the American Psychological Association or one of its allied publishers.

tive validities well exceeded the upper limit of the so-called


pattern of results. First, we were able to confirm the predictive personality coefficient, a term introduced by Mischel (1968) “to
importance of self-reported personality, as measured by a Big Five describe the correlation between .20 and .30 which is found
questionnaire: The explicit self-concept of personality is capable persistently when virtually any personality dimension inferred
of predicting actual behavior for all Big Five dimensions. Second, from a questionnaire is related to almost any conceivable criterion
it was shown that implicit personality also predicts actual behavior. involving responses sampled in a different medium—that is, not by
This was true for neuroticism and extraversion and not for open- another questionnaire” (p. 78). This neatly confirms that if “behavior
ness or conscientiousness. There was only a slight tendency for is appropriately aggregated over occasions and/or situations, the .30
implicit agreeableness to predict actual behavior. Third, reflecting personality barrier will routinely be breached” (Epstein, 1983, p. 180).
the hypothesized independent influence on actual behavior, the Our findings support and substantially extend prior research on the
predictive validity of implicit measures of personality remained predictive validity of direct measures of personality: In a nutshell,
stable when controlling for explicit measures. The prediction of self-reports of personality do indeed contribute to predictions of
actual behavior can thus be enhanced by including implicit mea- what people actually do.
sures of personality. We additionally demonstrated that valence
does not seem to pose a threat for the predictive validities reported.
Results are in line with our proposed model and underline the
The Predictive Validity of Indirect Measures of
relevance of direct and indirect personality measures in the pre- Personality
diction of actual behavior. Why then do we need indirect measures of personality when direct
measures are capable of predicting social behavior? It is necessary to
The Predictive Validity of Direct Measures of Personality bear in mind that the explicit personality– behavior relations observed
in our study are, although significant, far from being perfect. While
In the present study, we extended prior research on the behav-
direct measures of personality are necessary, they are presumably not
ioral correlates of directly measured personality by employing one
sufficient to accurately predict actual behavior. As outlined in the
large design to validate all Big Five dimensions. In sum, we were
BPMP, indirect measures of personality are thought to tap into other,
able to show that a variety of social behaviors are theoretically
more associatively structured aspects of the self (the implicit self-
definable and directly observable for each of the Big Five dimen-
concept of personality) that trigger social behavior in concert with
sions. By aggregating over these single behavioral criteria, we
propositional representations of the self (the explicit self-concept of
found clear evidence for the predictive validity of the explicit
personality) via a common pathway (behavioral schemata). We could
self-concept of personality: Explicit measures of neuroticism, ex-
show that, in accordance with our model, indirect measures of per-
traversion, openness, agreeableness, and conscientiousness each
sonality may valuably complement direct measures of personality:
predicted corresponding behavioral aggregates. Moreover, predic-
Implicit neuroticism and implicit extraversion indeed predicted actual
behavior. Moreover, this influence of the implicit self-concept of
Table 6 personality proved to be independent from the influence of the explicit
Correlations Between Personality Measures and Valence self-concept of personality.
This is in line with prior studies that have revealed incremental
Dimension and Explicit valence Implicit valence predictive validity of indirect measures of anxiety (Egloff &
absolute mean confound confound Schmukle, 2002; Schnabel et al., 2006) and shyness (Asendorpf et al.,
2002). For two aspects of personality, neuroticism and extraversion, it
Dimension
Neuroticism ⫺.45ⴱⴱⴱ ⫺.21ⴱ is now well documented that indirect measures constitute an impor-
Extraversion .31ⴱⴱ .20ⴱ tant addition in predicting actual behavior. This finding nicely paral-
Openness .38ⴱⴱⴱ .30ⴱⴱ lels research on the relation between personality and basic affective
Agreeableness .50ⴱⴱⴱ .19ⴱ and motivational tendencies. Neuroticism and extraversion have both
Conscientiousness .35ⴱⴱⴱ .10
Absolute mean .40 .20
been closely linked to basic affective dimensions and approach–
avoidance tendencies (Elliot & Trash, 2002; Watson, Wiese, Vaidya,
ⴱ ⴱⴱ ⴱⴱⴱ
p ⬍ .05. p ⬍ .01. p ⬍ .001. & Tellegen, 1999), which foster the automatic execution of behavior
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 543

