Moderated
consistency between direct, indirect, and behavioral indicators of dispositions
Traits are useful as descriptive personality
constructs to the extent that individual differences in behavior and other indicators,
such as physiological reactions, are consistent across equivalent situations
and stable across time. Whether and how well this assumption is met has been a
matter of repeated controversy in the history of personality psychology and
attitude research (Kenrick & Funder, 1988; Schmitt, 1990). The first
controversy in the 1930s was initiated by research on the cross-situational
consistency of moral behavior (Hartshorne & May, 1928) and by LaPierre’s
(1934) study on attitude-behavior consistency. The second debate occurred due
to Deutscher’s (1966) review of the consistency between words and deeds,
Mischel’s (1968) review of the predictive validity of personality and
achievement measures, and Wicker’s (1969) review of attitude-behavior
consistency.
Both controversies followed a characteristic sequence
of arguments and studies. First, the degree of consistency was debated,
especially the degree of behavioral consistency across situations, and the
predictability of behavior from trait measures (the “how much” question).
Second, boundary conditions of consistency were proposed and identified
empirically (the “when” question). Finally, elaborate theoretical ideas were
proposed for explaining person-situation interactions and subjected to targeted
empirical tests (the “why” question; Kenrick & Funder, 1988; Schmitt, 1990;
Swann & Seyle, 2005).
Perhaps one of the most important results of both
controversies was the recognition that consistency is not a constant but a
variable (Baumeister & Tice, 1988; Chaplin, 1991; Fiske & Rice, 1955;
Schmitt, 1990; Snyder & Ickes, 1985). As a consequence of this insight, the
theory-guided identification of moderators of consistency became an important
goal for personality and attitude research.
Recently, the
consistency issue has been revived due to the low consistency that was found in
many studies between (a) direct self-report measures of attitudes and personality
traits and (b) indirect measures, such as the Implicit Association Test (IAT;
Greenwald, McGhee, & Schwartz, 1998). Our program of research and ongoing theoretical development are devoted
to understanding this finding based on theoretical and methodological insights
as well as on empirical findings that were the result of earlier consistency
controversies. With the assumption that consistency is a variable, our thinking
and research are guided by the goal of identifying those moderator variables that
shape consistency.
We began with a systematic analysis of the literature
on implicit-explicit consistency and a meta-analysis of available findings
(Hofmann, Gawronski, Gschwendner, Le, & Schmitt, 2005). This meta-analysis was
limited to the IAT because, at the beginning of our literature review in 2002,
IAT measures were by far the most-often used indirect measures in studies on
implicit-explicit consistency. Based on a sample of 126 studies, we obtained a
mean correlation between direct and indirect (IAT) measures for the same
construct of .24, with approximately half of the variability across
correlations attributable to the moderator variables that we were able to
consider. Correlations systematically increased as a function of (a) increasing
spontaneity of self-reports and (b) increasing conceptual correspondence
between measures. These results suggest that direct and indirect measures are
systematically related, but that higher-order inferences and lack of conceptual
correspondence can reduce the convergence of automatic associations and
explicit self-reports.
A moderated consistency model for implicit, explicit,
and behavioral indicators of latent dispositions
Next, we developed a theoretical model that could
account for the systematic variability in implicit-explicit consistency that
was identified by our meta-analysis and in narrative reviews conducted by
ourselves and other authors (Gschwendner, Hofmann, & Schmitt,
2006a; Hofmann, Gschwendner, Nosek, & Schmitt, 2005; Nosek, 2005, 2007).
Based on theoretical advances and new findings, our model was refined in
several steps (Hofmann, Gschwendner, Nosek, & Schmitt, 2005; Gschwendner,
Hofmann, & Schmitt, 2006a, 2006b).
Consistent
with a predominant view in the current literature, the model assumes two types of mental representations of latent
dispositions such as attitudes, personality traits, motives, self-concept
dimensions, self-esteem, and beliefs: Explicit dispositions are assumed to be
represented in a propositional format (Gawronski & Bodenhausen, 2006),
introspectively accessible, and measurable via direct self-report. Explicit
dispositions serve as a self-knowledge base for deliberate thinking and
reasoned judgment. Implicit dispositions are assumed to be stored in an associative
format as object-attribute links (e.g., me – anxious; spiders - disgusting).
These associations are unconscious and introspectively opaque. Implicit
dispositions provide the knowledge base for quick evaluations and intuitive
judgments that occur outside a person’s conscious awareness. This is why
implicit dispositions can be measured only indirectly. In line with
dual-process models (Smith & DeCoster, 2000; Strack & Deutsch, 2004),
we assume that explicit dispositions feed into reasoned action based on the
controlled processing of relevant information, such as the anticipation of
consequences that a certain behavior will have. In contrast, implicit
dispositions are assumed to affect behavior automatically via approach and
avoidance impulses and the activation of behavioral schemata or scripts.
The most recent version of our model is depicted in
the Figure below. It differs slightly in terminology from previous versions of
the model (Hofmann, Gschwendner, Nosek, & Schmitt, 2005;
Gschwendner, Hofmann, & Schmitt, 2006a). Unlike these earlier versions, the
current model includes behavior in addition to direct and indirect measures.
