Implicit Social Influence

Article (PDF Available) · January 2007with 456 Reads
Cite this publication
Abstract
Thesis (PhD) - Indiana University, Psychology, 2007 Previous research has shown that people hold two kinds of attitudes, explicit attitudes, which are voluntary evaluations of things, and implicit attitudes, which are automatic evaluations that occur spontaneously and are difficult or impossible to control. Prior work has shown that social influence, whether it is intentional persuasion or incidental influence, usually leads the recipient of the influence to change his or her attitudes to be closer to the attitudes of the source of the influence. This work has focused on the effect of the explicit attitudes of the source of influence but ignored the possible effect of the source's implicit attitudes. Three studies examine the independent effect of the source's implicit attitude on a recipient in different social influence settings. In the first study, the implicit and explicit attitudes of a source towards a target were measured, and in the second two studies the implicit and explicit attitudes of the source were manipulated. In the first study, the recipient watched the source give a persuasive message about the target, in the second study the source described the target directly to the recipient, and in the third study, the recipient watched the source interacting with the target. Results revealed that implicit attitudes have an influence on a recipient, but in unexpected ways. In the first study, the sources' implicit attitudes led to a contrast effect on the recipients' explicit attitudes. In the second and third study the manipulation of the sources' attitudes did not work as expected, and the influence of the sources' implicit attitudes on the recipient was not detected. Thus, a person's implicit attitudes can influence another person's attitudes, but they must be strong and possibly naturally occurring. The conditions in which implicit attitudes lead to influence deserve further research.
Figures - uploaded by Winter Mason
Author content
All content in this area was uploaded by Winter Mason
Content may be subject to copyright.
No caption available
… 
No caption available
… 
No caption available
… 
No caption available
… 
IMPLICIT SOCIAL INFLUENCE
Winter Mason
Submitted to the faculty of the University Graduate School
in partial fulfillment of the requirements
for the degree
Doctor of Philosophy
in the Department of Psychological and Brain Sciences
and the Department of Cognitive Science
Indiana University
August 2007
ii
Accepted by the Graduate Faculty, Indiana University, in partial fulfillment of the
requirements for the degree of Doctor of Philosophy
Doctoral Committee
Eliot R. Smith, Ph. D.
Robert L. Goldstone, Ph. D.
Zakary Tormala, Ph. D.
Olaf Sporns, Ph. D.
August 9, 2007
iii
First, I would like to thank my parents, Priscilla Campbell and Brian Mason,
for continuing to believe in me despite some rough spots and wasted tuition
money. In more ways than one hundred, I wouldn’t be here without them.
Many thanks to my dissertation committee. Despite a sudden change in topic,
they continued to encourage me towards my goals.
Eliot Smith, thank you for listening, gently correcting me when I went astray,
and proudly patting me on the back when I did something right. Your sincerity
helped me as much as your brilliance. Rob Goldstone, thank you for introducing
me to the field that has excited me most, and putting up with my slack moments
and manic moments. I imagine one day that will actually pay off! Zak Tormala,
thank you for schooling me, not just on attitudes and persuasion, but also on the
basketball court. Olaf Sporns, thank you for inspiring me with your modest
brilliance, encouraging me in my original dissertation idea, and sticking with me
when the topic leapt out of your domain.
I also want to especially thank Sarah Queller, who brought me into the world
of social psychology and showed me the good science there.
I would like to thank the members of my lab (past and future): Charlie Seger,
Riki Conrey, Elizabeth Collins, Austin Chapman, and Elise Percy Hall, for their
helpful questions and comments on the work and the manuscript. A big thanks to
Riki for giving it to me straight; her wit and wisdom (and dirt) were often my best
counsel. I’d like to give a special thanks to Charlie and Elizabeth for actually
reading the whole thing, and for their invaluable feedback. Charlie, even though I
won’t be around, I’ll be happy to return the favor!
iv
I would like to thank my two closest friends in Indiana, Joe Anderson and
Eddie Fernandez, for helping me stay focused when I needed to be and well
entertained when I didn’t. Thanks also to Shawn Alexander, for being a true
friend, and for reminding me that what I was doing wasn’t really working.
And finally, I want to thank pancakes, for being warm and sweet and keeping
me happy every day while writing this dissertation.
v
Winter Mason
Implicit Social Influence
Previous research has shown that people hold two kinds of attitudes, explicit
attitudes, which are voluntary evaluations of things, and implicit attitudes, which
are automatic evaluations that occur spontaneously and are difficult or impossible
to control. Prior work has shown that social influence, whether it is intentional
persuasion or incidental influence, usually leads the recipient of the influence to
change his or her attitudes to be closer to the attitudes of the source of the
influence. This work has focused on the effect of the explicit attitudes of the
source of influence but ignored the possible effect of the source’s implicit
attitudes. Three studies examine the independent effect of the source’s implicit
attitude on a recipient in different social influence settings
In the first study, the implicit and explicit attitudes of a source towards a
target were measured, and in the second two studies the implicit and explicit
attitudes of the source were manipulated. In the first study, the recipient watched
the source give a persuasive message about the target, in the second study the
source described the target directly to the recipient, and in the third study, the
recipient watched the source interacting with the target.
Results revealed that implicit attitudes have an influence on a recipient, but in
unexpected ways. In the first study, the sources’ implicit attitudes led to a contrast
effect on the recipients’ explicit attitudes. In the second and third study the
vi
manipulation of the sources’ attitudes did not work as expected, and the influence
of the sources’ implicit attitudes on the recipient was not detected. Thus, a
person’s implicit attitudes can influence another person’s attitudes, but they must
be strong and possibly naturally occurring. The conditions in which implicit
attitudes lead to influence deserve further research.
_______________________________________
_______________________________________
_______________________________________
_______________________________________
vii
Table of Contents
Section Page
Title Page i
Signature Page ii
Acknowledgments iii
Abstract v
Table of Contents vii
List of Figures ix
List of Appendices x
I. Implicit Social Influence 1
a. What are Attitudes? 2
b. What are Implicit Attitudes? 4
c. Measurements of Implicit Attitudes 6
d. Implicit and Explicit Attitudes are Influenced by Different Sources 10
e. Different Effects of Implicit and Explicit Attitudes on Behavior 14
f. Summary: Implicit and Explicit Attitudes are Dissociable 17
g. Social Influence from Nonverbal Behavior 18
h. Social Influence from Implicit Attitudes 21
II. Study 1 23
a. Methods 24
i. Phase I Participants & Procedure 24
ii. Phase II Participants & Procedure 26
b. Results 28
i. Distribution of Sources’ Attitudes 29
ii. Distribution of Recipients’ Initial Attitudes 29
iii. Recipient Ratings of Source and Message 30
iv. The Influence of the Source on the Recipients’ Attitudes 31
c. Discussion 34
III. Study 2 36
a. Methods 37
i. Participants 37
ii. Procedure 37
iii. Implicit Association Test 39
b. Results 41
i. Manipulation of Sources’ Implicit and Explicit Attitudes 42
ii. Effect of Source Attitudes on Recipients’ Impression 44
c. Discussion 45
IV. Study 3 48
a. Methods 49
i. Phase I Participants & Procedure 49
ii. Phase II Participants & Procedure 52
b. Results 52
c. Sources’ Awareness of Primes 52
i. Manipulation of Sources’ Implicit and Explicit Attitudes 53
ii. Comparing Sources’ Attitudes in Study 2 and 3 55
viii
Table of Contents
Section Page
IV. c.
iii. Effect of Source on Target 56
iv. Effect of Source Attitudes on Recipients’ Impression 57
d. Discussion 59
V. General Discussion 60
VI. References 71
ix
List of Figures
Figure 1. Pictorial representation of the affective misattribution procedure
Figure 2. Scatterplot of the a) sources’ and b) recipients’ explicit (abscissa) and
implicit (ordinate) attitudes, Study 1.
Figure 3. Change in recipients’ explicit attitudes as a function of the direction of
the source’s speech,
Figure 4. Change in recipients’ implicit attitudes as a function of the sources’
implicit attitudes, Study 1.
Figure 5. Sources’ a) explicit and b) implicit attitudes as a function of the
behaviors and primes in the impression formation task, Study 2.
Figure 6. Scatterplot of the a) sources’ and b) recipients’ explicit (abscissa) and
implicit (ordinate) attitudes, Study 2.
Figure 7. Recipients’ explicit attitude as a function of the sources’ implicit
attitudes and (median split) explicit attitudes, Study 2.
Figure 8. Recipients’ implicit attitudes as a function of the source’s (median-split)
explicit attitudes, Study 2.
Figure 9. A still from a video watched by recipients in Phase II of Study 3.
Figure 10. Sources’ a) explicit and b) implicit attitudes as a function of the
behaviors and primes in the impression formation task, Study 3.
Figure 11. Effect of primes on sources’ implicit attitudes in Study 2 and Study 3.
Figure 12. Scatterplot of the a) sources’ and b) recipients’ explicit (abscissa) and
implicit (ordinate) attitudes, Study 3.
Figure 13. Recipients’ implicit attitudes as a function of the sources’ (median
split) explicit attitudes.
x
List of Appendices
Appendix A. Explicit attitude measure from Study 1.
Appendix B. Speeches read by sources in Study 1.
Appendix C. Explicit attitude measure from Study 2
Appendix D. Behaviors used to form impression of target in Study 3
Appendix E. Explicit attitude measure from Study 3.