(Strack & Deutsch, 2004). Behaving in a nervous and strained manner reflective processes seem to be more pronounced than for neuroticism
(neuroticism), as well as behaving expressively and cheerfully (ex- and extraversion. Open persons are often described as deliberately
traversion), seems to be strongly influenced by impulsive processes, considering novel and diverse ideas as well as acting accordingly.
specifically by the interplay of approach and avoidance tendencies, Conscientiousness is characterized by a careful consideration and
the automatic perception of situational cues, and the automatic acti- observation of one’s own present behaviors and the planning of future
vation of respective behavioral schemata. This can lead to chronic actions. On the other hand, one can argue that there is a certain degree
associative representations of the self, as outlined in the BPMP. of impulsivity in creative and imaginative acting (high openness) as
Consequently, IATs measuring these associations (e.g., an extraver- well as in unreliable and disorganized behaviors (low conscientious-
sion IAT) predict actual behavior (e.g., extraverted behavior) above ness). We would, however, consider these behaviors as not
and beyond explicit measures. being mainly driven by automatic motivational tendencies, as
Regarding agreeableness, the situation is less clear. We found some
described in the BPMP. Individual differences in creative and
evidence that implicit agreeableness might improve the prediction of
imaginative acting might not be due to automatic approach or
agreeable behavior when controlling for explicit agreeableness. How-
avoidance tendencies (e.g., impulsively approaching a creative
ever, this incremental predictive validity was smaller than that found
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

solution). Instead, they might result from mentally impulsive,


This document is copyrighted by the American Psychological Association or one of its allied publishers.

for neuroticism and extraversion and only marginally significant.


There are at least two reasons that might explain both the weaker automatic processes (innovative rearrangement of existing
predictive validity found in our study and the discrepant findings of schemata) that are then applied in a controlled fashion when the
prior studies (Banse & Fischer, 2002; Schnabel et al., 2006; Steffens situation demands it. Disorganized behaviors might not involve
& Schulze-König, 2006). First, while some behavioral aspects of approach or avoidance tendencies either (e.g., automatically
agreeableness are likely subject to the influence of impulsive pro- avoiding an organized desk). In contrast, they might best be
cesses (e.g., aggressive behavior), this is less likely to be true of other explained by the absence of effortful and controlled processes.6
aspects (e.g., modest behavior). As a consequence, associative repre- Future research might benefit from paying attention to this
sentations of the self-concept of agreeableness are less strongly af- interesting pattern of results. Our results do not definitely exclude
fected by a recurrent course of automatic behavior activation. Second, the possibility that using other indirect measures of openness and
social situations that provoke differences in agreeable behavior (situ- conscientiousness (particularly those highlighting the potentially
ations in which the concerns of others are more salient than one’s own impulsive aspects of these traits) might help to predict actual
concerns) might be less effective in activating one’s self-concept than behavior. Building on the present study, such research will not
would situations associated with neurotic and extraverted behavior. only help to better predict that elusive dependent variable called
This in turn may result in a less pronounced development of differ- actual behavior but might lead us toward a better understanding of
ences in self-trait associations during the automatic activation of
agreeable behavior. Future studies using different indirect measures of
agreeableness might improve our understanding of implicit agreeable- 6
ness. To test our categorization of neuroticism and extraversion as being
more impulsive than are openness and conscientiousness, we additionally
When it came to predicting actual behavior, implicit measures of
obtained expert ratings of the Big Five trait adjectives used in this study.
openness and conscientiousness, in contrast to those of neuroticism Eight trained observers rated each of the 50 trait adjectives on a
and extraversion, did not appear to be informative. A correlation reflectivity–impulsivity scale ranging from 1 (completely reflective) to 6
between indirectly measured conscientiousness and a single indi- (completely impulsive). In order to estimate the consensuality of impulsiv-
cator of conscientious behavior (number of errors made during a ity ratings, we calculated both the reliability of single ratings using the
self-paced concentration test) was found in a previous study (Ste- intraclass correlation ICC(2, 1) and the corresponding reliability of ratings
ffens & Schulze-König, 2006). Although a multitude of behavioral averaged across observers ICC(2, k; Shrout & Fleiss, 1979). Experts
indicators were used, the present study was not able to replicate substantially agreed in their impulsivity ratings, ICC(2, 1) ⫽ .52, p ⬍ .001,
this finding. While one can easily subject any trait to indirect ICC(2, k) ⫽ .90, p ⬍ .001. More importantly, the Big Five dimension to
measurement with an IAT by simply choosing two opposing trait which the adjectives belonged had a significant effect on impulsivity
ratings, F(4, 49) ⫽ 5.76, p ⬍ .001. Contrast tests revealed that trait
categories and some appropriate items, our study showed that these
adjectives belonging to neuroticism (M ⫽ 4.45, SD ⫽ 1.01) and extraver-
measures cannot always be assumed to be predictive of actual sion (M ⫽ 4.19, SD ⫽ 0.94) were judged to be significantly more
behavior. impulsive than were those belonging to openness (M ⫽ 2.59, SD ⫽ 0.72)
What might be the reason for the lack of predictive value of and conscientiousness (M ⫽ 3.06, SD ⫽ 1.59), t(45) ⫽ 4.62, p ⬍ .001. The
indirectly measured openness and conscientiousness? According to mean impulsivity score for agreeableness was in between (M ⫽ 3.38, SD ⫽
the BPMP, associative representations of the self are relevant for the 0.51). A second set of trained observers rated— on a scale ranging from 0
prediction of behavior only when they are developed on the basis of (not at all) to 5 (very much)—all 52 behavioral criteria with respect to the
a recurrent course of impulsive behavioral activation, including auto- extent to which they are subject to impulsive behavioral activation. Again,
matic approach or avoidance tendencies. We propose that the typical experts substantially agreed in their impulsivity ratings, ICC(2, 1) ⫽ .50,
activation of conscientious and open behavior implies less impulsive p ⬍ .001, ICC(2, k) ⫽ .89, p ⬍ .001. The Big Five dimension to which the
behavioral criteria belonged also had a significant effect on impulsivity
and automatic processes than does the typical activation of neurotic
ratings, F(4, 51) ⫽ 11.81, p ⬍ .001. Contrast tests revealed that behaviors
and extraverted behavior. As a consequence, indirect measures of belonging to neuroticism (M ⫽ 3.23, SD ⫽ 0.80) and extraversion (M ⫽
conscientiousness and openness are less predictive of actual behavior. 3.42, SD ⫽ 0.94) were judged to be significantly more impulsive than were
In fact, when looking at the conceptual descriptions of openness and those belonging to openness (M ⫽ 1.86, SD ⫽ 1.14) and conscientiousness
conscientiousness as well as at the behaviors related to these traits, (M ⫽ 0.95, SD ⫽ 0.57), t(47) ⫽ 6.50, p ⬍ .001. The mean impulsivity
one finds that impulsive processes seem to be less pronounced and score for agreeableness was again in between (M ⫽ 2.28, SD ⫽ 1.01).
544 BACK, SCHMUKLE, AND EGLOFF

