Canadian Psychology © 2009 Canadian Psychological Association

2009, Vol. 50, No. 3, 141–150 0708-5591/09/$12.00 DOI: 10.1037/a0013848

Ten Frequently Asked Questions About Implicit Measures and Their

Frequently Supposed, But Not Entirely Correct Answers

Bertram Gawronski
The University of Western Ontario

Self-report measures are often criticised for their susceptibility to self-presentation and their inability to
capture mental contents that are inaccessible to introspection. Over the past decade, researchers have
attempted to overcome these problems by means of implicit measures, which infer mental contents from
participants’ performance on experimental paradigms. In the present article I provide an overview of the
currently available implicit measures and discuss 10 common assumptions about these measures. I argue
that many of these assumptions are either inconsistent with the available evidence or theoretically
problematic for conceptual reasons. Nevertheless, implicit measures have proven their usefulness in
predicting behaviours that are difficult to predict with traditional self-report measures. Thus, even though
implicit measures may not be able to provide the information that is sometimes attributed to these
measures, they represent a valuable addition to the toolbox of psychological instruments in understanding
the determinants of human behaviour.

Keywords: associative processes, implicit measures, measurement, methods, validity

How pathetically scanty my self-knowledge is compared with, say, gists to search for alterative means to peek into people’s inner
my knowledge of my room. There is no such thing as observation of mental lives that do not rely on self-reports. One of the most
the inner world, as there is of the outer world. significant advances in this regard has been the development of
—Franz Kafka, The Third Notebook implicit measures. In contrast to traditional self-report measures,
implicit measures infer mental contents from participants’ perfor-
What Kafka alluded to in the above quote is the impossibility of mance on experimental paradigms, most often speeded categori-
directly observing the holy grail of psychology: the inner world of zation tasks. Over the past decade, implicit measures have inspired an
human beings. Psychological entities, such as attitudes, emotions, overwhelming amount of research across all disciplines of psychol-
beliefs, or motives, are not directly observable but have to be ogy, which clearly attests to the popularity of these measures.
inferred from people’s behaviours that are assumed to reflect these The main goal of the present article is to provide a brief
entities. However, Kafka’s concern is even more far-reaching, in overview of the currently available implicit measures and some
that he questioned our ability to observe our own inner mental life. frequently asked questions about these measures. The latter dis-
This suspicion has been echoed by psychologists in the second half cussion is inspired by the deplorable disconnect between basic
of the 20th century, who argued that we have no introspective research on the mechanisms underlying implicit measures and the
access to the psychological processes that guide our own behav- somewhat wider reception of research using these measures. Ad-
iour (e.g., Nisbett & Wilson, 1977). Instead, what is sometimes dressing 10 rather common assumptions about implicit measures,
regarded as the “privilege of self-knowledge” may be nothing else it is argued that many of these assumptions are either inconsistent
than naı̈ve theories about ourselves that are based on the same with the available evidence or theoretically problematic for con-
kinds of behavioural observations that form the basis of our ceptual reasons. In the remainder of the article, I briefly discuss a
knowledge about other individuals (Wilson & Dunn, 2004). more parsimonious interpretation of what implicit measures may
Needless to say, such introspective limits pose serious problems actually tell us, which has important implications for the under-
for the use of self-report measures in psychological research, and standing of human behaviour.
this problem seems even more far-reaching than the common
concern about self-presentational distortions (e.g., Crowne &
Marlowe, 1960; Paulhus, 1984). These issues have led psycholo- Implicit Measures

The history of implicit measures dates back to earlier attempts to

overcome the aforementioned problems of self-reports. One of the
Preparation of this article was supported by grants from the Canada most prominent examples in this regard is the Thematic Apper-
Research Chairs Program and the Social Sciences and Humanities Re-
ception Test (TAT), which was particularly designed to assess
search Council of Canada. I thank Rainer Banse and Wilhelm Hofmann for
helpful comments on an earlier version of this article.
implicit motives that are not accessible to introspection (for a
Correspondence concerning this article should be addressed to review, see McClelland, Koestner, & Weinberger, 1989). Another
Bertram Gawronski, Department of Psychology, The University of prominent precursor is the emotional Stroop task, which became a
Western Ontario, Social Science Centre, London, ON N6A 5C2, popular instrument in research on psychopathology (for a review,
Canada. E-mail: bgawrons@uwo.ca see Williams, Mathews, & MacLeod, 1996). Even though these

141
142 GAWRONSKI

measures can be regarded as pioneers in the area of implicit Sanbonmatsu, Powell, & Kardes, 1986). This task employs the
measurement (for earlier approaches, see Webb, Campbell, basic idea of sequential priming effects obtained in cognitive
Schwartz, & Sechrist, 1966), the type of implicit measures devel- psychology (Neely, 1977), using priming effects on evaluative
oped in the past decade differs from these precursors in one decisions as an indicator of automatic evaluation (for reviews,
important aspect. Whereas the TAT and the emotional Stroop task see Fazio, 2001; Klauer & Musch, 2003). In the standard para-
are both concerned with individual mental concepts (e.g., achieve- digm, participants are briefly presented with a prime stimulus (e.g.,
ment, anxiety), the more recently developed implicit measures are a Black face) that is immediately followed by a positive or nega-
concerned with assessing mental associations between concepts tive target word. Participants’ task is to quickly determine whether
(e.g., math-male; black-negative). The current review focuses pri- the target word is positive or negative by pressing one of two
marily on the latter type of measures, which have gained an response keys. To the degree that the prime stimulus leads to faster
enormous level of popularity across various areas of psychology in responses to positive words, the prime stimulus is assumed to be
the past decade. associated with positive valence. However, if the prime stimulus
facilitates responses to negative words, it is assumed to be asso-
Implicit Association Test ciated with negative valence (for scoring procedures, see
Wittenbrink, 2007). The evaluative priming task can be used to
One of the most prominent implicit measures is the Implicit assess evaluative responses to any type of object that can be
Association Test (IAT) developed by Greenwald, McGhee, and presented as a prime stimulus in a sequential priming paradigm,
Schwartz (1998). The standard version of the IAT includes two and it has been successfully used with supraliminal (e.g., Fazio et
binary categorization tasks that are combined in an association- al., 1995) and subliminal presentations of the primes (e.g., Olson,
congruent and an association-incongruent manner. For instance, & Fazio, 2002). Even though the standard variant of the task
the race IAT commonly used in research on racial prejudice typically employs evaluative decisions about positive and negative
involves a categorization of Black and White faces in terms of target words, procedural modifications that have been proposed
their race and a categorization of positive and negative words in include the pronunciation of positive and negative target words
terms of their valence. In a prejudice-congruent block, participants (Bargh, Chaiken, Raymond, & Hymes, 1996) and the naming of
are asked to respond to Black faces and negative words with one positive and negative pictures as target stimuli (Spruyt, Hermans,
key and to White faces and positive words with another key. De Houwer, Vandekerckhove, & Eelen, 2007).
Conversely, in the prejudice-incongruent block, participants are
asked to respond to Black faces and positive words with one Semantic Priming Task
key and to White faces and negative words with another key.
The rationale underlying the IAT is that quick and accurate A somewhat less common, but very similar paradigm is
responses in the task should be facilitated when the key assign- Wittenbrink, Judd, and Park’s (1997) semantic priming task. The
ments combine concepts that are strongly associated in mem- basic setup of this measure is almost identical to Fazio et al.’s
ory. However, quick and accurate responses should be inhibited (1995) evaluative priming task, the only difference being that
when the key assignment is association incongruent (for scoring participants are presented with meaningful and meaningless letter
procedures, see Greenwald, Nosek, & Banaji, 2003). The al- strings as target stimuli and that participants’ task is to determine
most infinite number of possible dimensions that can be used in as quickly as possible whether the letter string is a meaningful
the IAT makes the task quite flexible in its application, includ- word or a meaningless nonword. To the degree that the presenta-
ing prejudice (e.g., Gawronski, Peters, Brochu, & Strack, 2008), tion of a given prime stimulus facilitates quick responses to a
stereotyping (e.g., Gawronski, Ehrenberg, Banse, Zukova, & meaningful target word, the prime stimulus is assumed to be
Klauer, 2003), self-esteem (e.g., Greenwald & Farnham, 2000), associated with the semantic meaning of the target word (for
self-concepts (e.g., Asendorpf, Banse, & Mücke, 2002), brand scoring procedures, see Wittenbrink, 2007). For instance, in an
evaluations (e.g., Forehand & Perkins, 2005), phobic or anxiety- application of the task to racial stereotypes, Wittenbrink et al.
provoking stimuli (e.g., Teachman, Gregg, & Woody, 2001), al- found facilitated responses to trait words related to the stereotype
cohol (Wiers, Van Woerden, Smulders, & De Jong, 2002), and of African Americans (e.g., athletic, hostile) when participants
drugs (Wiers, Houben, & de Kraker, 2007), to name just a small were subliminally presented with the word black before the pre-
subset of earlier applications. More recently, a number of proce- sentation of the target words. Different to Fazio et al.’s evaluative
dural modifications have been proposed for the IAT, including IAT priming task, Wittenbrink et al.’s paradigm is primarily concerned
variants that make the task amenable for assessing associations of a with semantic rather than evaluative associations of a given object.
single concept rather than relative associations of two concepts
(Karpinski & Steinman, 2006) and variants that avoid blocked pre- Affect Misattribution Procedure
sentations of association-congruent and association-incongruent trials
by combining them in a single block (Teige-Mocigemba, Klauer, & A relatively recent, but already very popular measure is Payne,
Rothermund, 2008; see also, Rothermund, Teige-Mocigemba, Gast, Cheng, Govorun, and Stewart’s (2005) Affect Misattribution Pro-
& Wentura, in press). cedure (AMP). In this task, participants are briefly presented with
a prime stimulus, which is followed by a brief presentation of a
Evaluative Priming Task neutral Chinese character (see also Murphy & Zajonc, 1993). The
Chinese character is then replaced by a black-and-white pattern
Another prominent implicit measure is Fazio, Jackson, Dunton, mask, and participants’ task is to indicate whether they consider
and Williams’ (1995) evaluative priming task (see also, Fazio, the Chinese character as visually more pleasant or visually less
 SPECIAL ISSUE: IMPLICIT MEASURES 143