1
Implicit Social Influence
Social influence is a powerful force in human psychology. The knowledge,
beliefs, attitudes, opinions, emotions, and behaviors of other people shape the way
an individual perceives, understands, and acts on the world. This influence can
come in the form of direct persuasion (Petty & Cacioppo, 1996), learning
(Deutsch & Gerard, 1955; Kaplan & Miller, 1987), modeling (Bandura, Ross, &
Ross, 1963), conformity (Asch, 1951), obedience to authority (Milgram, 1963),
unconscious mimicry (Chartrand & Bargh, 1999), peer pressure (Brown, Clasen,
& Eicher, 1986), and on and on. When seen from a broader perspective, social
influence underlies the very fabric of culture. As people learn new things,
develop new styles or fashions, or create new ways of thinking or doing things, it
is the force of social influence that causes these developments to eventually
permeate a people and then serve as the starting point for future developments.
Central to many studies of social influence and a determining factor in the spread
or stagnation of culture itself are peoples’ attitudes towards the elements of
culture – innovations, styles, food, etc. As revealed by the many studies of social
influence, typically when there is influence between two people it means that the
attitudes of the recipient of social influence are brought closer to the attitudes of
the source of influence.
Roughly twenty years ago, research on attitudes demonstrated that people
could automatically evaluate stimuli, and that this evaluation was not always the
same as the attitude expressed in response to explicit questioning. Since then,
there has been a surge of interest in the distinction between implicit and explicit
2
attitudes. As a result, evidence has accumulated showing that automatic
evaluations and deliberative evaluations can be dissociated in how they are
shaped, how they change, and how they guide behavior. One implication of this
research is that a person’s automatic evaluation and deliberative evaluation can
influence behavior simultaneously and sometimes incongruently. However, no
work to date has shown whether implicit and explicit attitudes can have
independent effects on other people. The three studies reported in this manuscript
seek to demonstrate that the influences of a person’s automatic and deliberative
attitudes on another person are also dissociable.
This introduction first introduces the historical struggle to define attitudes,
especially the difference between implicit and explicit attitudes, reviews evidence
supporting the dissociation of these attitudes in measurement and influence on
behavior, reviews work on the influence of nonverbal behaviors on attitudes, and
finally ties it all together to show the literature supports the hypothesis that
implicit attitudes can have an independent effect on others’ attitudes.
What Are Attitudes?
Defining implicit attitudes and how they differ from explicit attitudes is
regrettably not as simple as one would like. The meaning of implicit attitudes has
been the center of a debate amongst attitude researchers that has brought the very
construct of ‘attitude’ under scrutiny (Gawronski, in press). Therefore, to
elaborate on the distinction between implicit and explicit attitudes, it is necessary
to briefly review the various definitions of “attitudes” that have been used in the
literature.
3
Thurstone (1928) said an attitude is, “the sum total of a man’s inclinations and
feelings, prejudice or bias, preconceived notions, ideas, fears, threats, and
convictions about any specified topic.” More recently, Eagly and Chaiken (1993)
defined an attitude as “a psychological tendency that is expressed by evaluating a
particular entity with some degree of favor or disfavor” (Eagly & Chaiken, 1993,
p. 1). Revealing the influence of cognitive psychology, Fazio, Chen, McDonel,
and Sherman (1982) defined an attitude as an association between an object and
an evaluation.
While there are some common elements in these definitions, there is still
disagreement about the best way to think of attitudes, including how an attitude is
represented in memory (Abelson & Prentice, 1989; E. R. Smith & Conrey, 2007),
whether attitudes are necessarily bipolar (Osgood, Suci, & Tannenbaum, 1957) or
composed of separable positive and negative components (Cacioppo, Gardner, &
Berntson, 1997), whether there are multiple types or bases of attitudes (Breckler,
1984; Zanna & Rempel, 1988), and whether attitudes are enduring dispositions
(Ajzen & Fishbein, 1977; Bem, 1970; Cialdini, Petty, & Cacioppo, 1981) or
constructed with each elicitation (Schwarz & Bohner, 2001; Wilson & Hodges,
1992). Despite these remaining ambiguities in the definition of attitudes, some of
which were precipitated by the struggle to develop a theoretical account of
implicit and explicit attitudes, it is possible to extract a general definition that
seems well agreed upon: an attitude is an evaluative response to a stimulus.
What Are Implicit Attitudes?
4
Indirect measurements of attitudes have been in use in social psychology for
over half a century (Campbell, 1950). For most of that time, their utility has been
to avoid reactivity or self-presentation effects, and the assumption was that the
indirect measurements were tapping a cleaner version of the same construct as the
direct measurements. However, it was recognized that some evaluations were
occurring faster than possible with deliberative thought (Fazio, Sanbonmatsu,
Powell, & Kardes, 1986), and that the elicitation of an attitude could occur with
the subliminal presentation of a stimulus (Greenwald, Klinger, & Liu, 1989).
This led some researchers to hypothesize the existence of implicit attitudes
(Greenwald & Banaji, 1995), evaluations that could be activated automatically
and / or outside of conscious awareness.
This implication, that there were unconscious attitudes that could be different
from the conscious attitudes, stirred up theoretical and empirical research on the
topic (Fazio, in press), and is still hotly debated. For instance, Schwarz and
colleagues (Schwarz, in press; Schwarz & Bohner, 2001) assert that since all
attitudes are constructed spontaneously, the empirical evidence for implicit and
explicit attitudes is simply the result of different contextually sensitive attitude
constructions, rather than evidence for two separate but stable attitude
representations. Wilson, Lindsey, and Schooler (2000) suggest that implicit
attitudes are well-learned evaluative responses to a stimulus, while explicit
attitudes (when they differ) are recently learned responses that have not been
practiced sufficiently to become automatic. Fazio and colleagues (Fazio, 1990;
Fazio & Towles-Schwen, 1999) argue that there are two distinct processes, one
5
automatic and one controlled, that lead to differences in the attitude-behavior
relation, and thus the differences in implicit and explicit measures. Gawronski
and Bodenhausen (2006) suggest that the explicit attitude is contained within a
propositional framework (Strack & Deutsch, 2004), so the evaluation of the object
can be compared with other beliefs and tested for veracity, whereas the implicit
attitude is contained in an associative framework, subject to the underlying
structure and rules of pattern activation (McClelland & Rumelhart, 1985; E. R.
Smith, 1996; E. R. Smith & Conrey, 2007; E. R. Smith & DeCoster, 2000). For
instance, one could implicitly associate good feelings with fatty foods, which is
merely an association and has no truth or falsehood to it, whereas the explicit
attitude that “fatty foods are good,” can be checked and corrected with the
knowledge that fatty foods lead to heart disease and obesity, which are bad.
Nevertheless, the general agreement about how to appropriately conceptualize
implicit attitudes is reaching consensus. First, as already mentioned, the
responses that are elicited with very fast presentation or in situations with little
possibility for conscious control can be different than those given when there is
the opportunity to reflect. This is most strongly and reliably evident in the
different responses to implicit and explicit attitude measures. Second, there is
evidence that different sources of influence affect implicit and explicit attitudes
differentially. Third, and finally, implicit and explicit attitudes seem to guide
different kinds of behavior in different contexts.
In the following three sections, this introduction reviews the history and
development of implicit attitude measures, and especially how they differ from
6
explicit attitude measures, then describes research showing that implicit and
explicit attitudes can be influenced by different sources, and reviews work
showing that implicit and explicit attitudes can have different effects on behavior,
concluding that implicit and explicit attitudes are indeed dissociable.
Measurements of Implicit Attitudes
In 1928, Thurstone’s manuscript, “Attitudes Can Be Measured,” marked the
beginning of an era in attitudes research. Thurstone developed a scale that
respondents could use to indicate their preference, and argued that the responses
on this scale were an adequate indicator of the person’s underlying attitude. Likert
(1932) and Osgood et al. (1957) developed similar scales. All of these were not
much different than opinion surveys that simply and straightforwardly asked
participants to state their preference with respect to the attitude object. The
research that accompanied the creation and testing of these scales showed that
there was a reliable correlation between the expressed attitude and the person’s
behavior towards the attitude object. However, this correspondence did not
always exist (LaPiere, 1934; Wicker, 1969), which Thurstone recognized but
attributed to, at best, self-presentation concerns, and at worst, plain dishonesty.
Thus, it was apparent that it was desirable to develop indirect measures of
attitudes, to avoid the problems with self-report. As mentioned, most of the early
indirect measures of attitudes were designed to circumvent these self-presentation
and honesty concerns, such as the now-famous “bogus pipeline,” (Jones & Sigall,
1971) in which participants were persuaded to believe they were attached to what
was essentially a lie detector to encourage fidelity in the responses. Jones and
7
Sigall focused their method on socially sensitive issues, specifically attitudes
towards minority groups. However, even when responses could safely be
assumed to be honest, they still did not always correspond with later behavior.
Taking the lead from memory researchers who distinguished between
automatic and controlled processes (Schneider & Shiffrin, 1977), Fazio et al.
(1986) reasoned that if object-evaluation associations were stored in memory,
their recall could also be dissociated into automatic and controlled processes.
Indeed, they showed that priming with positive and negative words (i.e.,
cockroach or puppy) led to facilitation on the categorization of the valence of
words that were also positive or negative. Devine (1989) brought the effect into
the domain of racial attitudes, and showed that unconscious priming of a
stereotype could lead participants to utilize the stereotype in subsequent
information processing. Advances in technology allowed researchers to
subliminally present pictures rather than words, and thus was born one standard
measure of automatic attitudes, affective priming (Fazio, Jackson, Dunton, &
Williams, 1995). Fazio and colleagues interpreted the facilitation in
categorization caused by primes to be a valid and reliable measure of person’s
attitude, although they tried to remain agnostic as to whether it was tapping a
person’s “true” attitude by leaving a question mark at the end of the article’s title,
“Variability in automatic activation as an unobtrusive measure of racial attitudes:
A bona fide pipeline?”