and a refined framework of personality and social behavior in would be extremely interesting to further examine how much of
general. the information captured by other-perceptions is also reflected in
the implicit self-concept of personality. We would expect the
Not a Question of Valence variance of other-perception to be at least partly explained by the
implicit self-concept of personality. Most importantly, however,
In the present study, we indirectly measured personality with the use of all three important sources of information might lead to
five independent IATs, each designed to measure one of the Big even better predictions concerning how people actually behave.
Five dimensions. This approach has previously been questioned in Such studies would also contribute to a deeper theoretical un-
light of the potentially confounding role of valence (e.g., Banse & derstanding of personality and actual social behavior, and in doing
Greenwald, 2007; Schnabel et al., 2008). It has been argued that, so expand the BPMP. Other-perceptions interact with the explicit
due to the positive–negative valence polarity of each of the single and implicit self-concepts of the target in two important ways (cf.,
personality IATs (e.g., agreeable vs. disagreeable), results reflect Swann, 1987; Swann & Bosson, 2008). On the one hand, the
nothing other than multiple measures of implicit self-esteem. Ac- impressions which perceivers form are influenced by the explicit
cordingly, Banse and Greenwald (2007) recommend that “this and implicit self-concepts of the target person, given that these
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

approach might wisely be rejected.” At the very least, such a impressions are based on the observable cues (actual behavior)
valence confound “may represent a threat to the validity of IAT produced by the target’s explicit and implicit self (Brunswik, 1956;
measures” (Schnabel et al., 2006, p. 388), masking or artificially Funder, 1999). On the other hand, reputation might influence the
producing significant validity coefficients for personality IATs, explicit and implicit self. What others think about a target person
depending on the sign of the correlation between valence and the influences their behavior toward that person (via impulsive and
behavioral indicators under consideration. reflective processes). This behavior of others constitutes important
If these objections indeed hold—that is, if our Big Five IAT situational as well as self-related cues for the target person. These
predominantly reflects valence—then certain empirical conse- cues are subsequently processed as described in the BPMP and
quences should be observable. First, personality IATs should ultimately influence the target’s explicit and implicit self-concepts.
strongly correlate with a self-esteem IAT that purely measures The role of self-perception can be viewed in a similar way.
self-valence associations. Second, controlling for valence should According to self-perception theory (Bem, 1967, 1972), people
substantially change the validity coefficients of the personality infer what they are like by observing what they do. This important
IATs. In the present study, we were able to show that (a) corre- personality process fits neatly into the BPMP. Self-perception
lations with valence were rather modest for implicit measures and processes emerge when a person’s own actual behavior serves as
stronger for explicit measures, and even more importantly, (b) a situational cue that is perceived by the acting person him- or
controlling for valence did not change validity coefficients in any herself. With respect to the BPMP (see Figure 1), this would be
way for either explicit or implicit measures of personality. In sum, represented by adding a feedback loop from actual behavior to
explicit and implicit personality measures are not just multiple perception. Interestingly, self-perception can accordingly be con-
measures of self-esteem, and their predictive validities are not ceptualized in a reflective and an impulsive, automatic way, lead-
explained by valence. ing to calibration of the explicit or implicit self-concept of per-
sonality. These dual-process feedback mechanisms can potentially
Expanding the BPMP: Other-Perception (Reputation) and be used to increase self-knowledge: By increasing the salience of
Self-Perception people’s automatic behavior, controlled self-perception might lead
to a more accurate explicit self-concept of personality as well as
In describing the BPMP, we focused on the reflective and greater consistency between explicit and implicit self-concepts
impulsive processes that lead from situational cues to actual social (Egloff, Weck, & Schmukle, 2008).
behavior. We further showed how the explicit and implicit self-
concepts of personality can be understood as resulting from the Future Prospects for the Prediction of Actual Behavior
typical functioning of these processes. In the following, we discuss
two important aspects of personality that were not explicitly ad- We presented a behavioral process model that clearly specified
dressed in our model but that might be incorporated in a refined why and how the self-concept of personality should prove relevant
and expanded behavioral process model of personality: other- when it comes to predicting what people actually do. The present
perception (reputation) and self-perception. study further contributed a meaningful amount of empirical data
How others see us (reputation) certainly contains relevant in- underscoring the fact that personality really does matter. More-
formation with respect to what we are like. Importantly, other over, it provided clear evidence that using direct and indirect
people have even more direct access to a person’s actual behavior measures of personality as well as a multitude of theoretically
than does the individual him- or herself (Funder, 1999; Hofstee, derived and aggregated behavioral indicators substantially in-
1994). As a consequence, other-reports can predict behavior above creases the accuracy of predictions with respect to actual behavior.
and beyond self-reports of personality (e.g., Kolar, Funder, & In the following, we describe a number of future prospects for
Colvin, 1996; Vazire & Mehl, 2008). As we were able to show, research on personality and actual behavior and further show how
indirect tests of personality constitute a further valuable source of the BPMP can fruitfully be utilized for a better understanding and
information. Future studies should examine relations between all prediction of what people actually do.
three kinds of personality measures— other-reports, direct tests of First, it might be of particular interest to compare the predictive
personality, and indirect tests of personality—as well as their joint validity of different direct and indirect measures of personality.
contribution to the prediction of behavior. As a case in point, it With regard to predicting self-reported behaviors, a number of
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 545