pleasant than the average Chinese character. The typical finding is related to the other pole of the attribute dimension (e.g., negative
that the neutral Chinese characters tend to be evaluated more words). In a second block, the classification of the particular
favourably when participants have been primed with a positive attribute poles as targets (“go”) and distracters (“no-go”) is re-
than when they have been primed with a negative stimulus. More versed. GNAT trials typically include a response deadline, such
interesting, even though this task may appear rather obvious and that participants are asked to show a “go” response to the targets
easy to control, priming effects in the AMP have been shown to be before the expiration of that deadline (e.g., 600 ms). The resulting
resistant against deliberate attempts to control the influence of the data are analysed by means of signal detection theory (Green &
prime stimuli even when participants were given explicit informa- Swets, 1966), such that differences in sensitivity scores (d⬘) be-
tion about how the prime stimuli may influence their responses in tween the two pairings of “go” trials (e.g., apples-positive vs.
the task (Payne, Cheng, et al., 2005). As with Fazio et al.’s (1995) apples-negative) are interpreted as an index of associations be-
evaluative priming task, the AMP can be used to assess evaluative tween the concept of interest and the respective attributes. Like the
responses toward any kind of stimuli that can be used as primes in IAT, the GNAT is relatively flexible in its application, in that
the task. targets and distracters may include a large variety of concepts and
attributes, including social groups (e.g., Nosek & Banaji, 2001),
Extrinsic Affective Simon Task self-related associations (e.g., Boldero, Rawlings, & Haslam,
2007), and anxiety-provoking or phobic stimuli (e.g., Teachman,
Another interesting, but less common implicit measure is De 2007).
Houwer’s (2003a) Extrinsic Affective Simon Task (EAST). In the
critical block of the task, participants are presented with target Approach-Avoidance Tasks
words (e.g., beer) that are presented in two different colors (e.g.,
yellow vs. blue) and positive and negative words that are presented A final class of implicit measures can be subsumed under the
in white ink colour. Participants are instructed to categorize the general label approach-avoidance tasks. The general assumption
presented words in terms of their valence when they are presented underlying these tasks is that positive stimuli facilitate approach
in white ink colour, and to categorize them in terms of their ink reactions and inhibit avoidance reactions, whereas negative stimuli
colour when they are coloured. For instance, in an EAST applica- facilitate avoidance reactions and inhibit approach reactions. In the
tion to evaluations of alcoholic beverages (e.g., De Jong, Wiers, first published demonstration of such effects, Solarz (1960) found
van de Braak, & Huijding, 2007), participants may be presented that participants were faster in pulling a lever toward them (ap-
with positive and negative words (e.g., spider, sunrise) presented proach) in response to positive compared to negative words. Con-
in white ink colour and with names of alcoholic and nonalcoholic versely, participants were faster in pushing a lever away from them
beverages (e.g., beer, juice) that are presented in yellow ink colour (avoidance) in response to negative compared to positive words.
on some trials and in blue ink colour on others. Participants’ task Expanding on these findings, Chen and Bargh (1999) showed that
is to press a left-hand key when they see a white word of negative these effects emerged regardless of whether approach-avoidance
valence or a word printed in blue ink colour and to press a responses were mapped with valence as the response-relevant
right-hand key when they see a white word of positive valence or stimulus feature (e.g., positive-approach, negative-avoidance vs.
a word printed in yellow ink colour. To the degree that participants negative-approach, positive-avoidance) or the particular type of
show faster (or more accurate) responses to a coloured word (e.g., responses were mapped with a valence-irrelevant feature (e.g.,
beer) when the required response to this word is combined with a approach to positive and negative words printed in yellow, avoid-
positive as compared to a negative response, it is inferred that ance to positive and negative words printed in blue). However, in
participants have positive associations with the object depicted by contrast to earlier interpretations of these effects as being due to
the coloured word. Even though the EAST originally has been direct, inflexible links between motivational orientations and par-
designed as a measure of evaluative responses, a number of recent ticular motor actions (e.g., Strack & Deutsch, 2004), accumulating
studies demonstrated its applicability to other domains, such as the evidence suggests that congruency effects in approach-avoidance
assessment of self-related associations (e.g., Teige, Schnabel, tasks depend on the positive or negative meaning (i.e., positive vs.
Banse, & Asendorpf, 2004; see also, Schmukle & Egloff, 2006). negative) that is assigned to a particular motor action (i.e., extensor
contraction vs. flexor contraction) in the description of the task.
Go/No-Go Association Task For instance, Eder and Rothermund (2008) found that participants
are faster in pulling a lever (flexor contraction) in response to
Nosek and Banaji’s (2001) Go/No-Go Association Task (GNAT) positive words and faster in pushing a lever (extensor contraction)
has been inspired by the basic logic of the IAT (Greenwald et al., in response to negative words when the required motor responses
1998) with an attempt to make the task suitable for an assessment of were described as pull (i.e., positive meaning attributed to flexor
absolute associations of a single concept rather than relative associa- contraction) and push (i.e., negative meaning attributed to extensor
tions of two concepts (see also, Karpinski & Steinman, 2006). In a contraction). However, these effects were reversed when the
nutshell, the GNAT uses a go/no-go task in which participants are same motor responses were described as upward (i.e., positive
asked to show a “go” response to different kinds of target stimuli meaning attributed to extensor contraction) and downward (i.e.,
(e.g., by pressing the space bar) and a “no-go” response to dis- negative meaning attributed to flexor contraction; for related
tracter stimuli (i.e., no button press). In one block of the task, the findings, see Markman & Brendl, 2005; Seibt, Neumann,
targets include stimuli related to the concept of interest (e.g., Nussinson, & Strack, 2008). Even though most studies have
apples) and stimuli related to one pole of a given attribute dimen- used variations of the abovementioned standard paradigm, de-
sion (e.g., positive words); the distracters typically include stimuli rivates of that paradigm include approach-avoidance tasks with
144 GAWRONSKI