Behavior was added because the consistency issue is not limited to direct and
indirect measures of dispositions, but rather encompasses all sorts of relevant
indicators, including overt behavior. Moreover, overt behavior is often
considered to be the ultimate criterion for the usefulness of a construct and
the validity of measures. This is especially true for new constructs, such as
implicit constructs, and new measures, such as the IAT (Greenwald,
Poehlman, Uhlmann, & Banaji, in press). Finally, the current model differs
from an earlier version of the model (Gschwendner, Hofmann, & Schmitt,
2006b) in that it differentiates between manifest behavior,
behavioral plans and intentions, and behavioral schemata or scripts. In line
with action theories, such as the theory of planned behavior (Ajzen, 1987), we
assume that behavioral plans and intentions mediate the effect of explicit dispositions
on manifest behavior. Further, and in line with the Reflective Impulsive Model
(RIM) proposed by Strack and Deutsch (2004), we assume that behavioral schemata
and scripts mediate the effect of implicit dispositions on manifest behavior.
The model specifies nine effects (depicted as
single-headed arrows) and three correlations (depicted as double-headed
arrows). Most importantly, from our primary research interest, it is assumed
that the size of all effects and correlations can depend on moderator
variables. Moreover, we deem it likely that the size of each effect and
correlation depends on more than one moderator. For this reason, moderators
were combined into groups in the model. Each moderator group contains several
factors, including personality traits or characteristics of the situation, in
which measurement or behavior occurs. Moderator groups also include attributes
of the construct at issue, such as the social desirability of a personality
trait or an attitude, and attributes of their indicators, such as the degree of
controllability (versus automaticity) of a specific behavior, or the
reliability of a measure. These differentiations between types of moderators
are not included in the model in order to keep it principled and parsimonious.
The model also does not specify joint moderator effects. We assume, for
instance, that functionally equivalent moderators, such as the chronic and
acute activation of associations, will moderate the effect of implicit
dispositions on indirect measures jointly in a synergistic fashion
(Gschwendner, Hofmann, & Schmitt, 2006; 2008).
It is important to understand that implicit-explicit
consistency cannot be observed directly on the level of constructs (circular
frames in the model), but only on the level of manifest indicators (rectangular
frames in the model). Note that most moderators are assumed to affect causal
links between latent constructs (A, B, C, D) and the effects of these latent
constructs on their manifest indicators (E, F, G, H). These assumptions cannot
be tested directly. Only their implications on the correlations among the
manifest indicators can be tested. This is not a specific limitation of our
model, but rather an implied limitation of any latent variable model.
Note next that although the model contains a
substantial number of variables and paths, it is still a simplification. This
is especially true for the moderators of the model. First, moderators are not
specified as separate variables, but are only summarized as groups. Second, moderators
are constructs that either need to be measured (personality factors) or varied
experimentally (properties of the situation, the behavior, and the measures).
Thus, a more complete version of the model would have to include latent and
manifest moderator variables and, in addition, possible interactions between
them. These details are not included in the model in order to keep it general
and flexible. Thus the model is not a detailed effect model for a specific
disposition, but a theoretical framework that can be applied to large variety
of dispositions.
We will first briefly describe the nine causal paths
that are numbered accordingly in the Figure.
1.
In line with other authors, we assume that
object-attribute associations form the elementary basis of propositional
thinking and explicit self-knowledge (Gawronski & Bodenhausen,
2006; Strack & Deutsch, 2004). This assumption is
represented by the causal path from the implicit to the explicit disposition
(Path 1).
2.
Propositional thinking may in turn shape the structure
of the underlying associative representations (Strack & Deutsch, 2004).
Autosuggestion and self-instruction are examples. In the domain of attitudes,
the intentional imagination of counter-attitudinal exemplars can weaken
previous associations between attitude objects and values. The same process can
lead to changes in implicit stereotypes, implicit self-esteem, and implicit
self-concept (Gawronski & Bodenhausen, 2006). This type of process is represented in the model
by a causal path from the explicit disposition to the implicit disposition
(Path 2).
3.
Explicit dispositions are assumed to cause behavior.
This assumption is consistent with results from a vast number of studies
showing that behavior can be predicted from personality factors, attitudes,
self-concept dimensions, motive traits, self-esteem, and beliefs. In line with
action theories, such as the theory of planned behavior (Ajzen, 1987), and also
in line with dual process theories such as the RIM (Strack & Deutsch,
2004), we assume that explicit dispositions do not affect behavior directly,
but rather indirectly via action plans and behavioral intentions (Path 3).
People reason about the benefits and costs of the behavioral options they
consider, and depending on these and other factors, they make choices. The
preferred choice is turned into a behavioral intention. Moreover, many goals
require sequences of behavioral steps. These need to be designed and
represented mentally as action plans.
4.