different personality questionnaires have already been compared matic perceptions and evaluations of specified situational cues,
(e.g., Grucza & Goldberg, 2007). A comparison of the predictive their automatic approach or avoidance tendencies, the ease of
validities of different personality inventories with regard to actual activation of different concepts, and so forth.
behavior may also prove worthwhile. In conducting such a com- For both the comparison of existing direct and indirect measures
parison, researchers may find it useful to determine how different of personality and the development and evaluation of new mea-
characteristics of personality scales, such as for example the num- sures, actual behavior is in our view the most convincing criterion.
ber of items or the abstractness of scales (factors vs. facets), relate Most importantly, therefore, we encourage researchers to conduct
to their predictive validities. Moreover, traits that are not ade- systematic and realistic behavioral studies with a variety of rele-
quately represented in the five-factor model, such as honesty, vant social situations that allow measurement of multiple behav-
could also be included (Ashton et al., 2004; Saucier et al., 2005). ioral indicators. In light of the extraordinary personal, financial,
Similarly, a comparison of indirect tests of personality might and time resources that are necessary for thorough behavioral
include shorter variants of the IAT (Sriram & Greenwald, in press) studies, one might ask: Is it worth the effort? In our eyes, there is
and different approaches to measuring multiple constructs with the simply no alternative if we hope to learn more about individual
IAT (Greenwald, 2005), as well as other alternative indirect mea- differences in what people actually do. As the considerations in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