left–right responses and visual depictions of their respective Question 2: Do Implicit Measures Overcome the Problem
meanings (Brendl, Markman, & Messner, 2005) and variants in of Social Desirability?
which motor movements are used to assess self-related associ-
ations (Schnabel, Banse, & Asendorpf, 2006). Another widespread assumption about implicit measures is that
they overcome the problem of social desirability that has plagued
the developers of self-report measures (e.g., Fazio et al., 1995).
Frequently Asked Questions About Implicit Measures This speculation is based on the notion that it is much easier to
adjust one’s responses on self-report measures compared to im-
Implicit measures have been used in almost all subdisciplines of plicit measures (e.g., Banse, Seise, & Zerbes, 2001; Egloff &
psychology. However, interpretations of what these measures can Schmukle, 2002; Kim, 2003; Schnabel et al., 2006; Steffens,
tell us sometimes go far beyond the available data. In the following 2004). However, empirical research suggests that the situation is a
sections, I discuss 10 questions and their frequently assumed, but little more complex. In a nutshell, the social desirability hypothesis
not entirely correct, answers on what kind of information we can implies that correlations between implicit measures and self-
infer from implicit measures. Even though the conclusions drawn reports should be high when social desirability is low, whereas
in this discussion are to a large degree shared by researchers that correlations between the two measures should be significantly
have investigated the mechanisms underlying implicit measures reduced when social desirability is high (see Gawronski et al.,
(for a review, see Wittenbrink & Schwarz, 2007), misconceptions 2007). Even though some studies found empirical support for this
about their meaning are still relatively common in the wider prediction (e.g., Nosek, 2005; Riketta, 2006), other studies failed
discourse about implicit measures. to find any effect of social desirability (e.g., Egloff & Schmukle,
2003; Hofmann, Gawronski, et al., 2005) or even found the oppo-
Question 1: Do Implicit Measures Provide a Window to site pattern (e.g., Hofmann, Gawronski, et al., 2005; Riketta,
the Unconscious? 2006). Although the picture seems somewhat more straightforward
for specific motivational influences such as motivation to control
A common assumption about implicit measures is that these prejudiced reactions (see Gawronski et al., 2007), correlations
measures assess unconscious mental associations that are not ac- between self-reports and implicit measures have been shown to
cessible to conscious introspection (e.g., Banaji, Lemm, & vary as a function of several nonmotivational, cognitive factors,
Carpenter, 2001). This assumption is based on the methodological suggesting that the correspondence between implicit measures and
fact that implicit measures do not presuppose conscious introspec- self-reports is far more complex than just a matter of social
tion, as it is the case for traditional self-report measures. However, desirability and self-presentation (for a review, see Hofmann,
from a logical point of view, this does not necessarily imply that Gschwendner, et al., 2005).
the associations assessed by implicit measures are indeed uncon-
scious. The latter is an empirical question that needs to be tested as Question 3: Are Implicit Measures Immune to Faking?
such (De Houwer, 2006; Gawronski & Bodenhausen, 2007). In
fact, recent reviews indicate that there is quite some evidence to Another assumption, which is directly related to the last one, is
the contrary (e.g., Gawronski, Hofmann, & Wilbur, 2006; that implicit measures are immune to faking. Obviously, the par-
Gawronski, LeBel, & Peters, 2007). For instance, one finding that ticular nature of implicit measures make them far less susceptible
is commonly interpreted as evidence for the unconscious nature of to faking than traditional self-report measures (e.g., Banse et al.,
the associations assessed by implicit measures is their weak cor- 2001; Egloff & Schmukle, 2002; Kim, 2003; Schnabel et al., 2006;
respondence to self-report measures (for a meta-analysis, see Steffens, 2004). However, empirical evidence suggests that they
Hofmann, Gawronski, Gschwendner, Le, & Schmitt, 2005). How- are not entirely immune to faking, such that deliberate attempts to
ever, low correlations to self-report measures can be due to a control one’s responses have been shown to alter the scores re-
number of reasons other than lack of introspective access, includ- vealed by some implicit measures (e.g., Degner, in press; Fiedler
ing measurement error (e.g., Cunningham, Preacher, & Banaji, & Blümke, 2005; Klauer & Teige-Mocigemba, 2007; Lowery,
2001), lack of conceptual correspondence (e.g., Hofmann, Hardin, & Sinclair, 2001; Schnabel et al., 2006; Steffens, 2004;
Gawronski, et al., 2005; see also Payne, Burkley, & Stokes, 2008), Teige-Mocigemba & Klauer, in press; but see Payne, Cheng, et al.,
self-presentational concerns (e.g., Nosek, 2005), and many other 2005). Typically, these effects are much smaller compared to those
factors (for a review, see Hofmann, Gschwendner, Nosek, & on self-report measures. However, to say that implicit measures
Schmitt, 2005). Moreover, recent research has shown that corre- are “immune” to faking seems inaccurate in the light of these
lations between implicit measures and self-reported attitudes in- findings.
crease if participants are instructed to introspect on their feelings
toward the attitude object (e.g., Gawronski & LeBel, 2008; see Question 4: Can Implicit Measures Be Used
also, C. T. Smith & Nosek, 2007). These findings are quite as a Lie Detector?
difficult to explain by the unconsciousness account. If the associ-
ations assessed by implicit measures are indeed unconscious, it is Based on their lower susceptibility to faking, another common
certainly possible that introspection instructions influence the assumption about implicit measures is that they can be used as
mean values of self-reported attitudes. However, an increase in some kind of lie detector when a person is not willing to reveal a
correlations between the two measures seems rather surprising particular opinion or belief (e.g., Nier, 2005; Sartori, Agosta,
from an unconsciousness point of view (for a more elaborate Zogmaister, Ferrara, & Castiello, 2008). Needless to say, this
discussion, see Gawronski et al., 2006). assumption seems problematic in the light of the abovementioned
 SPECIAL ISSUE: IMPLICIT MEASURES 145