Implicit dispositions are also assumed to cause
behavior. Although this assumption has been tested much less often than the
previous one, a recent review provides clear evidence in support of Path 4 of
our model (Greenwald, Poehlman, Uhlmann, & Banaji, in press). In line with
dual process models such as the RIM (Strack & Deutsch, 2004), we assume
that implicit dispositions do not feed directly into behavior, but via
behavioral schemata or scripts instead. More specifically, it is assumed that
the activation of object-attribute links automatically co-activates behavioral
schemata or scripts.
5.
Path 5 represents a common notion of all latent trait
theories. It is assumed in these theories that self-reports represent a
specific form of behavior that is causally determined by the knowledge that
people have about their traits. Moreover, it is assumed that the process that
transforms the explicit disposition into self-reports is a conscious and
controlled process similar to the formation of intentions and behavioral plans
that precede behavior (Path 3).
6.
Accordingly, Path 6 of our model represents the
assumption that implicit dispositions cause the behavior that people show in
indirect measurement procedures such as the IAT.
7.
Path 7 reflects a common notion of action theories
that behavior is the controlled execution of previously-formed plans and
intentions (Ajzen, 1987).
8.
Path 8 reflects the core assumption of dual process
theories such as the RIM (Strack & Deutsch, 2004) that activated behavioral
schemata feed automatically into behavior without conscious awareness.
9.
Path 9 was derived from self-observation theory (Bem,
1972). Behavior is never entirely consistent with explicit dispositions because
it is jointly determined by controlled and automatic processes. People are able
to detect inconsistencies between their explicit dispositions and their
behavior. Such inconsistencies generate cognitive dissonance that needs to be
resolved. Changing explicit assumptions about the self is one way of dissolving
such inconsistencies. Because this process is more likely to occur in cases
where behavior was driven automatically, it may be an important route by which
implicit associations feed into explicit self-knowledge (cf. Hofmann,
Gschwendner, Nosek & Schmitt, 2005).
We will now turn
to moderators that affect the strength of these paths and, in addition, the
size of the correlations between direct, indirect, and behavioral indicators of
a disposition.
Moderator Group A: Explicit self-knowledge can draw upon implicit self-knowledge to the
extent that implicit self-knowledge is accessible. The accessibility of
associations depends on their strength. Several studies have shown, for
instance, that strong attitudes are more accessible than weak attitudes (Petty
& Krosnick, 1995).
Compared to weak object-value associations, strong associations should feed
more easily into explicit self-knowledge. Support for this assumption comes from
a study by Nosek (2005). The notion of associative strength can be generalized
from attitudes to other dispositions such as stereotypes, personality traits,
and self-concepts because these dispositions are also represented as
associations between objects and attributes. A second important moderator is
awareness. People differ in their motivation and ability to introspect
(Fenigstein, Scheier, & Buss, 1975). The strength of Path 1 of our model
should vary accordingly (cf. Hofmann, Gschwendner, Nosek, & Schmitt, 2005).
Moderator Group B: Auto-suggestion, self-instruction, and
selective exposure to object-attribute combinations contribute to the
translation of elaborate assumptions about the self into simpler representations
of self-knowledge. Again, the effectiveness of this process should depend on
motivational factors and on abilities. People who dislike their impulsive
reactions to certain objects, such as certain kinds of food, sometimes
consciously engage in counter-impulsive auto-suggestions. People who dislike
their attitudes / stereotypes / self-concept can engage in intentional
imagination of counter-attitudinal exemplars of these attitudes / stereotypes /
self-concept. The effectiveness of this process will depend on knowledge about
the underlying psychological principle, on motivational strength, and probably
also on personality factors such as self-control and openness to experience.
Moderator Group C: Making plans requires both
the motivation and the opportunity to deliberate about relevant information.
Both can vary depending on stable individual differences between people and as
a function of situational factors. If people have no time to deliberate, they
cannot make plans. Further, making plans and carefully considering the pros and
cons of various behavioral alternatives requires cognitive capacity and
ability. More generally, all factors that contribute to the availability and
usability of control resources will affect Path 3. In addition, some people
rely more on intuition than on deliberation (Betsch, 2004; Epstein, Pacini,
Denes-Raj, & Heier, 1996) and this will also contribute to how carefully action
plans are made on the basis of relevant information. In the domain of attitudes,
people may know that they hold negative attitudes that are politically
incorrect. To the extent that they are motivated to control prejudiced
reactions (Dunton & Fazio
1997),
they will refrain from translating their attitudes into consistent behavioral
intentions. Additional moderators in this group are reviewed by Friese,
Hofmann, and Schmitt (in press).
Moderator Group D: The strength
with which behavioral schemata are activated in a specific situation depends on
how closely they are linked with objects. This in turn will depend on how often
both have been activated simultaneously during the person’s learning history.
Social learning and modeling may play an important role here. Consider a person
who holds negative attitudes toward strangers and who repeatedly observes how
peers with the same negative attitude treat strangers. Very likely, this person
will not only develop a behavioral script vis à vis strangers, but will also
develop an associative link between the attitude object “stranger” and this
script (Castelli, De Dea, & Nesdale, 2008).
Moderator Group E: People are not
always willing to disclose their self-knowledge honestly. Rather, they tend to
adjust self-reports to personal goals. Pervasive motivational sources of these
goals are self-presentation, social desirability, and impression management.