Nonetheless, the question sparked a veritable watershed in research on
measures of automatic attitudes (Dovidio, Kawakami, Johnson, Johnson, &
8
Howard, 1997; Greenwald, McGhee, & Schwartz, 1998; Nosek, 2001; Payne,
Cheng, Govorun, & Stewart, 2005; von Hippel, Sekaquaptewa, & Vargas, 1997).
Most prominent is the Implicit Association Task (IAT; Greenwald et al., 1998),
which compares the facilitation (or inhibition) when the same response is required
for categorizing target stimuli (e.g., African-American faces) and positively
valenced words to the facilitation (or inhibition) when the same response is
required for target stimuli and negative words. If the target category is strongly
associated with positivity or negativity relative to a contrast category, there will
be a large difference in facilitation (or inhibition) between the two response-
pairings. Greenwald and colleagues (1998) firmly asserted that the processes
leading to responses on the IAT were highly efficient, uncontrollable, and
unconscious, and therefore fully automatic.
Another measure, the Affect Misattribution Paradigm (Payne et al., 2005) uses
the affect elicited by a stimulus in an indirect measure of attitudes. This
procedure relies on the affect misattribution effect in which a person is uncertain
of the source of affect and erroneously attributes the cause to something else
within awareness (Dutton & Aron, 1974; Murphy & Zajonc, 1993). In this
procedure, exemplars from a target category (e.g. Republicans) and a foil category
(e.g., trees) are briefly presented prior to a Chinese ideogram, and participants are
told to ignore the prime and evaluate the ideogram. Even if they are explicitly
told to correct for the bias of the prime, participants under-correct and attribute
some of the affect elicited by the prime to the ideogram. Thus, ideograms that
follow the target category will tend to be categorized more positively if the target
9
category elicits positive affect. Payne et al. (2005) describe their measure as an
implicit measure, but the authors remained agnostic about the automaticity of the
processes that led to responses on the measure.
Bargh (1994) outlined four criteria for a process to be automatic, what he
called the “Four Horsemen of Automaticity”: 1) increased efficiency, 2) lack of
control, 3) lack of intention, and 4) lack of awareness. It was clear that responses
on explicit measures did not have any of these qualities: the responses required
controlled, intentional deliberation within the subject’s awareness. It was much
less clear whether the processes that led to responses in the implicit attitude
measures, such as facilitation in affective priming, met these four criteria. It is
safe to say that the responses were highly efficient, since facilitation occurs very
rapidly. The evidence also indicated that the responses were uncontrollable and
unintentional (Kim, 2003; Payne et al., 2005). However, the evidence that the
responses were unconscious was mixed, at best (Bargh & Pietromonaco, 1982;
Devine, 1989; Winkielman, Berridge, & Wilbarger, 2005), and was certainly not a
requirement for the differential responses on explicit and implicit measures of
attitudes. Indeed, after Greenwald and Banaji (1995) postulated a separate,
unconscious attitude, it became increasingly unclear whether the “implicit” in the
term, “implicit attitude measure” was meant to apply to the measure (participants
were unaware that their attitude was being measured), or to the attitude
(participants were unaware of their attitude).
Thus, although there is still disagreement as to whether implicit attitudes are
necessarily unconscious – and the evidence appears to point towards the negative
10
– there are clearly times a person is unaware of their automatic attitude (Bargh &
Pietromonaco, 1982). It does, however, seem safe to say that implicit attitudes
meet the other three criteria for automaticity as defined by Bargh (1994). As
mentioned, this is in contrast to explicit attitudes (as measured by traditional
“pencil and paper” questionnaires), which meet none of these criteria. Of course,
this does not mean that explicit and implicit attitudes cannot be the same; in fact,
for many attitude objects they are highly correlated (see Nosek, 2007). However,
it does mean that they can be different, both in how they are elicited and the
contexts in which they arise.
Implicit and Explicit Attitudes Are Influenced by Different Sources
Not only are the attitudes dissociable in the attitude measures, there is
evidence that the implicit and explicit attitudes can be influenced independently
as well. For instance, some research shows that implicit attitudes can be changed
with no effect on explicit attitudes. Lowery, Hardin, and Sinclair (2001) found
that the mere presence of an African-American experimenter affected White
participants’ responses on a Black-White IAT without affecting their responses on
explicit measures such as the Modern Racism Scale (McConahay, Hardee, &
Batts, 1981), the Internal and External Motivation to Respond Without Prejudice
scales (Plant & Devine, 1998), or the Social Dominance Orientation scale (Pratto,
Sidanius, Stallworth, & Malle, 1994).
In another similar study, Dasgupta and Greenwald (2001) showed that
repeated exposure to admired Black persons and disliked White persons reduced
automatic prejudice as measured by an IAT without affecting responses on feeling
11
thermometers or semantic differential scales. Similarly, Sinclair and colleagues
(Sinclair, Lowery, Hardin, & Colangelo, 2005) showed that participants’
(negative) automatic attitudes towards Blacks was attenuated when interacting
with an experimenter wearing a t-shirt that proclaimed anti-racism ideals.
Presumably, participants’ affiliative goals led them to “socially tune” their
automatic, implicit attitudes to the expressed attitudes of the experimenter.
Using a very different methodology, Olson and Fazio (2006) showed that
evaluative conditioning could influence participant’s implicit attitudes without
affecting explicit attitudes. The participants in the experimental condition saw
photographs of Black people repeatedly paired with positively valenced words,
while the control participants saw the same photographs and words, but the two
were not consistently paired together. On a subsequent measure of implicit
attitudes, participants in the experimental condition showed a decreased
association between Blacks and negative words compared to the control
condition. There was no difference between conditions on the explicit ratings of
prejudice.
Other research showed a change in explicit measures with no change in the
implicit. Gawronski and Strack (2004) showed that cognitive dissonance
reduction changed explicit attitudes but had no effect on implicit attitudes (see
also Schooler, 1990). Using a standard cognitive dissonance induction procedure,
they had participants write a counter-attitudinal essay in either a low or high
situational pressure condition. In the low situational pressure condition,
attributing one’s behavior to external causes was much harder than the other, high
12
situational pressure condition. Relative to the control condition and the high-
pressure condition, the low-pressure condition caused the explicit attitude to
change in the direction of the essay, but did not affect the implicit attitude,
thereby reducing the correlation between implicit and explicit attitudes.
In work by Gregg, Seibt, and Banaji (2006), participants learned about two
novel groups, such that one was always described as positive and the other was
always described as negative, which was sufficient to induce congruent explicit
and implicit attitudes toward the novel groups. In their third experiment, some of
the participants were told that there was a mix-up, and the descriptors of one
group were supposed to be applied to the other group, and then asked them to
complete the attitude measures again. In this study, the explicit attitude reversed
(in accord with instructions), but the implicit attitude did not reverse, thereby
inducing conflicting implicit and explicit attitudes in the participants.
Most importantly, still other work showed that both attitudes could be
simultaneously affected in opposite directions. Deutsch and Strack (2002; as
reported in Strack & Deutsch, 2004) had participants repeatedly choose between a
red or blue “door” that had different consequences. One door led to a fast
presentation of a very negative photograph, followed by a long presentation of a
pleasant photograph. The other door had a fast presentation of a positive
photograph and a long presentation of a negative photograph. When given time to
choose between the doors, participants consistently chose the door that had the
long presentation of a pleasant photograph. However, when under extreme time
pressure, the choice reversed. Thus, the briefly presented photograph influenced
13
participants’ automatic evaluations of the doors, but the more persistent
photograph influenced participants’ explicit attitudes.
Using more conventional measures of implicit and explicit attitudes, Rydell,
McConnell, Mackie, and Strain (2006) simultaneously conditioned participants
through subliminal priming to have one implicit attitude while training them
through a categorization procedure to have the opposite explicit attitude about a
person. Later measurements of the attitudes using explicit questionnaires and the
IAT (Greenwald et al., 1998) revealed the participants’ implicit attitudes had been
influenced by the primes while their explicit attitudes had been influenced by the
categorization procedure, resulting in incongruent implicit and explicit attitudes.
All of these examples show that explicit and implicit attitudes can be
manipulated independently. This implies that either the attitudes are represented
independently or that the automatic and controlled processes that lead to
differential evaluative responses are independent. In either case, there appears to
be strong evidence that not only can implicit and explicit attitudes be elicited and
measured with different methods, but they can be changed by different means as
well.
Different Effects of Implicit and Explicit Attitudes on Behavior
Most important to the research at hand, however, is whether (and how)
implicit and explicit attitudes differentially guide behavior. Clearly, because the
responses on attitude measures are themselves behaviors, there are at least some
contexts in which implicit and explicit attitudes have dissociable effects on
behavior. However, it is useful to delineate the contexts in which these attitudes
14
influence behavior, what behaviors are guided by automatic or controlled
evaluations, and whether they can simultaneously influence behavior (when the
attitudes are dissociated).
Thirty years ago, Icek Ajzen and Martin Fishbein presented a summary of
research on the attitude-behavior relationship, beginning by noting that, “Reports
of rather low or nonsignificant relations between attitudinal predictors and
behavioral criteria have been accumulating for more than 40 years,” (Ajzen &
Fishbein, 1977, p. 1). However, they went on to outline the conditions under
which there was a correspondence between attitudes and behavior. This work
came before the implicit / explicit distinction, and indeed, the attitude measures
they focused on would be considered explicit. In their review, they suggested that
the reason explicit measures sometimes failed to capture subsequent behavior was
because the measure was aimed at a different behavior than the one being
predicted. For instance, an explicit attitude measure might show a positive
evaluation towards birth control, but then poorly predict future use. However, if
the measure instead focused on attitudes towards personally using birth control in
high arousal situations, the favorability might drop considerably and the attitude-
behavior correlation might correspondingly increase.