research, showing that deliberate attempts to fake responses can contingent on the subjectively preferred conceptualization of the
alter the scores revealed by implicit measures (e.g., Degner, in “true self.”
press; Fiedler & Blümke, 2005; Klauer & Teige-Mocigemba,
2007; Lowery et al., 2001; Schnabel et al., 2006; Steffens, 2004; Question 6: Do Implicit Measures Reflect Early
Teige-Mocigemba & Klauer, 2008). Such effects imply the poten- Socialization Experiences?
tial for omission errors such that implicit measures may fail to
detect the sought-after “true response” (see Fiedler, Schmid, & Addressing potential sources of the associations assessed by
Stahl, 2002). In addition, it is important to note that there can be implicit measures, there has been a strong intuition that these
ambiguity regarding the sources of the mental associations as- measures may tap early socialization experiences (e.g., Greenwald
sessed by implicit measures. For instance, a number of studies & Banaji, 1995; Rudman, 2004). In line with this assumption,
tested whether implicit measures of child-sex associations can there have been a number of studies that successfully related the
successfully identify convicted child molesters (e.g., Gray, Brown, associations assessed by implicit measures to different types of
MacCulloch, Smith, & Snowden, 2005; Nunes, Firestones, & early childhood experiences. Such findings have been reported for
Baldwin, 2007; see also, Kamphuis, de Ruiter, Janssen, Spiering, attitudes toward smoking (Rudman, Phelan, & Heppen, 2007),
2005). Results showed that implicit measures were indeed quite attitudes toward overweight (Rudman et al., 2007), racial prejudice
successful in discriminating between pedophiles and nonpedo- (Sinclair, Dunn, & Lowery, 2005), and self-esteem (DeHart, Pel-
philes. However, it is important to note that child-sex associations ham, & Tennen, 2006). However, qualifying the generality of the
may have their roots in a number of factors other than pedophilia, long-term socialization hypothesis, there are several studies that
for instance when a person has been the target of sexual abuse as successfully induced variations in implicit measures with rather
child (see Gray et al., 2005). Such cases imply the risk of false simple manipulations in the lab (e.g., Gawronski, Walther, &
alarms, which also challenge the suitability of implicit measures as Blank, 2005; Gregg, Seibt, & Banaji, 2006). Thus, even though
a lie detector (see Fiedler et al., 2002). For these reasons, any early socialization experiences may indeed be one potential source
claims that implicit measures could be used as some kind of lie of the associations assessed by implicit measures, these associa-
detector should be treated with caution. tions can have their roots in a variety of factors other than early
socialization experiences. Accordingly, interpreting observed vari-
ations in implicit measures as directly reflecting early socialization
Question 5: Do Implicit Measures Reflect the experiences implies a logical error, such that an empirically accu-
“True Self”? rate “A implies B” relation (i.e., variations in early socialization
experiences imply variations in implicit measures) is used to infer
The questions pertaining to social desirability, faking, and lie the reciprocal, in this case incorrect, relation “B implies A” (i.e.,
detection all have the connotation that implicit measures may variations in implicit measures imply variations in early socializa-
somehow reveal a person’s true attitudes, opinions, or beliefs, tion experiences).
which this person may not be willing to disclose in a self-report
measure (e.g., Fazio et al., 1995; Olson & Fazio, 2004). However,
Question 7: Do Implicit Measures Reflect Highly
the abovementioned issues already indicated that such a claim is
Robust Associations?
not without problems. In addition, it is important to note that there
are two possible interpretations of what could be regarded as a A widespread assumption that is directly related to the last one
person’s “true self,” and those interpretations directly contradict is that implicit measures reflect mental associations that, once
each other (Gawronski, Peters, & LeBel, 2008). The first interpre- formed, are highly robust and stable over time (e.g., Wilson,
tation implies that the “true self” is revealed when intentional Lindsey, & Schooler, 2000). This assumption is consistent with
control over one’s responses fails. This interpretation, which is in research showing that self-reported evaluations are sometimes
line with the notion implied by the abovementioned questions, more susceptible to attitude change manipulations compared to
suggests that any disruption of people’s success at controlling their evaluations assessed by implicit measures (e.g., Gawronski &
behaviour (e.g., due to alcohol intoxication) reveals the true nature Strack, 2004; Gregg et al., 2006; Rydell & McConnell, 2006).
of a person (see Payne, Jacoby, & Lambert, 2005). The second However, there are also several studies that found exactly the
possible interpretation stands directly in contrast to the first one, in opposite. In these studies, implicit measures showed evidence for
that the “true self” is reflected in what a person consciously intends change, whereas explicit measures did not (e.g., Gawronski &
to do or say. From this perspective, any unintentional act is LeBel, 2008; Gibson, 2008; Karpinski & Hilton, 2001; Olson &
attributed to forces outside the person that may not be under Fazio, 2006). Moreover, a number of studies have shown that
control of the person’s “true self” (see Arkes & Tetlock, 2004). implicit measures can be highly context-sensitive (e.g., Dasgupta
Thus, whereas the first interpretation equates the “true self” with & Greenwald, 2001; Lowery et al., 2001; Rydell & Gawronski, in
unintentional behaviour, the second one equates the “true self” press; Wittenbrink, Judd, & Park, 2001), which further challenges
with intentional behaviour (Gawronski et al., 2008). As these the assumption that implicit measures reflect highly robust asso-
interpretations are a matter of definition and cultural worldviews ciations. Based on a comprehensive review of the available evi-
rather than empirical observation, any claims about the “true self” dence, Gawronski and Bodenhausen (2006) argued that the par-
can be regarded as a matter of subjective preference for one or the ticular pattern of contextually induced changes on self-report and
other view. Thus, even though responses on implicit measures implicit measures depends on (a) the type of process that is
clearly fall into the unintentional spectrum of possible behaviours, affected by a given manipulation, the activation of associations in
any depiction of implicit measures as revealing the “true self” are memory, or the propositional validation of activated information;
146 GAWRONSKI

and (b) whether changes in one process lead to indirect changes in Fazio, 2003; Teige et al., 2004). The only two measures that have
the other one. This interplay of associative processes (reflected in consistently shown reliability estimates that are acceptable from a
implicit measures) and propositional processes (reflected in self- psychometric point of view (e.g., Cronbach’s alpha values in the
report measures) can produce a variety of predictable outcomes, range of .80) are the IAT (Greenwald et al., 1998) and the AMP
including changes in implicit but not self-report measures (e.g., (Payne, Cheng, et al., 2005). Another problem in this context is
Gawronski & LeBel, 2008; Olson & Fazio, 2006), changes in that it is still not common practise to report reliability estimates for
self-report but not implicit measures (e.g., Gawronski & Strack, implicit measures, which makes the evaluation of a given research
2004; Gregg et al., 2006), corresponding changes in self-report and finding rather difficult. Adherence to the conventions typically
implicit measures (e.g., Gawronski & LeBel, 2008; Olson & Fazio, employed for other measures would certainly be helpful in this
2001), and even opposite effects on self-report and implicit mea- regard.
sures (e.g., Rydell, McConnell, Mackie, & Strain, 2006). In addi-
tion, Gawronski and Bodenhausen’s (2006) conceptualization Question 10: Are All Implicit Measures Created Equal?
specifies the conditions under which implicit measures do versus
do not vary as a function of the context, which have been empir- A final assumption is that implicit measures are generally in-
ically confirmed in a recent study by Rydell and Gawronski (in terchangeable, such that all implicit measures will produce the
press). same outcome, at least as long as these measures employ the same
materials (e.g., Sherman, Rose, Koch, Presson, & Chassin, 2003).
Question 8: Do Implicit Measures Simply Reflect This assumption is incorrect for at least two reasons. First, implicit
measures differ to the extent to which they assess category-related
Cultural Associations?
or exemplar-related associations (Olson & Fazio, 2003). For ex-
One concern that has been raised about implicit measures is that ample, the standard variant of Fazio et al.’s (1995) evaluative
these measures may simply reflect cultural or extrapersonal asso- priming task does not require participants to process the category
ciations that are not necessarily endorsed by the individual. This membership of a given prime stimulus (e.g., to explicitly catego-
argument has been made at different levels of generality, ranging rize a black face as Black), making the task more amenable to
from criticisms of particular kinds of implicit measures (e.g., idiosyncratic features of the prime rather than the category of that
Olson & Fazio, 2004) to implicit measures in general (e.g., Arkes prime (e.g., Livingston & Brewer, 2002; Olson & Fazio, 2003).
& Tetlock, 2004). However, in evaluating this criticism, there are This situation is different in the IAT (Greenwald et al., 1998),
a number of issues that need to be considered. First, implicit which explicitly requires participants to categorize the presented
measures— by definition— do not assess the endorsement of eval- stimuli in terms of their category membership (e.g., Olson &
uations or beliefs (De Houwer, 2006). Instead, these measures Fazio, 2003). Thus, the findings revealed by the two measures may
simply assess mental associations that may or may not be explic- sometimes diverge, due to the different associations they assess
itly endorsed. As such, the objection that implicit measures assess (i.e., exemplar related vs. category related). Second, different
associations that are not necessarily endorsed by the individual is implicit measures employ different kinds of mechanisms to assess
true in a trivial sense. Second, the claim that only some of these mental associations (De Houwer, 2003b). Thusly, to the degree
associations are inherently personal whereas others are inherently that a given manipulation interacts with these mechanisms,
extrapersonal (e.g., Olson & Fazio, 2004) implies that the repre- implicit measures that are based on different mechanisms may
sentation of associations in memory can differ as a function of show different effects of the same manipulation (Gawronski,
their source (Gawronski & Bodenhausen, 2006). Such a claim not Deutsch, LeBel, & Peters, 2008). Even though research on this
only presupposes a theory of mental representation that allows for question is still scarce, there is evidence that otherwise identical
such differences; it also requires a precise and unambiguous def- implicit measures that are based on different mechanisms can
inition of the terms personal and extrapersonal, as applied to show opposite effects of the same experimental manipulation (e.g.,
mental associations. However, neither of the two requirements has Deutsch & Gawronski, 2009; Gawronski & Bodenhausen, 2005).
been met so far (for a discussion, see Gawronski et al., 2008), As such antagonistic effects may often be due to secondary vari-
which makes conceptual claims about the personal versus extrap- ations in the mechanism underlying these measures rather than
ersonal nature of the associations assessed by implicit measures genuine variations in the to-be-assessed construct, replications
problematic. with multiple-different measures would be helpful to establish the
precise nature of a given effect.
Question 9: How Reliable Are Implicit Measures?
What Can We Learn From Implicit Measures?
Moving from conceptual to methodological questions, a com-
mon question from psychometricians concerns the reliability of Given the problems with the reviewed assumptions, one may
implicit measures. In the context of the first question, I already naturally ask the question: What can we learn from implicit mea-
have pointed out that low correlations between implicit measures sures? Drawing on generalised dual-process models of human
and traditional self-reports can be due to measurement error. information processing, several theorists argued that implicit mea-
Unfortunately, large proportions of measurement error seem to be sures provide a proxy for the activation of associations in memory
a critical issue for quite a number of implicit measures, showing regardless of whether these associations are regarded as accurate
internal consistency scores that are clearly unsatisfactory from a or inaccurate. Traditional self-report measures, in contrast, are
psychometric point of view (e.g., Banse, 1999, 2001; Bosson, assumed to reflect the outcome of a propositional validation pro-
Swann, & Pennebaker, 2000; Cunningham et al., 2001; Olson & cess, which aims at assessing the (subjective) validity of these
 SPECIAL ISSUE: IMPLICIT MEASURES 147