sures (see Banse & Greenwald, 2007; Bosson et al., 2000; De this article might suggest, we are certain that research on person-
Houwer, 2006; Nosek et al., 2007, for recent overviews). In ality and actual behavior is to become even more exciting as we
choosing individual differences for indirect measurement, one come closer to understanding what people actually do and why
should not restrict the selection to those aspects that are well they differ in their behavior.
established in the direct measurement of personality. It may well
be interesting to examine those aspects of the implicit self-concept References
of personality that theoretically relate to actual behavior on ac-
count of the impulsive dynamics of their development, as outlined Anastasi, A. (1968). Psychological testing (3rd ed.). New York: Macmil-
in the BPMP. lan.
Asendorpf, J. B. (1988). Individual response profiles in the behavioral
The goal of the present study was to analyze the predictive
assessment of personality. European Journal of Personality, 2, 155–167.
validity of personality at the level of broad behavioral aggregates, Asendorpf, J. B., Banse, R., & Mücke, D. (2002). Double dissociation
each of which comprised several behavioral indicators measured in between implicit and explicit personality self-concept: The case of shy
meaningful social situations. In real-life settings, these behavioral behavior. Journal of Personality and Social Psychology, 83, 380 –393.
indicators are almost always the result of both impulsive and Ashton, M. C., Lee, K., Perugini, M., Szarota, P., De Vries, R. E., Di Blas,
reflective processes. Other research settings might, however, allow L., et al. (2004). A six-factor structure of personality-descriptive adjec-
more controlled behavioral indicators to be disentangled from tives: Solutions from psycholexical studies in seven languages. Journal
more spontaneous indicators. Such studies might allow an analysis of Personality and Social Psychology, 86, 356 –366.
of whether self-reports predict more controlled behaviors whereas Back, M. D., Schmukle, S. C., & Egloff, B. (2005). Measuring task-
indirect tests predict more spontaneous behaviors (double dissoci- switching ability in the implicit association test. Experimental Psychol-
ogy, 52, 167–179.
ation; Asendorpf et al., 2002; Egloff & Schmukle, 2002). This
Back, M. D., Schmukle, S. C., & Egloff, B. (2006). Who is late and who
approach could then be compared with that of other models of is early? Big Five personality factors and punctuality in attending
predictive validity (Perugini, 2005). Of further interest is a con- psychological experiments. Journal of Research in Personality, 40,
sideration of potential moderators of the predictive validity of 841– 848.
explicit and implicit self-concepts of personality. Impulsivity (vs. Back, M. D., Schmukle, S. C., & Egloff, B. (2008). How extraverted is
controlledness) as a trait, for instance, might influence how well honey.bunny77@hotmail.de? Inferring personality from e-mail add-
actual behavior is predicted by direct or indirect measures of resses. Journal of Research in Personality, 42, 1116 –1122.
personality (cf. Hofmann, Gschwendner, Friese, Wiers, & Schmitt, Banse, R., & Fischer, I. (2002, July). Implicit and explicit aggressiveness
2008; Hofmann, Gschwendner, & Schmitt, 2005). and the prediction of aggressive behaviour. Paper presented at the 11th
In addition, the BPMP may be useful in developing new direct Conference on Personality of the European Association of Personality
Psychology, Jena, Germany.
and indirect measures that are more closely related to behavioral
Banse, R., & Greenwald, A. G. (2007). Personality and implicit social
activation. According to our model, both the explicit and the cognition research: Past, present and future. European Journal of Per-
implicit self-concept of personality mainly reflect condensed past sonality, 21, 371–382.
behavioral activation (which predicts future behavior). Classical Bargh, J. A., & Chartrand, T. L. (1999). The unbearable automaticity of
direct and indirect tests of personality, as it were, measure by- being. American Psychologist, 54, 462– 479.
products of typical behavior. An attempt may also be made to Baumeister, R. F., Vohs, K. D., & Funder, D. C. (2007). Psychology as the
measure more proximate individual differences in the processes science of self-reports and finger movements: Whatever happened to
that are involved in reflective and impulsive behavioral activation actual behavior? Perspectives on Psychological Science, 2, 396 – 403.
(see Figure 1). With regard to reflective processes, it may be Beier, M. E., & Ackermann, P. L. (2001). Current-events knowledge in
possible, for example, to directly measure how people differ in adults: An investigation of age, intelligence, and nonability determi-
nants. Psychology and Aging, 16, 615– 628.
their typical conscious perceptions and categorizations of different
Bem, D. J. (1967). Self-perception: An alternative interpretation of cogni-
situational cues or their typical processes of thinking, deciding, tive dissonance phenomena. Psychological Review, 74, 183–200.
and intending given specified situations (Endler, Hunt, & Rosen- Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances
stein, 1962; Mischel & Shoda, 1995). Similarly, typical impulsive in experimental social psychology (Vol. 6, pp. 1– 62). New York: Aca-
processes might be measured by means of indirect procedures. demic Press.
These could, for example, reflect how people differ in their auto- Berry, D. S., & Hansen, J. S. (2000). Personality, nonverbal behavior, and
546 BACK, SCHMUKLE, AND EGLOFF

interaction quality in female dyads. Personality and Social Psychology Goldberg, L. R. (1992). The development of markers for the Big-Five
Bulletin, 26, 278 –292. structure. Psychological Assessment, 4, 26 – 42.
Borkenau, P., & Liebler, A. (1992). Trait inferences: Sources of validity at Gosling, S. D., John, O. P., Craik, K. H., & Robins, R. W. (1998). Do
zero acquaintance. Journal of Personality and Social Psychology, 62, people know how they behave? Self-reported act frequencies compared
645– 657. with on-line codings by observers. Journal of Personality and Social
Borkenau, P., Mauer, N., Riemann, R., Spinath, F. M., & Angleitner, A. Psychology, 74, 1337–1349.
(2004). Thin slices of behavior as cues of personality and intelligence. Gosling, S. D., Ko, S. J., Mannarelli, T., & Morris, M. E. (2002). A room
Journal of Personality and Social Psychology, 86, 599 – 614. with a cue: Personality judgments based on offices and bedrooms.
Bosson, J. K., Swann, W. B., & Pennebaker, J. W. (2000). Stalking the Journal of Personality and Social Psychology, 82, 379 –398.
perfect measure of implicit self-esteem: The blind men and the elephant Greenwald, A. G. (2005). The Multifactor Trait IAT. Unpublished manu-
revisited. Journal of Personality and Social Psychology, 79, 631– 643. script, University of Washington.
Brunswik, E. (1956). Perception and the representative design of experi- Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition:
ments. Berkeley: University of California Press. Attitudes, self-esteem, and stereotypes. Psychological Review, 102,
Creed, A. T., & Funder, D. C. (1997). Social anxiety: From the inside and 4 –27.
Greenwald, A. G., Banaji, M. R., Rudman, L. A., Farnham, S. D., Nosek, B. A.,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

outside. Personality and Individual Differences, 25, 19 –33.