Early research on attitude-behavior consistency showed that the more a person
engaged in reflective processing, both when considering their attitudes (Carver &
Scheier, 1978; Snyder & Swann, 1976) and when engaging in the behavior (F. J.
Smith, 1977; but see Wilson & Schooler, 1991), the higher the correspondence
between the measured attitude and the behavior. This supports the hypothesis that
15
explicit attitudes guide behavior when there is opportunity for deliberation (Fazio
et al., 1982).
Conversely, those behaviors that are automatic and relatively uncontrollable
tend to be guided by implicit attitudes. This includes speeded reactions (Strack &
Deutsch, 2004) and behaviors that are normally spontaneous, such as nonverbal
behaviors. Chen and Bargh (1999), replicating a study by Solarz (1960) found
that participants were faster at pushing a lever away from themselves when
presented with a negative stimulus and pulling it towards themselves in response
to a positive stimulus than the opposite motor movements. The speeded reactions
required in the task indicate that the automatic evaluations were influencing their
reaction times for the different behaviors.
In a more real world example, Fazio et al. (1995) tested the predictive validity
of the affective priming procedure, in which a Black experimenter rated the
friendliness of her interaction with participants, paying special attention to
nonverbal behaviors. The experimenter’s rating was significantly correlated with
the facilitation scores from the implicit attitude measure, but was not related to the
participants’ scores on the MRS (McConahay et al., 1981), an explicit attitude
measure.
Dovidio and colleagues (Dovidio, Gaertner, Kawakami, & Hodson, 2002;
Dovidio, Kawakami, & Gaertner, 2002) also found that a priming measure
correlated with the participants’ nonverbal friendliness towards a Black
confederate, as measured by the confederates themselves, judges’ ratings of the
videotaped interaction, and judges’ ratings of the participants’ nonverbal
16
behavior. However, their verbal friendliness correlated with the explicit measure
of racial attitudes, and with the participants’ rating of their own friendliness.
This mimicked results from a previous study that measured nonverbal
behaviors more carefully. Dovidio et al. (1997) measured participants’ implicit
racial attitudes using an affective priming procedure, and measured their explicit
attitudes using the MRS and the Old-Fashioned Racism scale (ORS; McConahay
et al., 1981). They then had participants interact with a Black and White
interviewer, which they video recorded. They found that explicit attitudes, as
measured by the MRS and ORS, predicted participants’ explicit ratings of the
interviewers, such that higher scores on the explicit scales led to lower liking of
the Black relative to the White experimenter. The explicit scales did not,
however, predict the participants’ nonverbal behaviors. Instead, the lower the
facilitation scores for Whites vs. Blacks (i.e., the more automatic prejudice
towards Blacks), the less eye-contact and more blinking they had with the Black
relative to the White interviewer.
Wilson et al. (2000) reported a correlation between a priming measure of
implicit prejudice and the number of times White participants touched a Black
confederate’s hand when passing back and forth a pen. Bessenoff and Sherman
(2000) used an affective priming procedure to measure participants’ attitudes
towards fat persons, as well as an explicit measure with three subscales (Anti-fat
Questionnaire; Crandall, 1994). The implicit measure predicted the distance the
participant would place their own chair from a fat woman, which was not
predicted by any of the subscales in the questionnaire.
17
As is evident with the differential responses on explicit and implicit measures,
automatic and controlled evaluations guide different behaviors. Perhaps more
interesting, and more relevant to the research at hand, automatic evaluations can
guide nonverbal behaviors even while deliberative evaluations are guiding more
deliberative behavior such as verbal statements (Dovidio et al., 2002; Dovidio et
al., 1997; McConnell & Leibold, 2001) or responses on explicit attitude scales.
Summary: Implicit and Explicit Attitudes Are Dissociable
The work reviewed here demonstrates quite thoroughly that implicit and
explicit attitudes are dissociable. This dissociation is not with respect to
consciousness, as was originally believed, but with respect to automaticity and
control. In other words, while explicit attitudes are deliberative and therefore
necessarily within awareness, implicit attitudes are automatic in that they are
activated spontaneously, efficiently, and largely uncontrollably in response to a
stimulus. However, they may or may not be outside of a person’s awareness
(Fazio & Olson, 2003). They are dissociated on the numerous explicit and
implicit measures of attitudes (Asendorpf, Banse, & Mücke, 2002; Bessenoff &
Sherman, 2000; Campbell, 1950; Fazio et al., 1995; Fazio & Olson, 2003;
Greenwald et al., 1998; Jones & Sigall, 1971; Nosek, 2001, 2007; Payne et al.,
2005). They can be affected or changed by different means (Bargh &
Pietromonaco, 1982; Dasgupta & Greenwald, 2001; Fazio, 1990; Gregg et al.,
2006; Herr et al., 1983; Insko & Oakes, 1966; Kerpelman & Himmelfarb, 1971;
Kim, 2003; Lowery et al., 2001; Olson & Fazio, 2001, 2006; Petty, Tormala,
Briñol, & Jarvis, 2006; Rydell et al., 2006; Wilson et al., 2000). And they affect
18
different aspects of behavior (Asendorpf et al., 2002; Bessenoff & Sherman,
2000; Chen & Bargh, 1999; Cunningham, Zelazo, Packer, & Van Bavel, in press;
Devine, 1989; Dovidio, Kawakami, & Gaertner, 2002; Fazio, 1990; Fazio et al.,
1995; Gawronski & Bodenhausen, 2006; Greenwald et al., 1998; McConnell &
Leibold, 2001; Strack & Deutsch, 2004; Wilson et al., 2000).
While it is possible that explicit and implicit attitudes are separate
representations (Greenwald & Banaji, 1995), it is also possible that explicit
attitudes are simply further “downstream” from implicit attitudes (i.e., occurring
later in cognitive processing of stimuli) (Cunningham, et al., in press; Gawronski
& Bodenhausen, in press), and the different measures and manipulations merely
affect what part of the “stream” is being tapped or changed. Regardless, what is
crucial to the studies at hand is that implicit and explicit attitudes can be
independently measured and manipulated, and affect different aspects of
behavior. Specifically, it is relevant that explicit attitudes tend to guide
conscious, controllable behavior, while implicit attitudes tend to guide automatic,
spontaneous, and (often) nonverbal behavior.
The remaining two sections of this introduction review research on the
influence of nonverbal behaviors on other people and their attitudes, and ties this
to the dissociation of implicit and explicit attitudes, which leads directly into the
research reported here.
Social Influence from Nonverbal Behavior
The final key to the work at hand is that nonverbal behaviors can influence
others’ attitudes and behavior. Some of the original work on the relationship
19
between nonverbal behaviors and attitudes comes from Albert Mehrabian (1967),
who controlled the nonverbal behaviors of an experimenter towards a participant
and a confederate. He found, perhaps unsurprisingly, that the orientation of the
experimenter’s head towards the participant predicted his / her rating of how
much the experimenter liked him / her.
Indeed, most research on the effect of nonverbal behaviors on attitudes has
focused on increased rapport and liking. Scheflen (1964) noted that posture
sharing and increased rapport were correlated, and speculated that the effect could
be bidirectional. LaFrance (1979) used a cross-lagged panel technique to show
that rapport at time 1 was strongly correlated with posture sharing at time 2, but
the correlation was even stronger between posture sharing at time 1 and rapport at
time 2, lending evidence to Scheflen’s hypothesis. Using an experimental
method, Chartrand and Bargh (1999) demonstrated that when a confederate
mimicked the behavior of a target it increased liking for the confederate, and
enhanced the “smoothness” of the interaction with the participant.
Increased rapport is very important in some industries, especially service
industries, and a series of studies examined the effects of nonverbal behaviors on
customers’ behaviors. Crusco and Wetzel (1984) showed that a light touch by a
server significantly increased the tips received relative to customers not touched.
Similarly Patterson (1986) showed that interpersonal touch increased compliance
to a request for help. Applying mimicry to tipping, van Baaren, Holland,
Steenaert, and van Knippenberg (2003) found that servers imitating their
customers also led to greater tips.
20
Most relevant to the current work is the role of nonverbal behaviors in social
influence. Woodall and Burgoon (1981) found that when verbal prosody and
nonverbal behaviors were synchronized it increased the persuasiveness of the
speaker. Burgoon, Birk, and Pfau (1990) hypothesized that nonverbal and
prosodic cues can lead to the perception of credibility, which has been shown to
increase persuasiveness (Hass, 1981; Hovland, Janis, & Kelley, 1953; Kelman,
1961). They videotaped participants giving a persuasive speech, and the
participants’ nonverbal behaviors were coded. Later, a second set of participants
watched these videos, and rated the speakers on credibility and persuasiveness.
Burgoon et al. found that immediacy of the speaker, increased eye contact, facial
pleasantness and kinesic expressiveness were all correlated with persuasiveness.