associations (e.g., Gawronski & Bodenhausen, 2006; Strack & Résumé

Deutsch, 2004; see also Beevers, 2005; Lieberman, Gaunt, Gilbert,
& Trope, 2002; Sloman, 1996; E. R. Smith & DeCoster, 2000). Les mesures autorapportées sont souvent critiquées pour leur sus-
Even though implicit measures do not provide an entirely pure ceptibilité à la désirabilité sociale et leur incapacité à capturer les
reflection of activated associations (Conrey, Sherman, Gawronski, contenus mentaux inaccessibles à l’introspection. Au cours de la
Hugenberg, & Groom, 2005), this rather simple conceptual dis- dernière décennie, les chercheurs ont tenté de remédier à ces
tinction between associative and propositional processes has been problèmes avec des mesures implicites, qui permettent d’inférer les
applied successfully to the determinants of variations in self-report contenus mentaux à partir de la performance des participants en
and implicit measures (Gawronski & Bodenhausen, 2006) as well contexte expérimental. Dans le présent article, je propose un survol
as the relation of self-report and implicit measures to overt behav- des mesures implicites actuellement disponibles et je discute de 10
iour (Strack & Deutsch, 2004). postulats communs à propos de ces mesures. J’avance que plu-
The most notable finding in this context is that activated asso- sieurs de ces postulats sont soit inconsistants avec les données
ciations assessed by implicit measures can influence overt behav- disponibles, soit problématiques sur le plan théorique pour des
iour, even when these associations are rejected as invalid in stan- raisons conceptuelles. Néanmoins, les mesures implicites ont
dard self-report measures. This notion is most prominently prouvé leur utilité pour prédire les comportements difficiles à
reflected in research showing that implicit measures predict spon- prédire avec les mesures autorapportées traditionnelles. Ainsi,
même si les mesures autorapportées ne permettent pas d’obtenir
taneous behaviours that are difficult to predict with standard self-
certaines informations qui leurs sont parfois attribuées, elles con-
report measures (e.g., Asendorpf et al., 2002; Egloff & Schmukle,
stituent un ajout valable à l’éventail des instruments de mesure
2002; Fazio et al., 1995; McConnell & Leibold, 2001). In line with
psychologiques visant à comprendre les déterminants du compor-
these findings, implicit measures have also been shown to be
tement humain.
superior over self-report measures in predicting behaviour un-
der conditions of reduced cognitive capacity (e.g., Hofmann, Mots-clés : processus associatifs, mesures implicites, instruments
Gschwendner, Castelli, & Schmitt, 2008; Hofmann, Rauch, & de mesure, méthodes, validité
Gawronski, 2007). Other studies have shown that implicit mea-
sures sometimes explain behavioural variation over and above the
variation explained by self-report measures (e.g., Perugini, 2005), References
which provides further evidence for the impact of activated asso-
Arkes, H. R., & Tetlock, P. E. (2004). Attributions of implicit prejudice, or
ciations on behaviour. Moreover, research adopting an individual
“Would Jesse Jackson “fail” the Implicit Association Test? Psycholog-
difference approach suggests that implicit measures may be better ical Inquiry, 15, 257–278.
predictors of behaviour for people preferring intuitive over ana- Asendorpf, J. B., Banse, R., & Mücke, D. (2002). Double dissociation
lytical processing styles (e.g., Conner, Perugini, O’Gorman, Ayres, between explicit and implicit personality self-concept: The case of shy
& Prestwich, 2007) and for individuals with low working memory behavior. Journal of Personality and Social Psychology, 83, 380 –393.
capacity (e.g., Hofmann, Gschwendner, Friese, Wiers, & Schmitt, Banaji, M. R., Lemm, K. M., & Carpenter, S. J. (2001). The social
2008). In addition, activated associations assessed by implicit unconscious. In A. Tesser & N. Schwarz (Eds.), Blackwell handbook of
measures have been shown to bias the processing of ambiguous social psychology: Intraindividual processes (pp. 134 –158). Oxford,
England: Blackwell.
information (e.g., Hugenberg & Bodenhausen, 2003, 2004), and
Banse, R. (1999). Automatic evaluation of self and significant others:
these biasing influences seem to occur outside of conscious aware-
Affective priming in close relationships. Journal of Social and Personal
ness (e.g., Gawronski, Geschke, & Banse, 2003; see also, Galdi, Relationships, 16, 803– 821.
Arcuri, & Gawronski, 2008). Finally, there is an accumulating Banse, R. (2001). Affective priming with liked and disliked persons: Prime
body of research showing that discrepancies between activated visibility determines congruency and incongruency effects. Cognition &
associations assessed by implicit measures and explicitly endorsed Emotion, 15, 501–520.
beliefs assessed with self-report measures can produce unique Banse, R., Seise, J., & Zerbes, N. (2001). Implicit attitudes towards
psychological states that promote behaviours aimed at reducing homosexuality: Reliability, validity, and controllability of the IAT.
these discrepancies. For instance, Jordan, Spencer, Zanna, Zeitschrift für Experimentelle Psychologie, 48, 145–160.
Bargh, J. A., Chaiken, S., Raymond, P., & Hymes, C. (1996). The auto-
Hoshino-Browne, and Correll (2003) found that individuals who
matic evaluation effect: Unconditional automatic activation with a pro-
displayed a particular pattern of self-esteem discrepancies in self- nunciation task. Journal of Personality and Social Psychology, 32,
report and implicit measures showed enhanced levels of defensive 104 –128.
behaviour. In a similar vein, Petty, Tormala, Briñol, and Jarvis Beevers, C. G. (2005). Cognitive vulnerability to depression: A dual
(2006) showed that individuals who displayed attitudinal discrep- process model. Clinical Psychology Review, 25, 975–1002.
ancies in self-report and implicit measures were more likely to Boldero, J. M., Rawlings, D., & Haslam, N. (2007). Convergence between
engage in elaborate processing of attitude-relevant information GNAT-assessed implicit and explicit personality. European Journal of
(see also Briñol, Petty, & Wheeler, 2006). Taken together, these Personality, 21, 341–358.
Bosson, J. K., Swann, W. B., & Pennebaker, J. W. (2000). Stalking the
findings indicate that even though implicit measures may not
perfect measure of implicit self-esteem: The blind men and the elephant
provide the type of information that is sometimes attributed to
revisited? Journal of Personality and Social Psychology, 79, 631– 643.
these measures, they represent a valuable addition to the toolbox of Brendl, C. M., Markman, A. B., & Messner, C. (2005). Indirectly measur-
psychological instruments to peek into people’s inner mental life, ing evaluations of several attitude objects in relation to a neutral refer-
which is essential in understanding the determinants of human ence point. Journal of Experimental Social Psychology, 41, 346 –368.
behaviour. Briñol, P., Petty, R. E., & Wheeler, S. C. (2006). Discrepancies between
148 GAWRONSKI