& Mellot, D. S. (2002). A unified theory of implicit attitudes, stereotypes,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

De Houwer, J. (2006). What are implicit measures and why are we using
them? In R. W. Wiers & A. W. Stacy (Eds.), The handbook of implicit self-esteem, and self-concept. Psychological Review, 109, 3–25.
cognition and addiction (pp. 11–28). Thousand Oaks, CA: Sage. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring
DePaulo, B. M., & Friedman, H. S. (1998). Nonverbal communication. In individual differences in implicit cognition: The implicit association test.
D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social Journal of Personality and Social Psychology, 74, 1464 –1480.
psychology (4th ed., Vol. 2, pp. 3– 40). New York: McGraw–Hill. Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding
Dijksterhuis, A., & Nordgren, L. F. (2006). A theory of unconscious and using the implicit association test: I. An improved scoring algorithm.
thought. Perspectives on Psychological Science, 1, 95–109. Journal of Personality and Social Psychology, 85, 197–216.
Draine, S. C. (2001). Inquisit: Version 1.32. Seattle, WA: Millisecond Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R. (in press).
Understanding and using the implicit association test: III. Meta-analysis of
Software.
predictive validity. Journal of Personality and Social Psychology.
Eaton, L. G., & Funder, D. C. (2003). The creation and consequences of the
Grucza, R. A., & Goldberg, L. R. (2007). The comparative validity of 11
social world: An interactional analysis of extraversion. European Jour-
modern personality inventories: Predictions of behavioral acts, infor-
nal of Personality, 17, 375–395.
mant reports, and clinical indicators. Journal of Personality Assessment,
Egloff, B., & Schmukle, S. C. (2002). Predictive validity of an implicit
89, 167–187.
association test for assessing anxiety. Journal of Personality and Social
Grumm, M., & Von Collani, G. (2007). Measuring Big-Five personality
Psychology, 83, 1441–1455.
dimensions with the implicit association test: Implicit personality traits
Egloff, B., Schwerdtfeger, A., & Schmukle, S. C. (2005). Temporal sta-
or self-esteem? Personality and Individual Differences, 43, 2205–2217.
bility of the Implicit Association Test–anxiety. Journal of Personality
Hofmann, W., Gschwendner, T., Friese, M., Wiers, R. W., & Schmitt, M.
Assessment, 84, 82– 88.
(2008). Working memory capacity and self-regulatory behavior: Toward
Egloff, B., Weck, F., & Schmukle, S. C. (2008). Thinking about anxiety
an individual differences perspective on behavior determination by
moderates the relationship between implicit and explicit anxiety mea-
automatic versus controlled processes. Journal of Personality and Social
sures. Journal of Research in Personality, 42, 771–778.
Psychology, 95, 962–977.
Elliot, A. J., & Trash, T. M. (2002). Approach-avoidance motivation in
Hofmann, W., Gschwendner, T., & Schmitt, M. (2005). On implicit-
personality: Approach and avoidance temperaments and goals. Journal explicit consistency: The moderating role of individual differences in
of Personality and Social Psychology, 82, 804 – 818. awareness and adjustment. European Journal of Personality, 19, 25– 49.
Endler, N. S., Hunt, J. M., & Rosenstein, A. J. (1962). An S-R inventory Hofstee, W. K. B. (1994). Who should own the definition of personality?
of anxiousness. Psychological Monographs, 76(17), 1–33. European Journal of Personality, 8, 149 –162.
Epstein, S. (1983). The stability of confusion: A reply to Mischel and Hong, R. Y., Paunonen, S. V., & Slade, H. P. (2008). Big Five personality
Peake. Psychological Review, 90, 179 –184. factors and the prediction of behavior: A multitrait-multimethod ap-
Epstein, S. (1994). Integration of the cognitive and the psychodynamic proach. Personality and Individual Differences, 45, 160 –166.
unconscious. American Psychologist, 49, 709 –724. John, O. P., & Srivastava, S. (1999). The Big Five Trait taxonomy: History,
Fast, L. A., & Funder, D. C. (2008). Personality as manifest in word use: measurement, and theoretical perspectives. In L. A. Pervin & O. P. John
Correlations with self-report, acquaintance report, and behavior. Journal (Eds.), Handbook of personality: Theory and research (2nd ed., pp.
of Personality and Social Psychology, 94, 334 –346. 102–138). New York: Guilford.
Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: Kenrick, D. T., & Funder, D. C. (1988). Profiting from controversy:
The MODE model as an integrative network. In M. P. Zanna (Ed.), Lessons from the person-situation debate. American Psychologist, 43,
Advances in experimental social psychology (Vol. 23, pp. 75–109). San 23–34.
Diego, CA: Academic Press. Kolar, D. W., Funder, D. C., & Colvin, C. (1996). Comparing the accuracy
Funder, D. C. (1999). Personality judgment: A realistic approach to person of personality judgments by the self and knowledgeable others. Journal
perception. San Diego, CA: Academic Press. of Personality, 64, 311–337.
Funder, D. C. (2001). Personality. Annual Review of Psychology, 52, Levesque, M. J., & Kenny, D. A. (1993). Accuracy of behavioral predic-
197–221. tions at zero acquaintance: A social relations analysis. Journal of Per-
Funder, D. C., Furr, R. M., & Colvin, C. R. (2000). The Riverside sonality and Social Psychology, 65, 1178 –1187.
Behavioral Q-Sort: A tool for the description of social behavior. Journal McCrae, R. R., & John, O. P. (1992). An introduction to the five-factor
of Personality, 68, 451– 489. model and its applications. Journal of Personality, 60, 175–215.
Goldberg, L. R. (1990). An alternative “description of personality”: The Mehl, M. R., Gosling, S. D., & Pennebaker, J. W. (2006). Personality in its
Big-Five factor structure. Journal of Personality and Social Psychology, natural habitat: Manifestations and implicit folk theories of personality in
59, 1216 –1229. daily life. Journal of Personality and Social Psychology, 90, 862– 877.
PREDICTING ACTUAL BEHAVIOR FROM PERSONALITY 547