With respect to actual persuasion, Ridgeway (1987) had participants watch
two confederates discuss a jury case in which a financial award had to be given to
a victim. One of the confederates had consistent nonverbal behaviors and speech
across conditions, but the other confederate employed one of four nonverbal
“styles” to accompany the same dialogue. She found that the “High-task” style,
which included moderate voice level, straight and relaxed posture, level
eyebrows, frequent eye contact with normal break-offs, and a rapid speech rate,
led participants to be influenced by the confederate: they awarded an amount
more similar to the “High-task” confederate than to the other confederate. Carli,
LaFleur, and Loeber (1995) used a similar paradigm as Ridgeway (1987), and
found that in addition to the “task” style, a “social” style – which included
moderately high eye contact, relaxed posture, friendly facial expression, and
21
moderately calm gestures – led participants to change their opinion towards the
confederate’s more than other nonverbal styles. Both of these show that
nonverbal behaviors can moderate persuasion.
Another more recent example Bailenson and Yee (2005) used immersive
virtual environments to control the behavior of the “confederate” to a degree that
is impossible with real human confederates. In this study, a participant donned
virtual reality (VR) goggles and was immersed in an environment in which it
appeared that another person was sitting opposite them at a table. The person
proceeded to deliver a persuasive message about campus security policy.
Although participants were told the other “person” was another participant
communicating through the VR environment, the “virtual confederate” was
actually a computer-generated avatar that was programmed to either mimic the
participant’s head movements with a 4 second delay, or “replay” the recorded
movements of another participant. Bailenson and Yee (2005) found that when the
avatar was mimicking the participants, they were more persuaded by the avatar’s
message than when the avatar was “replaying” the movements of another
participant.
Overall, these studies together show that nonverbal behavior carries
information that is independent from verbal information, and that these behaviors
can influence the perception of the individual as well as the persuasiveness of
their verbal statements, and can even directly affect opinion change.
Social Influence from Implicit Attitudes
22
It is evident from much research that implicit attitudes can guide nonverbal,
unintentional, subtle behavior, and that this kind of behavior has subtle influential
effects on recipients. However, despite recent compounding evidence that
implicit attitudes are dissociable and potentially independent from explicit
attitudes (Asendorpf et al., 2002; Rydell et al., 2006), no research has studied the
possibility of a person’s implicit attitude affecting others independent of her / his
explicit attitudes. One study provided correlational evidence that a person’s
implicit attitude can be related to another person’s attitudes or behavior. Chassin,
Presson, Rose, Sherman, and Prost (2002) showed that a mother’s implicit
attitudes towards smoking predicted her child’s own smoking behavior
independent of her explicit attitude towards smoking.
The literature just reviewed has shown that explicit and implicit attitudes can
be dissociable, that implicit attitudes can guide nonverbal behavior, and that one’s
nonverbal behavior can influence another’s attitudes and behavior. The research
described here seeks to show that one’s implicit attitudes can influence another
person’s attitudes independent of one’s explicit attitudes, and potentially outline
the conditions under which this could occur. Since most social influence is
assimilative (Mason, Conrey, & Smith, 2007) – the influence causes the recipient
to want to be more like the source – it is reasonable to expect the influence of a
person’s implicit attitudes will also be assimilative. Additional support for this
hypothesis comes from studies showing that assimilation occurs with the subtle or
subliminal presentation of a prime, even if the blatant presentation of the same
23
prime leads to contrast (Lombardi, Higgins, & Bargh, 1987; Strack, Schwarz,
Bless, Kübler, & Wänke, 1993).
However, it is not precisely clear whether to expect the effect of a person’s
implicit attitudes to be on another person’s explicit or implicit attitudes, or both.
It could be that the recipient of the influence would observe the source’s
behaviors and interpret their attitude correctly, consequently using that attitude as
information to determine their explicit attitude, either through a deliberative
process or a more heuristic one (Petty & Cacioppo, 1984). On the other hand, the
person might “socially tune” their attitude to the inferred attitude of the source,
thereby causing their implicit attitude to change in the direction of the source’s
implicit attitude (Sinclair, Lowery, Hardin, & Colangelo, 2005). Or there could
be a combination of effects, such that the source’s implicit attitude has an effect
on both the recipient’s explicit and implicit attitude. The exact mechanism and
pathways of influence are an open empirical question.
To this end, I present three studies. In Study 1, the existing attitudes of a
source are measured, and the independent influence of his / her explicit and
implicit attitudes are examined in a persuasive context. In Study 2, the attitudes
of the source toward an unknown person are independently manipulated, and then
the source directly communicates his / her attitude towards the target to a
recipient, and the recipient’s attitudes are subsequently measured. In Study 3, the
attitudes of the source are independently manipulated in a manner similar to Study
2, however, the recipient observes the source interacting with the target rather
than hearing the source talk about the target. Thus, in the Study 1 the attitudes of
24
the source are measured, and in the Studies 2 and 3 the attitudes of the source are
manipulated. In Studies 1 and 2 the source is talking about the target, while in
Study 3 the source is interacting with the target.
Study 1
The purpose of this study is to examine the independent effects of naturally
occurring implicit and explicit attitudes on another person’s attitudes. President
George W. Bush was chosen as the attitude object because he is likely to be
familiar to participants and evoke strong attitudes, both implicit and explicit.
Additionally, many of the participants in this study are likely to be students that
come from conservative families that support the President, but the on-campus
political atmosphere is definitively liberal, creating a norm that might contradict
the participants’ well-learned (and therefore automatic) attitudes. Therefore, it is
reasonable to expect at least some of the participants will have conflicting implicit
and explicit attitudes towards the President. The primary independent variables
were the sources’ measured attitudes. The dependent variable was the change in a
recipient’s attitudes after observing the source deliver a persuasive message about
the attitude object (President Bush).
The primary hypothesis was that the nonverbal behaviors of the speaker
would reveal his / her implicit attitudes, and either have a direct influence on the
recipient’s attitudes independent of either the source’s explicit attitudes or the
direction of the speech, or have a moderating effect on the attitude change caused
by the message content. Similar to Carli et al. (1995), the nonverbal behaviors of
25
the speaker could simply moderate the persuasiveness of the speaker, rather than
directly influence the recipients’ attitudes towards Bush.
Method
This study was conducted in two phases. Participants in the first phase had
their attitudes measured and were then videotaped delivering a pro-Bush and an
anti-Bush persuasive message. Participants in the second phase of the experiment
had their attitudes measured before and after watching one of the videos from
Phase I.
Phase I: Participants
Twenty-nine undergraduate students from Indiana University participated in
the first phase of the experiment for partial course credit. Three participants’ data
had to be dropped because they were unable or unwilling to deliver the persuasive
message. Two additional participants’ data had to be dropped due to video
recording errors, leaving 24 total participants in the first phase.
Phase I: Procedure
After agreeing to participate in the experiment, participants were informed
that this was a study about communication, and that they would complete a brief
computer-based task and answer some questions about their opinions on President
Bush.
All participants first completed a measure of their implicit attitudes toward
President Bush. To measure the participants’ implicit attitudes, we used the AMP
(Payne et al., 2005). To recap, in this measure participants are informed that they
will see a series of Chinese characters that represent different concepts. Their job
26
is to decide whether each Chinese character represents something that is good or
something that is bad. They are told that in addition, there will be another
photograph that is flashed briefly before the presentation of the Chinese character,
but that they are to ignore this picture because it is simply a warning that the
Chinese character would appear. Payne et al. showed that with this measure, even
if participants are fully aware of the purpose of the priming photographs and are
told to disregard them, participants are unable to fully correct and still tend to
attribute some of their affect from the preceding prime to the Chinese character.
For sixty trials, one of 6 photographs of President Bush, or one of 6
photographs of unknown individuals matched for appearance, posture, and
background appeared for 150 ms before being replaced by 1 of 40 randomly
selected Chinese characters. Participants then had as much time as they needed to
categorize the Chinese character as “good” or “bad” by pressing either the ‘a’ or
the ‘5’ key (see Figure 1).
After this the participants completed a measure of their explicit attitudes
toward President Bush. They were shown a series of 12 statements about George
W. Bush’s success or failure as President, and were asked to indicate their
agreement on a traditional seven-point Likert scale (see Appendix A). Half of the
statements were framed positively, and half were framed negatively, and we made
two lists of statements by reversing which questions were framed positively or
negatively. These two lists were counterbalanced between participants. After
this, they completed a “feeling thermometer” towards the President, which asks
27
them to give a number between 0 (very cold) to 100 (very warm) to indicate how
they feel about the President.
After they completed the measures of their attitudes, they were told they
would be videotaped reciting a speech about President Bush. They were given a
speech about President Bush that was either supportive or critical of his
Presidency (see Appendix B), and told they had as much time as they needed to
learn the contents of the speech, but that they would not be allowed to read from it
when they delivered the message to the video camera. Participants took from 2 -
8 minutes to learn the speech. After they said they were ready, the experimenter
started the camera and gave the participants as much time as they needed to recite
as much of the speech as they were able. The video recording ended when the
participant said they could not remember any more of the message (this portion
was edited out for later use). After this, the participants were given the other
speech (either pro- or anti-Bush), and given the same instructions. Both messages
were written to discuss similar issues and have a similar style, and the order in
which participants read the messages (pro-Bush or anti-Bush first) was
counterbalanced. The sources spoke for an average of 73 seconds (SD = 21.89),
and this was not different between speech directions. After the participants had
delivered this second speech, they were debriefed about the purpose of the
experiment and thanked for their participation.
Phase II: Participants
Seventy-two undergraduate students at Indiana University participated in the
second phase of the experiment for partial course credit. One participant had to
28
be dropped because they reported knowing the speaker, and one participant had to
be dropped because they could read the Chinese characters in the implicit
measure, leaving a total of 70 participants in the second phase.