The strength of these motives varies across individuals, across dispositions,
and across the measurement context. For instance, some individuals are more
motivated than others to control prejudiced reactions (Dunton & Fazio 1997), and several studies have
confirmed that this moderator indeed affects the validity of self-reports (for
a summary see Hofmann, Gschwendner, Nosek, & Schmitt, 2005). Furthermore,
some situations will more likely trigger adjustment processes than others.
Anonymous self-reports of socially undesirable attitudes, stereotypes, beliefs,
motives, and personality traits are more valid than public self-reports on the
same dispositions.
Moderator Group F: Indirect
procedures such as the IAT are intended to measure behavior that is driven
automatically by the activation of object-attribute associations. It follows
directly from this assumption that the validity of such measures depends on the
accessibility of the association at issue. This in turn depends on the strength
of the relevant associations and thus the person’s learning history. In
addition, the validity of measures for object-attribute associations will vary
depending on how strongly these associations have been pre-activated and how
strongly they are activated by the measurement procedure itself (Blair, 2002).
Pre-activation can be achieved via priming. Moreover, measurement procedures
differ in how strongly they activate associated concepts. For instance,
embedding object-stimuli in a congruent context can promote activation, and
thus increase the validity of the indirect measure (Gschwendner, Hofmann, &
Schmitt, 2008a).
Moderator Group G: Action plans and
behavioral intentions are not transformed automatically into behavior or chains
of behavioral steps. Rather, they require self-regulation. Self-regulation
consists of self-monitoring and self-management strategies such as the
adjustment of one’s behavior if goal discrepancy is beyond an acceptable limit.
Self-monitoring (Snyder, 1974) and self-management abilities such as
self-efficacy (Bandura, 1977) vary between individuals. In addition,
self-regulation requires the availability (opportunity) and willingness to
invest (motivation) control resources. In line with this reasoning, chronic and
acute control resources have been shown to influence the relative weight of
explicit and implicit dispositions on behavior (Hofmann, Gschwendner, Castelli, & Schmitt, 2008; Hofmann, Gschwendner, Wiers, Friese & Schmitt, 2008).
Moderator Group H: Behavioral
schemata and scripts feed automatically into behavior depending on how strongly
they are linked with the behavioral object, how strongly they are activated,
and how much they are overridden by controlled processes. The first factor depends
on the person’s learning history (cf. Moderator Group D). The second moderator
is a function of time. Activation fades away quickly, and therefore, schemata
will affect behavior automatically only if it follows in a short amount of
time. Whether and to what extent controlled processes override impulses depends
on the moderators from Group G. The person’s primary reliance on deliberation
versus intuition (cf. Moderator Group C) may also contribute to a moderator
effect at this point.
Moderator Group I: Our model
assumes that the self-observation of behavior will feed back into explicit
self-knowledge, especially if behavior was not consistent with previously held
assumptions about the self. Inconsistencies are more likely to occur to the
extent that behavior was driven by implicit associations and behavioral
schemata linked with these associations. This is more likely for behaviors that
are difficult to control such as certain kinds of nonverbal behavior
(Asendorpf, Banse, & Mücke, 2002). We assume that the feed-back effect will
be stronger for people who are willing and able to self-observe and who are
capable of interpreting their behavior adequately. Ability and willingness to
self-observe varies with self-monitoring (Snyder, 1974) and private self-consciousness
(Fenigstein, Scheier, & Buss, 1975). The ability to decode behavioral cues
varies between individuals (Hefter, Manoach, &
Barton, 2005). Moreover, it can be trained (Ekman & Friesen, 1978).
Furthermore, the self-observation of behavior requires the opportunity to do so
and this opportunity in turn may vary between social contexts and situations.
Finally, people differ in their need for consistency and their tolerance for
ambiguity. Compared to people with a high need for consistency and a low tolerance
for ambiguity, those with a low consistency need and a high ambiguity tolerance
will not be as motivated to support their explicit self-assumptions with the
appropriate self-observed behavior.
Moderator Group J: The strengths of
all direct and moderator effects we have mentioned thus far will, by
implication, affect the correlations between direct measures, indirect
measures, and behavior. In addition, the convergence of these indicators will
depend on features of the measurement instruments that are known to affect the
size of correlations. Important examples of these factors are the reliability
and the symmetry of the measure in terms of content and specificity. For
instance, the correlation between indirect and direct measures that capture
different facets of a construct will be lower than the correlation between two
such measures that capture the same facets (Schmitt & Borkenau, 1992). This
moderator is very relevant because the kinds of behavior that have typically
been used in studies of predictive validity often do not match in content and
specificity with the measures that are employed for their prediction. This has
been a repeated issue in research on attitude-behavior consistency (Ajzen &
Fishbein, 1977). One of our recent studies shows that the degree of similarity
in content and specificity systematically moderates the convergence among
direct, indirect, and behavioral indicators of dispositions (Gschwendner,
Hofmann, & Schmitt, in press).