explicit and implicit self-concepts: Consequences for information pro- On the automatic activation of attitudes. Journal of Personality and
cessing. Journal of Personality and Social Psychology, 91, 154 –170. Social Psychology, 50, 229 –238.
Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation: Fiedler, K., & Blümke, M. (2005). Faking the IAT: Aided and unaided
Immediate behavioral predispositions to approach or avoid the stimulus. response control on the Implicit Association Tests. Basic and Applied
Personality and Social Psychology Bulletin, 25, 215–224. Social Psychology, 27, 307–316.
Conner, M., Perugini, M., O’Gorman, R., Ayres, K., & Prestwich, K. Fiedler, K., Schmid, J., & Stahl, T. (2002). What is the current truth about
(2007). Relations between implicit and explicit measures of attitudes and polygraph lie detection? Basic and Applied Social Psychology, 24,
measures of behavior: Evidence of moderation by individual difference 313–324.
variables. Personality and Social Psychology Bulletin, 33, 1727–1740. Forehand, M. R., & Perkins, A. (2005). Implicit assimilation and explicit
Conrey, F. R., Sherman, J. W., Gawronski, B., Hugenberg, K., & Groom, contrast: A set/reset model of response to celebrity voice-overs. Journal
C. (2005). Separating multiple processes in implicit social cognition: of Consumer Research, 32, 435– 441.
The Quad-Model of implicit task performance. Journal of Personality Galdi, S., Arcuri, L., & Gawronski, B. (2008). Automatic mental associ-
and Social Psychology, 89, 469 – 487. ations predict future choices of undecided decision makers. Science, 321,
Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability 1100 –1102.
independent of psychopathology. Journal of Consulting Psychology, 24, Gawronski, B., & Bodenhausen, G. V. (2005). Accessibility effects on
349 –354. implicit social cognition: The role of knowledge activation versus re-
Cunningham, W. A., Preacher, K. J., & Banaji, M. R. (2001). Implicit trieval experiences. Journal of Personality and Social Psychology, 89,
attitude measurement: Consistency, stability, and convergent validity. 672– 685.
Psychological Science, 12, 163–170. Gawronski, B., & Bodenhausen, G. V. (2006). Associative and proposi-
Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic tional processes in evaluation: An integrative review of implicit and
attitudes: Combating automatic prejudice with images of admired and explicit attitude change. Psychological Bulletin, 132, 692–731.
disliked individuals. Journal of Personality and Social Psychology, 81, Gawronski, B., & Bodenhausen, G. V. (2007). What do we know about
800 – 814. implicit attitude measures and what do we have to learn? In
Degner, J. (in press). On the (un)controllability of affective priming: B. Wittenbrink & N. Schwarz (Eds.), Implicit measures of attitudes (pp.
265–286). New York: Guilford.
Strategic manipulation is feasible but can possibly be prevented. Cog-
Gawronski, B., Deutsch, R., LeBel, E. P., & Peters, K. R. (2008). Response
nition and Emotion.
interference as a mechanism underlying implicit measures: Some traps
DeHart, T., Pelham, B. W., & Tennen, H. (2006). What lies beneath:
and gaps in the assessment of mental associations with experimental
Parenting style and implicit self-esteem. Journal of Experimental Social
paradigms. European Journal of Psychological Assessment, 24, 218 –
Psychology, 42, 1–17.
225.
De Houwer, J. (2003a). The extrinsic affective Simon task. Experimental
Gawronski, B., Ehrenberg, K., Banse, R., Zukova, J., & Klauer, K. C.
Psychology, 50, 77– 85.
(2003). It’s in the mind of the beholder: The impact of stereotypic
De Houwer, J. (2003b). A structural analysis of indirect measures of
associations on category-based and individuating impression formation.
attitudes. In J. Musch & K. C. Klauer (Eds.), The psychology of evalu-
Journal of Experimental Social Psychology, 39, 16 –30.
ation: Affective processes in cognition and emotion (pp. 219 –244).
Gawronski, B., Geschke, D., & Banse, R. (2003). Implicit bias in impres-
Mahwah, NJ: Erlbaum.
sion formation: Associations influence the construal of individuating
De Houwer, J. (2006). What are implicit measures and why are we using
information. European Journal of Social Psychology, 33, 573–589.
them? In R. W. Wiers & A. W. Stacy (Eds.), The handbook of implicit
Gawronski, B., Hofmann, W., & Wilbur, C. J. (2006). Are “implicit”
cognition and addiction (pp. 11–28). Thousand Oaks, CA: Sage.
attitudes unconscious? Consciousness and Cognition, 15, 485– 499.
De Jong, P. J., Wiers, R. W., van de Braak, M., & Huijding, J. (2007). Gawronski, B., & LeBel, E. P. (2008). Understanding patterns of attitude
Using the Extrinsic Affective Simon Test as a measure of implicit change: When implicit measures show change, but explicit measures do
attitudes towards alcohol: Relationship with drinking behavior and al- not. Journal of Experimental Social Psychology, 44, 1355–1361.
cohol problems. Addictive Behaviors, 32, 881– 887. Gawronski, B., LeBel, E. P., & Peters, K. R. (2007). What do implicit
Deutsch, R., & Gawronski, B. (2009). When the method makes a differ- measures tell us? Scrutinizing the validity of three common assumptions.
ence: Antagonistic effects on “automatic evaluations” as a function of Perspectives on Psychological Science, 2, 181–193.
task characteristics of the measure. Journal of Experimental Social Gawronski, B., Peters, K. R., Brochu, P. M., & Strack, F. (2008). Under-
Psychology, 45, 101–114. standing the relations between different forms of racial prejudice: A
Eder, A. B., & Rothermund, K. (2008). When do motor behaviors cognitive consistency perspective. Personality and Social Psychology
(mis)match affective stimuli? An evaluative coding view of approach Bulletin, 34, 648 – 665.
and avoidance reactions. Journal of Experimental Psychology: Gen- Gawronski, B., Peters, K. R., & LeBel, E. P. (2008). What makes mental
eral, 137, 262–281. associations personal or extra-personal? Conceptual issues in the meth-
Egloff, B., & Schmukle, S. C. (2002). Predictive validity of an Implicit odological debate about implicit attitude measures. Social and Person-
Association Test for assessing anxiety. Journal of Personality and Social ality Psychology Compass, 2, 1002–1023.
Psychology, 83, 1441–1455. Gawronski, B., & Strack, F. (2004). On the propositional nature of cog-
Egloff, B., & Schmukle, S. C. (2003). Does social desirability moderate the nitive consistency: Dissonance changes explicit, but not implicit atti-
relationship between implicit and explicit anxiety measures? Personality tudes. Journal of Experimental Social Psychology, 40, 535–542.
and Individual Differences, 35, 1697–1706. Gawronski, B., Walther, E., & Blank, H. (2005). Cognitive consistency and
Fazio, R. H. (2001). On the automatic activation of associated evaluations: the formation of interpersonal attitudes: Cognitive balance affects the
An overview. Cognition and Emotion, 15, 115–141. encoding of social information. Journal of Experimental Social Psychol-
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). ogy, 41, 618 – 626.
Variability in automatic activation as an unobtrusive measure of racial Gibson, B. (2008). Can evaluative conditioning change attitudes toward
attitudes: A bona fide pipeline? Journal of Personality and Social mature brands? New evidence from the Implicit Association Test. Jour-
Psychology, 69, 1013–1027. nal of Consumer Research, 35, 178 –188.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). Gray, N. S., Brown, A. S., MacCulloch, M. J., Smith, J., & Snowden, R. J.
 SPECIAL ISSUE: IMPLICIT MEASURES 149