Phase II: Procedure
After agreeing to participate in the experiment, participants were informed
that the study was about communicating opinions. They were told that they
would complete a brief computer task and then answer some questions about
President Bush. After this, they would watch a video, do another computer-based
task, and then answer more questions about President Bush. These participants
completed the same AMP and explicit questions as participants in the first phase.
After this, they were randomly assigned one of the videos made during the first
phase, and watched the source delivering either the pro-Bush or anti-Bush
message.
Once they watched the video, they rated their agreement to three statements
on a seven-point Likert scale: “The speech was very supportive of the President,”
“The speaker has a very positive attitude towards the President,” and “The
speaker was very certain of his or her attitude towards the President.” Next, they
completed a second AMP and set of explicit questions so we could measure the
change in their attitudes. To encourage these participants to thoughtfully consider
their agreement with the statements in the second explicit measure, we had them
rate their agreement with the statements from the second (oppositely phrased) list
of statements. For example, if the participant first rated their agreement with the
statement, “President Bush’s presidency has been a success,” they would then rate
29
their agreement with the statement, “President Bush’s presidency has been a
failure.”
Results
The Affect Misattribution Paradigm (AMP) compares the affect elicited by
photographs of the President to the affect elicited by similar photographs of
strangers. To quantify this affective difference, the measure compares the
categorization of Chinese characters that were preceded by a photograph of the
President to those that were preceded by a photograph of someone else. The
actual AMP score is the difference in the proportion of characters categorized as
‘good’ when preceded by a photograph of the President and the proportion
categorized as ‘good’ when preceded by a photograph of someone else. Thus, the
more characters categorized as good after seeing a picture of Bush relative to the
number categorized as good after seeing a picture of someone else, the more
positive their implicit attitude towards Bush. The most favorable (pro-Bush)
AMP score is 1 and the most negative (anti-Bush) AMP score is -1.
The explicit questions that were framed negatively were reverse-coded so all
questions were scored such that 7 meant very positive attitudes towards Bush.
There were no significant differences between the two (oppositely phrased) lists
of twelve questions, so this counterbalancing variable was left out of all further
analyses. The twelve explicit questions were found to be very internally
consistent (Cronbach’s alpha = 0.95), and a principle component factor analysis
revealed all twelve questions loaded on a single component that accounted for
30
69.41% of the variance, so they were averaged together to yield a single measure
of the source’s explicit attitude towards Bush.
Distribution of the sources’ attitudes
The distribution of the sources’ implicit and explicit attitudes towards Bush is
shown in Figure 2a. The implicit attitudes are highly clustered around the neutral
point (0), but there were some participants who had very strong positive or
negative implicit attitudes towards President Bush. The average AMP score was -
0.042 (SD = 0.44), which was not significantly different from 0.
The sources had a wide range of explicit attitudes towards Bush, with a mean
of 3.53 (SD = 1.56), just slightly negative towards Bush, which was also not
significantly different from the neutral point (4). It is easily seen from Figure 2a
that participants’ implicit and explicit attitudes towards President Bush are not
perfectly congruent, and in fact the correlation between the measures was only
0.37, which was not significantly different from 0. However, there were no
participants that had strong implicit attitudes and strong explicit attitudes that
were incongruent.
The average rating of Bush on the feeling thermometer was 43.67 (SD = 28.5)
(slightly but not significantly cool towards Bush), and this measure was
moderately correlated with the sources’ AMP (r = 0.421, p < 0.05) and highly
correlated with the sources’ explicit rating of Bush (r = 0.872, p < 0.001).
Distribution of recipients’ initial attitudes
Again, the explicit questions were highly internally consistent (Cronbach’s
alpha = 0.96), and loaded on a single factor that accounted for 69.01% of the
31
variance, so the questions were averaged to create a single measure of explicit
attitudes towards Bush. The distribution of the recipients’ implicit and explicit
attitudes towards Bush is shown in Figure 2b. As can be seen, the recipients had a
wide range of implicit and explicit attitudes towards Bush. The average AMP
score was -0.03 (SD = 0.52), and the average explicit attitude was 3.79 (SD = 1.7),
comparable to the values of the sources’ implicit and explicit attitudes. The
average feeling thermometer was 45.93 (SD = 32.11). The random assignment of
participants to pro- or anti-Bush speeches and to source videos was effective:
there were no significant differences in the initial attitudes of the recipients in
either condition, and there were no significant correlations between the sources’
attitudes and the recipients’ initial attitudes. The recipients’ implicit and explicit
attitudes were significantly correlated, however (r = 0.707, p < 0.001). The
feeling thermometer was also significantly correlated with the recipients’ initial
implicit (r = 0.671, p < 0.001) and explicit (r = 0.921, p < 0.001) attitudes. It was
not necessary for the recipients to have congruent or incongruent attitudes, but it
can be seen in Figure 2b that incongruent implicit and explicit attitudes sometimes
existed.
Recipient ratings of the source and message
After watching the video of the source delivering the pro- or anti-Bush
message, they were asked to rate how positive the speech was towards Bush, how
positive the speaker was towards Bush, and how certain the speaker seemed to be
about his / her attitude. The recipients’ rating for each of these questions was
subjected to an ANCOVA, with the direction of speech as a fixed factor and the
32
speakers’ implicit and explicit attitudes and thermometer ratings as covariates.
For the rating of the positivity of the speech, the only significant factor was the
direction of the speech (F(1,64) = 552.831, p < 0.001), which simply indicates the
recipients understood the meaning of the speech. For the rating of the positivity
of the speaker, however, in addition to the main effect of the direction of the
speech (F(1, 64) = 248.113, p < 0.001), there was also an effect of the source’s
implicit attitude (F(1, 64) = 5.77, p < 0.05), such that more positive implicit
attitudes led to a lower perceived positivity of the speaker. The implications of
this result will be discussed later.
The source’s implicit attitude had a similar effect on the recipients’ ratings of
the certainty of the speaker. The more positive the source’s implicit attitude
towards Bush, the less certain they seemed to the recipient, (F(1, 64) = 6.027, p <
0.05). The analysis included the source’s explicit attitude and the direction of
speech, which means that this is not an effect of the sources’ implicit attitudes
conflicting with the message they were delivering or their explicit attitudes.
Instead it means that, ceteris paribus, the more negative the source was implicitly
towards Bush, the more certain they seemed to the recipient.
The influence of the source on the recipients’ attitudes.
The main hypothesis was that the implicit attitude of the source would have an
assimilative effect on the recipients’ attitudes independent of the effect of the
sources’ explicit attitudes or the content of the speech. The influence of the
source on the recipient is measured by the change in the recipients’ attitudes after
watching the video of one of the sources. Again, these differences were subjected
33
to an ANCOVA with the direction of speech as a fixed factor and the speakers’
implicit and explicit attitudes and thermometer ratings as covariates.
The change in the recipients’ AMP scores was normally distributed with a
mean of -0.013 (SD = 0.33). There were no significant predictors of the change in
the recipients’ implicit attitudes, implying that there is no clear cause for the
participants’ change in attitudes beyond noise in the measure.
The change in the recipients’ explicit scores was also normally distributed
with a mean of -0.057 (SD = 0.552). Again, there were no significant predictors
of the change in explicit attitude, although the direction of the speech trended in
the expected direction, such that pro-Bush speeches led to positive change and
anti-Bush speeches led to negative change (see Figure 3).
Six of the recipients failed to complete the second thermometer rating, so they
were excluded from the relevant analysis. The change in the recipient’s
thermometer was normally distributed with a mean of -0.936 (SD = 5.11). In this
case, the sources’ own thermometer rating had a marginally significant effect on
the recipients ratings (F(1, 58) = 2.826, p < 0.1), such that the warmer the source
felt towards the President, the more positive the change in the recipients’
thermometer rating.
Contrary to predictions, the implicit attitude of the source did not seem to
influence the change in the recipients’ attitudes. There were a large number of
sources with relatively neutral implicit attitudes, so it could be of interest to focus
the analysis only to those sources with non-neutral implicit attitudes. Because
AMP scores are symmetric around 0, the sources with an absolute AMP score less
34
than 0.2 were excluded. In this way the variability in the implicit attitudes is
retained while removing the large portion (N = 14) of sources with essentially no
meaningful implicit attitude toward Bush. The remaining 10 source participants
had an average AMP score of -0.057 (SD = 0.69), an average explicit rating of
3.25 (SD = 1.41) and an average thermometer rating of 36.2 (SD = 26.73), none of
which were significantly different from the neutral points on their respective
scales.
There were 25 recipients who watched the videos of these 10 sources with
non-neutral implicit attitudes, 13 in the pro-Bush condition and 12 in the anti-
Bush condition. The analysis of the ratings of the sources was similar to the
analysis with all of the sources. The only significant effect on the rating of the
positivity of the speech was the direction of the speech, F(1, 20) = 165.407, p <
0.001. The direction of the speech also significantly affected the rating of the
positivity of the speaker (F(1, 20) = 119.577, p < 0.001), and there was a reverse
effect of the source’s implicit attitude (F(1, 20) = 4.22, p = 0.05); the more
negative the sources’ implicit attitudes, the more positive the recipient rated the
speaker’s attitude towards Bush. And finally, the source’s implicit attitude was
the only (marginally) reliable predictor of the rating of the certainty of the speaker
(F(1, 20) = 3.656, p < 0.08), such that more positive the speaker’s implicit
attitudes towards Bush, the less certain he or she seemed to the recipient.