Empirical test of the moderated consistency model
Our model has not been tested in its entire
complexity. However, many of the assumed moderator effects have been
investigated in our own research and studies by others. The available evidence
has been summarized by us in three narrative reviews (Friese, Hofmann &
Schmitt, in press; Gschwendner, Hofmann, & Schmitt, 2006a; Hofmann,
Gschwendner, Nosek, & Schmitt, 2005). We will now briefly summarize our own empirical research.
Joint moderator effects of individual differences in awareness and
adjustment
Our first
empirical study (Hofmann, Gschwendner, & Schmitt, 2005) addressed
moderators from Group A (awareness) and Group E (adjustment). In two
experiments on attitudes of West Germans toward East Germans and Turks, a
number of dispositional moderators pertaining to awareness and adjustment were
tested. Concerning moderators affecting awareness, no reliable first-order
effects were found for Private Self-Consciousness or Attitudinal
Self-Knowledge. However, Attitude Importance generated the expected effect. Concerning
the influence of moderators on adjustment, consistent effects were obtained for
Motivation to Control Prejudiced Reactions. Social Desirability and
Self-Monitoring did not moderate the implicit–explicit relationship in the
expected direction. Some evidence was found for a second-order moderator effect
between awareness and adjustment, suggesting that adjustment effects may be
more pronounced under conditions of high awareness.
Synergistic moderator effects of person and situation factors of awareness
and adjustment
As the next
step, again using attitudes of Germans toward Turks as an application of our
model, the previous study was extended by manipulating moderators of awareness
(Group A) and adjustment (Group E) experimentally, in addition to measuring
individual differences in these moderators (Gschwendner, Hofmann, &
Schmitt, 2006c). Besides including experimentally manipulated moderators, this
study differed from the previous one in that some moderator constructs from
Groups A and E were replaced or measured with better measures. Results were as
follows. Concerning moderators of adjustment, no effects on explicit-implicit
consistency were obtained for situational variables. The expected synergistic
interaction of personal and situational variables was also not significant.
However, concerning moderators of awareness, a reliable first-order effect was
found for Private Self-Consciousness. Moreover, Private Self-Consciousness and
experimentally manipulated motivation to introspect showed the assumed
synergistic interaction moderator effect.
The effects of acute and chronic construct accessibility on the temporal
stability of the IAT
Another study was
devoted to Moderator Group F (Gschwendner, Hofmann, & Schmitt, 2008). We
tested the assumption that the validity, and thus the temporal stability, of
indirect measures increases with the accessibility of the associated concepts
during the measurement process. Adopting a procedure employed by Wittenbrink,
Judd, and Park (2001), accessibility was manipulated experimentally by
including or not including background pictures in the IAT. In Study 1, the
2-week stability of an IAT assessing anxiety was higher when IAT stimuli were
embedded in an anxiety-relevant background (e.g., a snake). In Study 2, this
context effect could be replicated in the domain of racial attitudes. Moreover,
the context effect in Study 2 was especially pronounced for participants with
high chronic access to the relevant concept.
The moderating role of situationally available control resources
Another study (Hofmann,
Gschwendner, Castelli, & Schmitt, 2008) tested the assumption that control
resources have an effect on how strongly explicit and implicit dispositions
determine behavior (Moderator Groups C, D, G, H). More specifically, we investigated
how implicit attitudes (Implicit Association Test) and explicit attitudes
(Blatant/Subtle prejudice) were related to interracial interaction behaviors of
Italians toward an African interviewer (Study 1) and of Germans toward a
Turkish interviewer (Study 2). For half of the interview questions,
participants’ control resources were reduced via a memory task. Across both
studies, the Race IAT was more predictive of behavior when participants were
mentally taxed than when untaxed. Conversely, explicit attitudes were somewhat
more predictive under full resources. Taken together, our findings suggest that
available control resources moderate the predictive validity of implicit and
explicit attitudes (for similar findings, see Hofmann, Gawronski, & Rauch,
2007).
The moderating role of chronically available control resources
Next, we (Hofmann,
Gschwendner, Friese, Wiers, & Schmitt, 2008) tested whether chronically
available control resources have the same effect that we had predicted and found
for situationally available control resources. Results were fully consistent
with our expectations. In two studies on sexual interest behavior (Study 1) and
the consumption of tempting food (Study 2), automatic attitudes toward the
temptation of interest had a stronger influence on behavior for individuals who
scored low rather than high in working memory capacity. Analogous results
emerged in Study 3 on anger expression in a provoking situation when a measure
of the automatic personality trait of angriness was employed. Conversely,
controlled dispositions such as explicit attitudes (Study 1) and
self-regulatory goals (Studies 2 and 3) were more effective in guiding behavior
for participants who scored high rather than low in working memory capacity.
The moderating effect of content and specificity similarity on the
consistency of direct, indirect, and behavioral construct measures
Our next study
looked at two moderators from Group J: content and specificity similarity (Gschwendner, Hofmann, & Schmitt, 2008b). In the first
session, different general and specific anxiety measures were administered,
among them an IAT for general anxiety, an IAT for spider anxiety and an IAT
assessing speech anxiety. In the second session, participants had to deliver a
speech. Behavioral indicators of speech anxiety were measured. Results showed,
in line with the moderator hypotheses, that (a) implicit and explicit anxiety
measures correlated significantly only on the same specification level and if
they measured the same content, and (b) specific speech anxiety measures best
predicted concrete anxious behavior.