Mierke, J., & Klauer, K. C. (2003). Method-specific variance in the Schmukle, S. C., Back, M. D., & Egloff, B. (2008). Validity of the
implicit association test. Journal of Personality and Social Psychology, five-factor model for the implicit self-concept of personality. European
85, 1180 –1192. Journal of Psychological Assessment, 24, 263–272.
Mischel, W. (1968). Personality and assessment. New York: Wiley. Schmukle, S. C., & Egloff, B. (2005). A latent state-trait analysis of
Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of implicit and explicit personality measures. European Journal of Psycho-
personality: Reconceptualizing situations, dispositions, dynamics, and the logical Assessment, 21, 100 –107.
invariance in personality structure. Psychological Review, 102, 246 –268. Schnabel, K., Asendorpf, J. B., & Greenwald, A. G. (2008). Using implicit
Moosbrugger, H., & Oehlschläger, J. (1996). Frankfurter Aufmerk- association tests for the assessment of implicit personality self-concept.
samkeits-Inventar: FAIR [Frankfurt Attention Inventory: FAIR]. Bern, In G. J. Boyle, G. Matthews, & H. Saklofske (Eds.), Handbook of
Switzerland: Huber. personality theory and testing (pp. 508 –528). London: Sage.
Murray, H. A. (1943). Thematic Apperception Test: Manual. Cambridge, Schnabel, K., Banse, R., & Asendorpf, J. B. (2006). Assessment of implicit
MA: Harvard University Press. personality self-concept using the implicit association test (IAT): Con-
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: current assessment of anxiousness and angriness. British Journal of
Verbal reports on mental processes. Psychological Review, 84, 231–259. Social Psychology, 45, 373–396.
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2007). The implicit Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

association test at age 7: A methodological and conceptual review. In J. A.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

assessing rater reliability. Psychological Bulletin, 86, 420 – 428.