Thus, even with the reduced number of sources, the basic results of the first
part of the analysis with all of the sources are replicated. When looking at the
change in recipients’ attitudes, however, there is an interesting difference. While
35
there is still no significant effect of the source on the change in the recipients’
implicit attitudes, when focusing only on those sources with strong implicit
attitudes there is a significant effect of those implicit attitudes on the recipients’
explicit attitudes. As can be seen in Figure 4, there is a standard persuasion effect
that is marginally significant, (F(1, 20) = 3.203, p < 0.10), such that pro-Bush
messages tend to raise the explicit rating of Bush and vice-versa. However, there
is also a distinct, significant negative trend (F(1, 20) = 6.149, p < 0.05) such that
sources with more positive implicit attitudes towards Bush had a more negative
effect on the recipients’ explicit ratings. When including the two-way interaction
between the sources’ implicit and explicit attitudes in the model, there are no
significant predictors of the change in the recipients’ explicit attitudes, implicit
attitudes, or thermometer ratings.
Discussion
In this study, the pre-existing attitudes of the source participants towards the
President were measured, and they delivered one persuasive message in favor and
one opposed to the President. The effect of this message on recipients’ attitudes
was measured, with the prediction that the implicit attitudes of the source would
have an effect on the recipients’ attitudes independent from the source’s explicit
attitudes or the content of the persuasive message. The attitudes of the sources
varied widely and their implicit and explicit attitudes were only marginally
correlated. There were many sources with relatively neutral implicit attitudes, but
there were a few that had strong explicit and neutral implicit, or strong implicit
and neutral explicit attitudes. However, there were no sources with conflicting
36
strong implicit and explicit attitudes, so the strongest test of the hypothesis was
not possible.
Unfortunately, there seemed to be no systematic change in either the
recipients’ explicit or implicit attitudes. There seemed to be no effect of the
content of the message, the source’s explicit attitude, the source’s implicit
attitude, or the source’s feeling about Bush (as measured by the thermometer) on
the recipients’ implicit or explicit attitudes. There did seem to be a small effect of
the source’s thermometer rating on the change in the recipients’ thermometer
rating, but overall it seemed that the main hypothesis was not confirmed.
Although the recipients’ attitudes did not seem to be affected by the source’s
implicit attitudes, there was evidence that they could detect the implicit attitude.
The recipients rated the positivity of the message, the positivity of the source, and
the certainty of the source. The implicit attitude of the source seems to have
affected the perceptions of the positivity and certainty of the source such that
more positive attitudes towards the source led to lower perceptions of positivity
and lower perceptions of certainty. These effects were independent of the
source’s explicit attitudes or the message content. Thus, the source’s implicit
attitude seems to have had an opposite effect on the perceptions of their attitude.
Additionally, when focusing only on those sources with non-neutral implicit
attitudes, the sources’ implicit attitudes predicted a change in the recipients’
explicit attitudes, independent of the sources’ explicit attitudes. Again, this effect
was opposite the expected direction, so those with more positive implicit attitudes
towards Bush caused the recipients of their messages to explicitly become more
37
negative towards Bush. This corresponds with their perception of the source’s
attitude, which was also negatively correlated with the sources’ implicit attitudes.
It is unclear why the sources’ implicit attitudes had a reverse effect on the
recipients. As can be seen in Figure 4, there is no interaction between the
source’s implicit attitudes and the direction of the speech, so the reverse change in
the recipients’ attitudes is not caused by sources delivering a speech that
contradicted their implicit attitudes.
Since the source’s attitudes were measured rather than manipulated, this is a
correlational study and it is possible that people who are pro-Bush differ from
those that are anti-Bush, such that their nonverbal behaviors communicate
different things. For instance, persons who hold anti-Bush attitudes may be more
authoritarian and therefore seem more certain of their attitudes. Similarly, some
variable that was not measured could be correlated with positive or negative
attitudes towards Bush and cause recipients to interpret their behavior opposite
from their true attitude, so the speakers are perceived to be more positive when in
fact they are more negative, subsequently affecting the persuasiveness of the
speech.
To gain greater control over these possible confounding variables, Study 2 is
an experimental design with the attitudes of the source manipulated rather than
measured.
Study 2
In order to have greater control over the sources’ attitudes, they had to be
about a novel attitude object – a fictitious person named “Bob” – and therefore
38
were not as well-established as naturally occurring attitudes, but the recipients’
attitudes were entirely based on the sources’ communication. This design also
allowed experimental control over the attitudes of the source, ensuring that their
implicit and explicit attitudes were truly orthogonal. In addition, rather than
watching recording the of source’s impression of the target, the recipient directly
interacted with the source. While this may have created more variability in the
source’s communication, it also allowed more information to be conveyed
through nonverbal communication.
The main hypothesis of this study is that the source will communicate her / his
implicit attitudes through nonverbal behaviors, independent of the verbal
communication to the recipient. This could be conveyed through facial
expressions, body language, or the manner in which she / he describes Bob to the
recipient. The source’s implicit attitude could directly affect the recipients’
attitudes towards Bob, or could moderate the effect of the source’s explicit
attitude.
Methods
Participants
Forty-seven pairs of undergraduate students from Indiana University
participated in this study for partial course credit, for a total of 94 participants.
Procedure
Participants were randomly assigned to the role of the source or the recipient.
After receiving basic information about the purpose of the experiment, the
39
participant assigned to the role of the recipient was asked to wait in a private
cubicle until their part of the experiment began.
The participant assigned to the role of the source was told that they would be
forming an impression of a person named ‘Bob’, and after answering some
questions about their impression and doing a short computer task, they would then
communicate their impression to the other participant, and that this discussion
would be video taped. They were then sat in front of a computer where they
began the task of forming an impression about ‘Bob’.
The purpose of this impression formation task was to independently
manipulate the source’s implicit and explicit attitudes towards the fictional person
‘Bob’. To accomplish this, we used the paradigm developed by Rydell et al.
(2006). In this paradigm, participants were presented with a picture of Bob and a
corresponding behavior, which they were asked to categorize as characteristic or
uncharacteristic of Bob. Half of the participants were given feedback that the
positive behaviors were characteristic of Bob, and the other half were given
feedback indicating the negative behaviors were characteristic of him. Unknown
to the participants, immediately prior to the presentation of the picture of Bob, an
evaluative prime was subliminally presented. Half of the participants were given
positive evaluative primes (e.g., “happy”, “sunshine”) and the other half was
given negative primes (e.g, “pain”, “guilt”). All participants were asked to
categorize the same 100 behaviors (50 positive and 50 negative), and
consequently were presented with 100 subliminal (all positive or all negative)
evaluative primes. This allowed the participants to form an explicit attitude
40
through the categorization of the behaviors, and develop an independent implicit
attitude through the classical conditioning by the primes (c.f., Olson & Fazio,
2001). Thus, when the primes and behaviors have congruent valences, the
participants’ explicit and implicit attitudes should be congruent, but when the
primes are positive but the behaviors are negative, or vice-versa, the attitudes of
the participant should be incongruent.
Once the sources formed their impressions of Bob (and their implicit attitudes
were correspondingly shaped), they were given an IAT measure of their implicit
attitudes towards Bob. The IAT was used in this study rather than the AMP
because the previous work using this paradigm used the IAT to measure the
influence of the primes.
Implicit Association Test.
In this IAT there were five blocks, only two of which were the “critical
blocks” that constitute the measure of the implicit attitude. At the beginning of
the task participants were told that we were interested in their ability to categorize
objects while distracted. Participants were informed that they would be
categorizing pictures and words, and that they should be as fast as possible while
minimizing errors.
In Block 1, participants were asked to categorize a randomly ordered
presentation of 10 positive and 10 negative words as “good” or “bad” by pressing
the ‘a’ or ‘5’ keys (respectively) on the keyboard.
In Block 2, they were instructed to categorize pictures as ‘Bob’ or ‘Not Bob’
using the same keys. For this block they were given corrective feedback after
41
their categorization response. They did this twenty times with a random
presentation of the picture of ‘Bob’ and 5 pictures of similar White males.
Block 3 was the first critical block, in which the participants had to respond to
the picture of ‘Bob’ and good words with the ‘a’ key and respond to pictures of
‘Not Bob’ and bad words with the ‘5’ key. This will be referred to as the
‘compatible’ block (although in actuality it was only compatible for half of the
participants). Participants categorized forty trials of alternating words and
pictures randomly sampled from all of the words and pictures. They were then
told that the pictures were going to be reversed, and that they would have a
chance to practice this new combination.
Block 4 was the same as Block 2, only now the keys were reversed, so
participants had to respond to pictures of ‘not Bob’ with the ‘a’ key and respond
to the picture of Bob with the ‘5’ key. As in Block 2, there were twenty trials, the
order of the pictures was randomized, and they received corrective feedback after
their categorization responses.
Block 5 was the second critical block, and was similar to Block 3, only now
pictures of ‘Not Bob’ and good words were paired with the ‘a’ key, and the
picture of ‘Bob’ and bad words were paired with the ‘5’ key. This will be
referred to as the ‘incompatible’ block. As in Block 3, participants had forty
categorization trials.
After they completed the IAT, they were asked to rate Bob on a series of six
bipolar evaluative scales, such as, “Rate Bob on a scale from very agreeable (9) to
very disagreeable (1).” This constituted the measure of the source’s explicit
42
attitude towards Bob. After this they completed a feeling thermometer towards
Bob, and then were told that this part of the experiment was over, and that they
should contact the experimenter.
When the participant did so, the experimenter informed the other participant
(assigned to the role of the recipient) that it was time for their participation. The
experimenter led both participants to a room with two chairs and a video camera.