Self-perception of automatic behavior as a potential source of
implict-explicit consistency
Hofmann, Gschwendner, and
Schmitt (under review) tested Path 9 of our model. Showing that the
self-perception of automatic behavior feeds back into explicit self-knowledge
and thus contributes to implicit-explicit consistency is a necessary
prerequisite for exploring moderators of such a process. This is true because
only ordinal moderator effects seem plausible. Ordinal moderator effects imply
a main effect of the independent variable (in this case: automatic behavior) on
the dependent variable (in this case: explicit assumptions about the self). So
far, we have not been able to demonstrate the assumed self-perception effect.
However, we were able to show that the automatic behavior of a target person
affects explicit judgments about the target person in neutral observers,
demonstrating both cue validity (implicit dispositions cause automatic
nonverbal behavior) and cue utilization (explicit dispositions can be inferred
from nonverbal behavioral cues). Across three studies in the domains of
extraversion and anxiety, we consistently obtained reliable cue validity and
cue utilization for neutral observers but not for self-perceivers. An
additional measure of state inferences in Study 2 showed that one reason for
the lack of the assumed self-perception effects among self-perceivers is their
reluctance to use their state inferences as a basis for more general trait
inferences. It seems that people have a “blind spot” with respect to the
nonverbal behavioral manifestations of their unconscious selves, even though
neutral observers may readily detect and utilize this information for dispositional
inferences.
Current research
Our current research follows three lines.
Explicit and implicit disgust sensitivity and disgust behavior
For human
beings, automatic information processing and impulsive behavior are adaptive capacities
because they require very few cognitive resources and occur very fast. The cost
of these advantages is the lack of behavioral flexibility. This cost seems
irrelevant in situations that are potentially dangerous. In line with this
reasoning, implicit anxiety has been shown to be a powerful predictor of
anxious behavior (Egloff & Schmukle, 2002; Gschwendner, Hofmann, &
Schmitt, in press). Besides anxiety, disgust is another emotion that serves
survival. For this reason, and because disgust, unlike anxiety, is an
under-researched emotion, we currently apply our model to disgust sensitivity
and disgust behavior. We want to replicate some of the moderator effects that
we were able to demonstrate in the previous studies. We are especially
interested in using disgust to test Path 9 of our model again, and to test for
potential moderators of this path (Moderator Group I). Disgust seems a suitable
domain for this purpose because disgust behavior contains automatic components
such as facial expression and rapid withdrawal from a disgusting object.
Extending previous attempts to demonstrate Path 9, our current research
includes more moderators from Group I.
Explicit-implicit consistency as an independent variable
Like most other
research, our own research has treated implicit-explicit consistency as a
dependent variable. Our model seeks to explain why the consistency of direct
and indirect measures for explicit and implicit dispositions varies across
individuals, situations, dispositions, and measures. Explicit-implicit
consistency may not only be the result of information processing, but also
affect information processing, behavior, and long-term outcomes. Research in
the self-esteem domain may illustrate the point. Some studies have shown that
various combinations of high versus low explicit versus implicit self-esteem
are associated with different outcomes (Schröder-Abé, Rudolph, & Schütz,
2007; Schröder-Abé, Rudolph, Wiesner, & Schütz, 2007). More specifically,
implicit-explicit self-esteem discrepancies seem to make people vulnerable to
threatening information about the self.
We continue
with this line of research in the domain of the intelligence self-concept.
First, we are interested in exploring whether or not the implicit intelligence
self-concept is related to a person’s performance in an objective intelligence
test. Whereas several studies have shown that peoples’ explicit intelligence
self-concept is highly correlated with their objective performance, no study
has yet tested whether people also have an implicit knowledge of their
intelligence. It is also yet unknown whether the explicit and implicit
intelligence self-concepts exhibit consistency or not, and whether the implicit
intelligence self-concept has a unique overlap with objective performance in
intelligence tests.
Second, we are
interested in exploring effects of explicit-implicit consistency on the
processing of self-relevant information. More specifically, we want to
determine whether a discrepant self-concept renders a person more susceptible
to (false) feedback on objective performance.
Objective personality tests
Cattell (1957)
proposed three kinds of data for measuring personality: Q-data (questionnaire
data), L-data (life-data, often obtained from observers), and T-data (test
data). Measuring personality traits with objective tests is attractive because,
unlike self-reports, behavior from objective tests cannot be easily distorted
to meet self-presentational motives. Despite the enormous energy that Cattell
and his students invested in developing objective personality tests, most of
the results from their impressive research program were rather disappointing.