Bargh (Ed.), Social psychology and the unconscious: The automaticity of Sriram, N., & Greenwald, A. G. (in press). The brief implicit association
higher mental processes (pp. 265–292). New York: Psychology Press. test. Experimental Psychology.
Ostendorf, F. (1994). Zur Taxonomie deutscher Dispositionsbegriffe [A Steffens, M. C., & Schulze-König, S. (2006). Predicting spontaneous
taxonomy of German dispositional terms]. In W. Hager & M. Hassel- Big-Five behavior with implicit association tests. European Journal of
horn (Eds.), Handbuch deutschsprachiger Wortnormen (pp. 382– 441). Psychological Assessment, 22, 13–20.
Göttingen, Germany: Hogrefe. Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of
Ozer, D. J., & Benet-Martinez, V. (2006). Personality and the prediction of social behavior. Personality and Social Psychology Review, 8, 220 –247.
consequential outcomes. Annual Review of Psychology, 57, 401– 421.
Swann, W. B., Jr. (1987). Identity negotiation: Where two roads meet.
Paunonen, S. V. (2003). Big Five factors of personality and replicated
Journal of Personality and Social Psychology, 53, 1038 –1051.
predictions of behavior. Journal of Personality and Social Psychology,
Swann, W. B., Jr., & Bosson, J. K. (2008). Identity negotiation: A theory
84, 411– 422.
of self and social interaction. In O. P. John, R. W. Robins, & L. A.
Pennebaker, J. W., Chung, C. K., Ireland, M., Gonzales, A., & Booth, R. J.
Pervin (Eds.), Handbook of personality: Theory and research (pp. 448 –
(2007). The development and psychometric properties of LIWC2007.
471). New York: Guilford Press.
Austin, TX: LIWC.net.
ten Berge, M. A., & De Raad, B. (1999). Taxonomies of situations from a
Pennebaker, J. W., Francis, M. E., & Booth, R. J. (2001). Linguistic Inquiry
trait psychological perspective: A review. European Journal of Person-
and Word Count: LIWC2001. Mahwah, NJ: Erlbaum.
ality, 13, 337–360.
Pennebaker, J. W., & King, L. A. (1999). Linguistic styles: Language use
Todd, P. M., Penke, L., Fasolo, B., & Lenton, A. P. (2007). Different cognitive
as an individual difference. Journal of Personality and Social Psychol-
processes underlie human mate choices and mate preferences. Proceedings
ogy, 77, 1296 –1312.
Perugini, M. (2005). Predictive models of implicit and explicit attitudes. of the National Academy of Sciences, USA, 104, 15011–15016.
British Journal of Social Psychology, 44, 29 – 45. Vazire, S., & Gosling, S. D. (2004). e-perceptions: Personality impressions
Perugini, M., & Banse, R. (2007). Personality, implicit self-concept and based on personal websites. Journal of Personality and Social Psychol-
automaticity. European Journal of Personality, 21, 257–261. ogy, 87, 123–132.
Pilkonis, P. A. (1977). The behavioral consequences of shyness. Journal of Vazire, S., & Mehl, M. R. (2008). Knowing me, knowing you: The accuracy
Personality, 45, 596 – 611. and unique predictive validity of self-ratings and other-ratings of daily
Riggio, H. R., & Riggio, R. E. (2002). Emotional expressiveness, extra- behavior. Journal of Personality and Social Psychology, 95, 1202–1216.
version, and neuroticisms: A meta-analysis. Journal of Nonverbal Be- Watson, D., Wiese, D., Vaidya, J., & Tellegen, A. (1999). The two general
havior, 26, 195–218. activation systems of affect: Structural findings, evolutionary consider-
Saucier, G. (2003). Factor structure of English-language personality type- ations, and psychobiological evidence. Journal of Personality and Social
nouns. Journal of Personality and Social Psychology, 85, 695–708. Psychology, 76, 820 – 838.
Saucier, G., Bel-Bahar, T., & Fernandez, C. (2007). What modifies the West, S. G., & Brown, T. J. (1975). Physical attractiveness, the severity of
expression of personality tendencies? Defining basic domains of situa- emergency and helping: A field experiment and interpersonal simula-
tion variables. Journal of Personality, 75, 479 –504. tion. Journal of Experimental Social Psychology, 11, 531–538.
Saucier, G., Georgiades, S., Tsaousis, I., & Goldberg, L. R. (2005). The Wilson, T. D., Lindsley, S., & Schooler, T. Y. (2000). A model of dual
factor structure of Greek personality adjectives. Journal of Personality attitudes. Psychological Review, 107, 101–126.
and Social Psychology, 88, 856 – 875. Wolf, M., Horn, A. B., Mehl, M. R., Haug, S., Pennebaker, J. W., & Kordy,
Saucier, G., & Goldberg, L. R. (2001). Lexical studies of indigenous H. (2008). Computergestützte quantitative Textanalyse: Äquivalenz und
personality factors: Premises, products, and prospects. Journal of Per- Robustheit der deutschen Version des Linguistic Inquiry and Word
sonality, 69, 847– 879. Count [Computer-aided quantitative text analysis: Equivalence and re-
Scherer, K. R. (1978). Personality inference from voice quality: The loud voice liability of the German adaptation of the Linguistic Inquiry and Word
of extraversion. European Journal of Social Psychology, 8, 467– 487. Count]. Diagnostica, 54, 85–98.
Schmukle, S. C. (2005). Unreliability of the Dot Probe Task. European Yang, Y., Read, S. J., & Miller, L. C. (2006). A taxonomy of situations
Journal of Personality, 19, 595– 605. from Chinese idioms. Journal of Research in Personality, 40, 750 –778.

(Appendix follows)
548 BACK, SCHMUKLE, AND EGLOFF

Appendix

Predicting Behavioral Aggregates From Explicit and Implicit Personality Measures: Convergent
and Discriminant Predictive Validities
Neuroticism Extraversion Openness Agreeableness Conscientiousness
Behavioral
aggregate Explicit Implicit Explicit Implicit Explicit Implicit Explicit Implicit Explicit Implicit
ⴱⴱⴱ ⴱⴱⴱ ⴱⴱ ⴱ ⴱ
Neuroticism .36 .35 –.26 –.19 –.06 .03 –.21 .10 –.01 –.03
Extraversion –.01 –.16† .38ⴱⴱⴱ .31ⴱⴱⴱ .17† .00 .14 –.07 –.04 –.09
Openness .03 –.12 .18ⴱ .18ⴱ .31ⴱⴱⴱ –.14 .05 –.19ⴱ –.17† –.16†
Agreeableness –.11 –.03 .13 –.01 .20ⴱ –.04 .35ⴱⴱⴱ .13 .16† –.15†
Conscientiousness –.01 .00 .01 –.11 .15 –.03 .09 .15 .30ⴱⴱⴱ –.08
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Note. Convergent validities are shown in bold.


p ⬍ .10. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Received July 30, 2008


Revision received January 26, 2009
Accepted February 19, 2009 䡲

You might also like