(2005). An implicit test of the associations between children and sex in Klauer, K. C., & Teige-Mocigemba, S. (2007). Controllability and resource
pedophiles. Journal of Abnormal Psychology, 114, 304 –308. dependence in automatic evaluation. Journal of Experimental Social
Green, D. M., & Swets, J. A. (1966). Signal detection theory and psycho- Psychology, 43, 648 – 655.
physics. New York: Wiley. Lieberman, M. D., Gaunt, R., Gilbert, D. T., & Trope, Y. (2002). Reflec-
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: tion and reflexion: A social cognitive neuroscience approach to attribu-
Attitudes, self-esteem, and stereotypes. Psychological Review, 102, tional inference. In M. P. Zanna (Ed.), Advances in experimental social
4 –27. psychology (Vol. 34, pp. 199 –249). New York: Academic.
Greenwald, A. G., & Farnham, S. D. (2000). Using the Implicit Associa- Livingston, R. W., & Brewer, M. B. (2002). What are we really priming?
tion Test to measure self-esteem and self-concept. Journal of Personality Cue-based versus category-based processing of facial stimuli. Journal of
and Social Psychology, 79, 1022–1038. Personality and Social Psychology, 82, 5–18.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L. (1998). Measuring Lowery, B. S., Hardin, C. D., & Sinclair, S. (2001). Social influence effects
individual differences in implicit cognition: The Implicit Association on automatic racial prejudice. Journal of Personality and Social Psy-
Test. Journal of Personality and Social Psychology, 74, 1464 –1480. chology, 81, 842– 855.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding Markman, A. B., & Brendl, C. M. (2005). Constraining theories of em-
and using the Implicit Association Test: I. An improved scoring algo- bodied cognition. Psychological Science, 16, 6 –10.
rithm. Journal of Personality and Social Psychology, 85, 197–216. McClelland, D. C., Koestner, R., & Weinberger, J. (1989). How do
Gregg, A. P., Seibt, B., & Banaji, M. R. (2006). Easier done than undone: self-attributed and implicit motives differ? Psychological Review, 96,
Asymmetry in the malleability of implicit preferences. Journal of Per- 690 –702.
sonality and Social Psychology, 90, 1–20. McConnell, A. R., & Leibold, J. M. (2001). Relations among the Implicit
Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. Association Test, discriminatory behavior, and explicit measures of
(2005). A meta-analysis on the correlation between the Implicit Asso- racial attitudes. Journal of Experimental Social Psychology, 37, 435–
ciation Test and explicit self-report measure. Personality and Social 442.
Psychology Bulletin, 31, 1369 –1385. Murphy, S. T., & Zajonc, R. B. (1993). Affect, cognition, and awareness:
Hofmann, W., Gschwendner, T., Castelli, L., & Schmitt, M. (2008). Affective priming with optimal and suboptimal stimulus exposure. Jour-
Implicit and explicit attitudes and interracial interaction: The moderating nal of Personality and Social Psychology, 64, 723–739.
role of situationally available control resources. Group Processes and Neely, J. H. (1977). Semantic priming and retrieval from lexical memory:
Intergroup Relations, 11, 69 – 87. Roles of inhibitionless spreading activation and limited-capacity atten-
Hofmann, W., Gschwendner, T., Friese, M., Wiers, R., & Schmitt, M. tion. Journal of Experimental Psychology: General, 106, 226 –254.
(2008). Working memory capacity and self-regulatory behavior: Toward Nier, J. A. (2005). How dissociated are implicit and explicit racial atti-
an individual differences perspective on behavior determination by tudes? A bogus pipeline approach. Group Processes and Intergroup
automatic versus controlled processes. Journal of Personality and Social Relations, 8, 39 –52.
Psychology, 95, 962–977. Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know:
Hofmann, W., Gschwendner, T., Nosek, B. A., & Schmitt, M. (2005). Verbal reports on mental processes. Psychological Review, 84, 231–259.
What moderates explicit-implicit consistency? European Review of So- Nosek, B. A. (2005). Moderators of the relationship between implicit and
cial Psychology, 16, 335–390. explicit evaluation. Journal of Experimental Psychology: General, 134,
Hofmann, W., Rauch, W., & Gawronski, B. (2007). And deplete us not into 565–584.
temptation: Automatic attitudes, dietary restraint, and self-regulatory Nosek, B. A., & Banaji, M. R. (2001). The Go/No-Go Association Task.
resources as determinants of eating behavior. Journal of Experimental Social Cognition, 19, 625– 666.
Social Psychology, 43, 497–504. Nunes, K. L., Firestones, P., & Baldwin, M. W. (2007). Indirect assessment
Hugenberg, K., & Bodenhausen, G. V. (2003). Facing prejudice: Implicit of cognitions of child sexual abusers with the Implicit Association Test.
prejudice and the perception of facial threat. Psychological Science, 14, Criminal Justice and Behavior, 34, 454 – 475.
640 – 643. Olson, M. A., & Fazio, R. H. (2001). Implicit attitude formation through
Hugenberg, K., & Bodenhausen, G. V. (2004). Ambiguity in social cate- classical conditioning. Psychological Science, 12, 413–147.
gorization: The role of prejudice and facial affect in racial categoriza- Olson, M. A., & Fazio, R. H. (2002). Implicit acquisition and manifestation
tion. Psychological Science, 15, 342–345. of classically conditioned attitudes. Social Cognition, 20, 89 –104.
Jordan, C. H., Spencer, S. J., Zanna, M. P., Hoshino-Browne, E., & Correll, Olson, M. A., & Fazio, R. H. (2003). Relations between implicit measures
J. (2003). Secure and defensive high self-esteem. Journal of Personality of prejudice: What are we measuring? Psychological Science, 14, 636 –
and Social Psychology, 85, 969 –978. 639.
Kafka, F. (1954). The third notebook, October 18, 1917. In F. Kafka, Olson, M. A., & Fazio, R. H. (2004). Reducing the influence of extra-
Dearest Father. Stories and other writings. New York: Schocken Books. personal associations on the Implicit Association Test: Personalizing the
Kamphuis, J. H., de Ruiter, C., Janssen, B., Spiering, M. (2005). Prelim- IAT. Journal of Personality and Social Psychology, 86, 653– 667.
inary evidence for an automatic link between sex and power among men Olson, M. A., & Fazio, R. H. (2006). Reducing automatically activated
who molest children. Journal of Interpersonal Violence, 20, 1351–1365. racial prejudice through implicit evaluative conditioning. Personality
Karpinski, A., & Hilton, J. L. (2001). Attitudes and the Implicit Associa- and Social Psychology Bulletin, 32, 421– 433.
tion Test. Journal of Personality and Social Psychology, 81, 774 –788. Paulhus, D. L. (1984). Two component models of social desirable respond-
Karpinski, A., & Steinman, R. B. (2006). The Single Category Implicit ing. Journal of Personality and Social Psychology, 46, 598 – 609.
Association Test as a measure of implicit social cognition. Journal of Payne, B. K., Burkley, M., & Stokes, M. B. (2008). Why do implicit and
Personality and Social Psychology, 91, 16 –32. explicit attitude tests diverge? The role of structural fit. Journal of
Kim, D. (2003). Voluntary controllability of the Implicit Association Test Personality and Social Psychology, 94, 16 –31.
(IAT). Social Psychology Quarterly, 66, 83–96. Payne, B. K., Cheng, S. M., Govorun, O., & Stewart, B. D. (2005). An
Klauer, K. C., & Musch, J. (2003). Affective priming: Findings and inkblot for attitudes: Affect misattribution as implicit measurement.
theories. In J. Musch & K. C. Klauer (Eds.), The psychology of evalu- Journal of Personality and Social Psychology, 89, 277–293.
ation: Affective processes in cognition and emotion (pp. 7– 49). Payne, B. K., Jacoby, L. L., & Lambert, A. J. (2005). Attitudes as
Mahwah, NJ: Erlbaum. accessibility bias: Dissociating automatic and controlled components. In
150 GAWRONSKI