Q-data and T-data converged so poorly that it became admissible to assume that
personality tests measure different personality traits than do personality
questionnaires. Since Cattell’s research program on objective personality
testing came to a halt, much progress has been made in understanding the many
explanations for the low consistencies among various indicators of the same
construct (see above). Given that the consistency issue from objective
personality research is identical in principle with other types of
consistencies, such as implicit-explicit consistency, it seems worthwhile to
apply our model to the domain of objective personality tests. This is easily
possible if we substitute “objective personality test behavior” for “behavior”
in our model (Gschwendner, Hofmann, & Schmitt, 2006b). We are currently
planning a series of studies intended to test whether the proposed application
of our model to objective personality tests will help in understanding the
reasons for the low convergence between Q-data and T-data in previous research.
We speculate that the systematic inclusion of moderator variables will teach us
some of these reasons. Risk propensity is the first behavioral domain we want
to look at.
References
Ajzen,
Ajzen, I. T., & Fishbein, M. (1977). Attitude-behavior relations: A theoretical
analysis and review of empirical research. Psychological
Bulletin, 84, 888-918
Asendorpf, J.
B., Banse, R., & Mücke, D. (2002). Double dissociation between implicit and
explicit personality self-concept: The case of shy behavior. Journal of Personality and Social
Psychology, 83(2), 380-393.
Bandura,
A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215.
Baumeister, R.
F., & Tice, D. M. (1988). Metatraits. Journal
of Personality, 56, 571-598.
Bem, D. J.
(1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology.
(Vol. VI, pp. 1-62).
Betsch,
C. (2004). Präferenz für Intuition und Deliberation (PID): Inventar zur
Erfassung von affekt- und kognitionsbasiertem Entscheiden [Preference for
intuition and deliberation (PID): An inventory for assessing affect- and
cognitionbased decision-making]. Zeitschrift
für Differentielle und Diagnostische Psychologie, 25, 179-197.
Blair, I. V.
(2002). The malleability of automatic stereotypes and prejudice. Personality and
Social Psychology Review, 6(3), 242-261.
Castelli, L., De Dea, C., &
Nesdale, D. (2008). Learning social attitudes:
Children’s sensitività to the nonverbal behavior of adult models during
interracial interactions. Personality and
Social Psychology Bulletin, 34, 1504-1513.
Cattell, R. B. (1957). Personality and motivation structure and measurement.
Chaplin, W. F.
(1991). The next generation of moderator research in personality psychology. Journal of Personality, 59, 143-178.
Deutscher,
Dunton, B. C.,
& Fazio, R. H. (1997). An individual
difference measure of motivation to control prejudiced reactions. Personality and Social Psychology Bulletin, 23, 316-326.
Ekman,
P. & Friesen, W.V. (1978). The facial
action coding system.
Egloff, B.,
& Schmukle, S. C. (2002). Predictive validity of an implicit association
test for assessing anxiety. Journal of Personality and Social Psychology, 83, 1441-1455.
Epstein,
S., Pacini, R., Denes-Raj, V., & Heier, H. (1996). Individual differences
in intuitive-experiential and analytical-rational thinking styles. Journal
of Personality and Social Psychology, 71, 390-405.
Fenigstein, A.,
Scheier, M. F., & Buss, A. H. (1975). Public and private
self-consciousness: Assessment and theory. Journal
of Consulting and Clinical Psychology, 42, 522-527.
Fiske, D. W.,
& Rice, L. (1955). Intra-individual
response variability. Psychological
Bulletin, 52, 217-250.
Friese,
M., Hofmann, W. & Schmitt, M. (in press). When and why do implicit
measures predict behavior? Empirical evidence for the moderating role of
opportunity, motivation, and process reliance. European Review of Social Psychology,
Gawronski, B., & Bodenhausen,
G. V. (2006). Associative and propositional processes in evaluation: An
integrative review of implicit and explicit attitude change. Psychological Bulletin, 132, 692-731.
Greenwald, A. G.,
McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The implicit
association test. Journal of
Personality and Social Psychology, 74, 1464-1480.
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 predictive
validity. Journal of Personality and
Social Psychology.
Gschwendner,
T., Hofmann, W., & Schmitt, M. (2006a). Moderatoren
der Konsistenz implizit und explizit erfasster Einstellungen und Persönlichkeitsmerkmale.
[Moderators of the consistency of implicitly and
explicitly assessed attitudes and personality traits]. Psychologische Rundschau, 57, 13-33.
Gschwendner, T.,
Hofmann, W., & Schmitt, M. (2006b). Ein Modell des Zusammenhangs impliziter
Reaktionszeitverfahren mit Fragebogenmaßen. In R. T. Proyer, T. M. Ortner &
K. D. Kubinger (Eds.), Theorie und Praxis
Objektiver Persönlichkeitstests (pp. 69-86). Bern: Verlag Hans Huber.
Gschwendner, T.,
Hofmann, W., & Schmitt, M. (2006c). Synergistic moderator effects of situation and person factors of
awareness and adjustment on the consistency of implicit and explicit attitudes. Journal of Individual Differences, 27, 47-56.
Gschwendner,
T., Hofmann, W. & Schmitt, M. (2008). Differential stability: The effects of
acute and chronic construct accessibility on the temporal stability of the
Implicit Association Test. Journal of
Individual Differences, 29, 70-79.
Gschwendner,
T., Hofmann, W. & Schmitt, M. (in press). Convergent and predictive
validity of implicit and explicit anxiety measures as a function of specificity
similarity and content similarity. European
Journal of Psychological Assessment.
Hartshorne, H.,
& May, M. A. (Eds.).
(1928). Studies
in the nature of character (Vol. I: Studies in deceit).
Hefter,
R., Manoach, D. & Barton, J. (2005). Perception
of facial expression and facial identity in subjects with social developmental
disorders. Neurology, 65, 1620-1625.
Hofmann, W.,
Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. (2005). A meta-analysis on the correlation between the
Implicit Association Test and explicit self-report measures. Personality and Social Psychology Bulletin, 31, 1369-1385.
Hofmann, W.,
Rauch, W., & Gawronski, B. (2007). And deplete us not into temptation:
Automatic attitudes, dietary restraint, and self-regulatory resources as
determinants of eating behavior. Journal
of Experimental Social Psychology, 43, 497-504.
Hofmann, W.,
Gschwendner, T., Castelli, L., & Schmitt, M. (2008). Implicit and explicit
attitudes and interracial interaction: The moderating role of situationally
available control resources. Group Processes and Intergroup Relations, 11,
69-87.
Hofmann, W.,
Gschwendner, T., Nosek, B. A., & Schmitt, M. (2005). What moderates implicit-explicit
consistency? European Review of Social
Psychology, 16, 335-390.
Hofmann, W.,
Gschwendner, T., & Schmitt, M. (2005). On implicit-explicit consistency:
The moderating role of individual differences in awareness and adjustment. European Journal of Personality, 19,
25-49.
Hofmann, W., Gschwendner, t., & Schmitt, M. (under
review). The road to the unconscious self not taken: Discrepancies between
self- and observer-inferences about implicit dispositions from nonverbal
behavioral cues
Hofmann,
W., Gschwendner, T., Wiers, R., Friese, M. & Schmitt, M. (2008). Working memory
capacity and self-regulatory behavior: Toward an individual differences
perspective on behavior determination by automatic versus controlled processes.
Journal of Personality and Social
Psychology, 95, 962-977.
Kenrick, D. T.,
& Funder, D. C. (1988). Profiting from controversy: Lessons from the
person-situation debate.
American Psychologist, 43, 23-34.
LaPiere, R. T.
(1934). Attitudes versus action. Social
Forces, 13, 230-237.
Mischel, W.
(Ed.). (1968). Personality
and assessment.New
Nosek, B. A. (2005). Moderators of the relationships between implicit and explicit
evaluation. Journal of Experimental
Psychology: General, 134, 565-584.
Nosek, B. A.
(2007). Implicit-explicit relations. Current
Directions in Psychological Science, 16, 65-69.
Petty,
R., & Krosnick, J. A. (1995). Attitude
strength. Antecedents and consequences.
Schmitt,
M. (1990). Konsistenz als Persönlichkeitseigenschaft?
Moderatorvariablen in der Persönlichkeits- und Einstellungsforschung.
Schmitt,
M. & Borkenau, P. (1992). The consistency of personality. In G.-V. Caprara
& G.L. Van Heck (Eds.), Modern personality
psychology. Critical reviews and new directions (pp. 29-55).
Schröder-Abé, M., Rudolph, A. & Schütz, A. (2007). High implicit self-esteem is not necessarily
advantageous: Discrepancies between implicit and explicit self-esteem and their
relationship with anger expression and psychological health. European Journal of Personality, 21,
319-339.
Schröder-Abé,
M., Rudolph, A., Wiesner, A. & Schütz, A. (2007). Self-Esteem Discrepancies
and Defensive Reactions to Social Feedback. International
Journal of Psychology, 42, 174-183.
Smith, E. C.,
& DeCoster, J. (2000). Dual-process models in social and
cognitive psycholoy: Conceptual integration and links to underlying memory
systems. Personality and Social
Psychology Review, 4, 108-131.
Snyder, M.
(1974). The self-monitoring of expressive behavior. Journal of Personality and Social
Psychology, 30, 526-537.
Snyder, M. & Ickes, W. (1985). Personality and social behavior. In G.
Lindzey & E. Aronson (Eds.), Handbook
of social psychology (3rd ed., Vol. II, pp. 883-947). New York: Random House.
Swann, W. B., & Seyle, C. (2005). Personality Psychology’s comeback and its
emerging symbiosis with social psychology. Personality
and Social Psychology Bulletin, 31, 155-165.
Strack, F.,
& Deutsch, R. (2004).
Reflective and impulsive determinants of social
behavior. Personality and Social
Psychology Review, 8(3), 220-247.
Wicker, A. W.
(1969). Attitudes versus actions: The relationship of verbal
and overt behavioral responses to attitude objects. Journal of Social Issues, 25, 41-78.
Wittenbrink,
B., Judd, C. M., Park, B. (2001). Spontaneous prejudice at the implicit level
and its relationship between questionnaire measures. Journal of Personality and Social
Psychology, 72, 262-274.