R. R. Hassin, J. S. Uleman, & J. A. Bargh (Eds.), The new unconscious compatibility with the meaning of eliciting verbal signs. Journal of
(pp. 393– 420). New York: Oxford University Press. Experimental Psychology, 59, 239 –245.
Perugini, M. (2005). Predictive models of implicit and explicit attitudes. Spruyt, A., Hermans, D., De Houwer, J., Vandekerckhove, J., & Eelen, P.
British Journal of Social Psychology, 44, 29 – 45. (2007). On the predictive validity of indirect attitude measures: Predic-
Petty, R. E., Tormala, Z. L., Brinol, P., & Jarvis, W. B. G. (2006). Implicit tion of consumer choice behavior on the basis of affective priming in the
ambivalence from attitude change: An exploration of the PAST Model. picture-picture naming task. Journal of Experimental Social Psychology,
Journal of Personality and Social Psychology, 90, 21– 41. 43, 599 – 610.
Riketta, M. (2006). Gender and socially desirable responding as modera- Steffens, M. C. (2004). Is the Implicit Association Test immune to faking?
tors of the correlation between implicit and explicit self-esteem. Current Experimental Psychology, 51, 165–179.
Research in Social Psychology, 11, 14 –28. Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of
Rothermund, K., Teige-Mocigemba, S., Gast, A., & Wentura, D. (in press). social behavior. Personality and Social Psychology Review, 8, 220 –247.
Minimizing the Influence of Recoding in the IAT: The IAT–RF. Quar- Teachman, B. A. (2007). Evaluating implicit spider fear associations using
terly Journal of Experimental Psychology. the Go/No-Go Association Task. Journal of Behavior Therapy and
Rudman, L. A. (2004). Sources of implicit attitudes. Current Directions in Experimental Psychiatry, 38, 156 –167.
Psychological Science, 13, 79 – 82. Teachman, B. A., Gregg, A. P., & Woody, S. R. (2001). Implicit associ-
Rudman, L. A., Phelan, J. E., & Heppen, J. B. (2007). Developmental ations for fear-relevant stimuli among individuals with snake and spider
sources of implicit attitudes. Personality and Social Psychology Bulletin, fears. Journal of Abnormal Psychology, 110, 226 –235.
33, 1700 –1713. Teige, S., Schnabel, K., Banse, R., & Asendorpf, J. B. (2004). Assessment
Rydell, R. J., & Gawronski, B. (in press). I like you, I like you not: of multiple implicit self-concept dimensions using the Extrinsic Affec-
Understanding the formation of context dependent automatic attitudes. tive Simon Task. European Journal of Personality, 18, 495–520.
Cognition and Emotion. Teige-Mocigemba, S., & Klauer, K. C. (2008). “Automatic” evaluation?
Rydell, R. J., & McConnell, A. R. (2006). Understanding implicit and Strategic effects on affective priming. Journal of Experimental Social
explicit attitude change: A systems of reasoning analysis. Journal of Psychology, 44, 1414 –1417.
Personality and Social Psychology, 91, 995–1008. Teige-Mocigemba, S., Klauer, K. C., & Rothermund, K. (2008). Minimiz-
Rydell, R. J., McConnell, A. R., Mackie, D. M., & Strain, L. M. (2006). Of ing method-specific variance in the IAT: The Single Block IAT. Euro-
two minds: Forming and changing valence-inconsistent implicit and pean Journal of Psychological Assessment, 24, 237–245.
explicit attitudes. Psychological Science, 17, 954 –958. Webb, E. J., Campbell, D. T., Schwartz, R. D., & Sechrist, L. (1966).
Sartori, G., Agosta, S., Zogmaister, C., Ferrara, S. D., & Castiello, U. Unobtrusive measures: Nonreactive research in the social sciences.
(2008). How to accurately assess autobiographical events. Psychological Chicago: Rand McNally.
Science, 19, 772–780. Wiers, R. W., Houben, K., & de Kraker, J. (2007). Implicit cocaine
Schmukle, S. C., & Egloff, B. (2006). Assessing anxiety with Extrinsic associations in active cocaine users and controls. Addictive Behaviors,
Simon Tasks. Experimental Psychology, 53, 149 –160. 32, 1284 –1289.
Schnabel, K., Banse, R., & Asendorpf, J. B. (2006). Employing automatic Wiers, R. W., Van Woerden, N., Smulders, F. T. Y., & De Jong, P. J.
approach and avoidance tendencies for the assessment of implicit per- (2002). Implicit and explicit alcohol related cognitions in heavy and light
sonality self-concept: The Implicit Association Procedure (IAP). Exper- drinkers. Journal of Abnormal Psychology, 111, 648 – 658.
imental Psychology, 53, 69 –76. Williams, J. M., Mathews, A., & MacLeod, C. (1996). The emotional
Stroop task and psychopathology. Psychological Bulletin, 120, 3–24.
Seibt, B., Neumann, R., Nussinson, R., & Strack, F. (2008). Movement
Wilson, T. D., & Dunn, E. W. (2004). Self-knowledge: Its limits, value,
direction or change in distance? Self- and object-related approach-
and potential for improvement. Annual Review of Psychology, 55, 493–
avoidance motions. Journal of Experimental Social Psychology, 44,
518.
713–720.
Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual
Sherman, S. J., Rose, J. S., Koch, K., Presson, C. C., & Chassin, L. (2003).
attitudes. Psychological Review, 107, 101–126.
Implicit and explicit attitudes toward cigarette smoking: The effects of
Wittenbrink, B. (2007). Measuring attitudes through priming. In
context and motivation. Journal of Social and Clinical Psychology, 22,
B. Wittenbrink & N. Schwarz (Eds.), Implicit measures of attitudes (pp.
13–39.
17–58). New York: Guilford.
Sinclair, S., Dunn, E., & Lowery, B. S. (2005). The relationship between
Wittenbrink, B., Judd, C. M., & Park, B. (1997). Evidence for racial
parental racial attitudes and children’s implicit attitudes. Journal of
prejudice at the implicit level and its relationships with questionnaire
Experimental Social Psychology, 41, 283–289.
measures. Journal of Personality and Social Psychology, 72, 262–274.
Sloman, S. A. (1996). The empirical case for two systems of reasoning.
Wittenbrink, B., Judd, C. M., & Park, B. (2001). Spontaneous prejudice in
Psychological Bulletin, 119, 3–22.
context: Variability in automatically activated attitudes. Journal of Per-
Smith, C. T., & Nosek, B. A. (2007). Affective focus increases the con-
sonality and Social Psychology, 81, 815– 827.
cordance between implicit and explicit attitudes. Manuscript submitted
Wittenbrink, B., & Schwarz, N. (Eds.). (2007). Implicit measures of
for publication.
attitudes. New York: Guilford.
Smith, E. R., & DeCoster, J. (2000). Dual-process models in social and
cognitive psychology: Conceptual integration and links to underlying
memory systems. Personality and Social Psychology Review, 4, 108 – Received June 20, 2008
131. Revision received August 7, 2008
Solarz, A. K. (1960). Latency of instrumental responses as a function of Accepted December 5, 2008 䡲