The experimenter informed the participants that this was a study on forming and
communicating impressions, and that the job of the source was to communicate
their impression about the person they just learned about. They were told to start
by saying his name, and then describe him as though they were going to introduce
him to the other participant. The recipients were given a picture of Bob and told
that their job was to form an impression of Bob based on what the source told
them, and that afterwards they would complete a short computer task and answer
some questions about Bob. The experimenter then began recording, and left the
room (but remained within earshot).
After the source finished describing ‘Bob’, the experimenter came back into
the room, stopped the video recording, thanked the source for their participation
and debriefed them about the purpose of the experiment. The recipient was
reminded that they would complete a short computer task and answer some
questions about Bob. The recipient then completed the same IAT described
above, rated Bob on the same 6 semantic differential scales, and completed the
same feeling thermometer. When they had completed these measures, they were
thanked for their participation and debriefed about the purpose of the experiment.
43
Results
Manipulation of source’s implicit and explicit attitudes
To be sure that the manipulation of the source’s implicit and explicit attitudes
was successful, the effect of the primes and behaviors on the source’s IAT scores
and explicit ratings of ‘Bob’ was examined. The prediction was that the primes
would affect the sources’ implicit attitudes independent of the behaviors, and the
behaviors would predict the explicit ratings independent of the primes. To test
this, we conducted a 2 (primes; positive / negative) x 2 (behaviors; positive /
negative) ANOVA on the IAT scores and the explicit ratings.
The explicit ratings consisted of six 9-point semantic differential scales with
different endpoints (see Appendix C). The ratings were highly consistent
(Cronbach’s alpha = 0.983), and a principal components analysis revealed all of
the questions loaded on a single factor that accounted for 92.07% of the variance,
so a single explicit rating was created from the average of the six scales (with the
appropriate questions reverse-scored). Across all subjects in all conditions the
average rating was 4.98 (SD = 3.704), very close to (and not significantly
different from) the neutral point (5) on the scale. The distribution of explicit
attitudes was highly non-normal (in fact, highly bimodal; see Figure 5a), and the
error variance was significantly unequal across conditions (Levene’s F(3, 43) =
6.7, p < 0.001), so the analyses were done on the ranks of the average ratings.
These rank values were subjected to a 2 (behaviors; positive / negative) x 2
(primes; positive / negative) ANOVA, and the behaviors were the only significant
predictor of the explicit rating ranks, F(1, 43) = 150.629, p < 0.001. As can be
44
seen in Figure 5a, the behaviors had a very strong effect (η2 = 0.778) in the
predicted direction and there was no significant effect of the Primes.
To calculate the IAT scores, we followed the procedure suggested by
Greenwald, Nosek, and Banaji (2003). The resulting score, D, is the difference in
reaction times on the compatible and incompatible trials (Blocks 3 & 5), divided
by the standard deviation of the reaction times from both blocks. This measure is
more consistent across varying response times and does not require the exclusion
of very long or very short reaction times. Across all conditions, the average D
was -0.3 (SD = 0.432), which is slightly negative towards Bob, and this was
significantly different from the neutral point, 0, t(46) = -4.794, p < 0.001. It is
unclear why the participants formed a negative impression towards Bob, and this
will be discussed later in the manuscript.
When the sources’ IAT scores were subjected to the 2 x 2 ANOVA, the
results were surprising. Instead of the primes predicting the IAT scores
independent of the behaviors, the behaviors were predicting the scores
independent of the primes, and in the opposite direction to that predicted (see
Figure 5b). This effect was reliable but small, F(1, 43) = 6.236, p < 0.05, η2 =
0.127. It appears from Figure 5b that the primes had the predicted effect with the
positive behaviors, but this trend was not reliable.
Surprisingly, this result actually replicates the findings of Rydell et al. (2006).
In their study, they only looked at the conditions with incongruent behaviors and
primes. Exclusively looking at the incongruent conditions, it could appear with
this data that behaviors predicted explicit ratings and the primes predicted the
45
implicit attitudes. However, when looking at both congruent and incongruent
conditions, it appears that the implicit attitudes were simply influenced by the
behaviors, but opposite the expected direction. The implications of this will be
elaborated in the discussion. Fortunately, even though the manipulation did not
have the predicted effect, it was still possible to test the main hypotheses. As can
be seen in Figure 6a, there are sources with congruent and incongruent implicit
and explicit attitudes, both positive and negative, so the independent effects of the
sources’ attitudes on the recipients were examined.
Effect of Source Attitudes on Recipients’ Impression
For the two measures of the recipients’ attitudes – the explicit questions and
the IAT – there are two possible predictors: the source’s explicit attitude and the
source’s implicit attitude. As with the sources’ explicit attitudes, the recipients’
explicit attitudes were clearly bimodal (see figure 6b). Nonetheless, the average
explicit rating was 4.993 (SD = 2.99), which was not significantly different from
the neutral point (5) or the sources’ explicit attitudes.
To examine the independent effects of the source’s explicit and implicit
attitudes on the recipients’ explicit attitudes, we conducted a simple regression
with the explicit attitudes as the dependent variable and the sources’ attitudes as
the predictors. The entire model accounted for 80.5% of the variance (a fairly
good fit), but the sources’ explicit attitudes were the only significant predictor of
the recipients’ explicit attitudes, β = 0.929, p < 0.001. The impact of the sources’
explicit attitudes is evident in Figure 7, and although it appears there is a trend for
46
more positive implicit attitudes in the source to lead to more positive explicit
attitudes in the recipient, this effect was not significant.
Next, the recipients’ implicit attitudes were examined. Three of the recipients
failed to follow the instructions of the IAT and did not reverse their responses to
the target and the foil in the practice block 4 or in the critical block 5, so had to be
excluded from this analysis. The average IAT score for the remaining recipients
was -0.398 (SD = 0.437), which was significantly different from the neutral point
(0), t(43) = -6.037, p < 0.001, but not significantly different from the sources’
implicit attitudes. This general negativity corresponds with source’s implicit
negativity, which will be discussed later. To determine the effect of the sources’
attitudes on the recipients’ implicit attitudes, the same regression was conducted
on the recipients’ IAT scores as was done for their explicit ratings. In this case the
model did not fit particularly well (R2 = 0.138), and the only significant predictor
was the sources’ explicit attitudes (β = -0.374, p < 0.05) (see Figure 8), which
was negatively related to the recipients’ IAT scores. The sources with more
positive explicit attitudes towards the target negatively influenced the recipients
implicit attitudes.
Discussion
In this study, the source’s attitudes were manipulated rather than measured.
To induce congruent or incongruent explicit and implicit attitudes, we borrowed a
paradigm from Rydell et al. (2006). The manipulation seems to have exactly the
predicted effect on the explicit ratings of the target – the feedback they received
while categorizing behaviors as characteristic or uncharacteristic of the target
47
influenced their explicit attitude towards the target in the expected direction, and
the primes had no effect on the ratings. This effect was strong and very reliable,
and accords with previous work on impression formation (Asch, 1946; Hamilton
& Sherman, 1996; Kerpelman & Himmelfarb, 1971).
The manipulation of the implicit attitudes did not have the predicted effect,
however. Based on the results of Rydell et al. (2006), the primes were expected
to affect the IAT scores independent of the behaviors. However, instead it was
found that the behaviors negatively predicted the IATs, so positive behaviors led
to more negative IATs and negative behaviors led to more positive IATs. In
addition to this effect of the behaviors, there was also an overall tendency for the
sources to have a negative implicit attitude towards the target. By observing the
recordings of the sources communicating their impression of the target to the
recipient, it is possible to make inferences about why this effect may have
occurred. In both conditions, the sources described the target as impossibly
extreme, so that when they were receiving feedback that the positive behaviors
were characteristic of ‘Bob’, they described him as the perfect human being; one
participant even said, “Basically, the only person more perfect than this guy is
Jesus.” And in the negative condition one participant explained, “This guy is the
worst possible person you can imagine.”
It is possible that this extremity in the good condition caused a comparative
reaction, so participants implicitly didn’t like the target because he was too good.
This is comparable to the findings of Herr et al. (1983), who found that moderate
primes led to assimilative judgments, whereas extreme primes led to contrast.
48
Another possibility is that although participants were told that the behaviors were
merely characteristic of the target and not necessarily actual behaviors of the
target, the impression formation task may have led the participants to incorporate
all behaviors in their impression. While the individual behaviors were believable,
when put in combination they may have seemed impossible, causing participants
to form an opposite automatic evaluation.
This result has implications for the work using the manipulation because it
does not contradict the findings from the original study (Rydell et al., 2006). In
that study, they were focused on showing that incongruent implicit and explicit
attitudes could be induced in participants and that these attitudes could be
reversed. Naturally, the researchers attributed the negative implicit attitudes to
the primes, but they did not have a congruent condition as a control. It is possible
that it was the behaviors causing the incongruent explicit and implicit attitudes, as
they did in this study. One goal of Study 3 was to make the behaviors of the
target more believable when taken as a whole, and see if this effect of the
behaviors on the implicit attitudes of the source is replicated.
Despite this unpredicted effect of the manipulation on the source’s attitudes,
there was still enough variation in the sources’ attitudes to examine the effect of
the sources’ attitudes on the recipients. Unfortunately, the predicted effect of the
sources’ implicit attitude on recipients’ attitudes independent of the sources’
explicit attitudes was not found. However, the recipients seemed to have formed
their impression of the target in the same way as the sources: sources with
positive explicit attitudes towards the target induced a negative implicit attitude in
49
the recipient, just like positive behaviors induced a negative implicit attitude in
the source. Additionally, there was an overall tendency for the recipients to have
a negative implicit attitude towards the target, just like the sources’ implicit
attitudes.
A likely explanation for this is that the source communicated their explicit
impression of the target so faithfully that the recipients developed the same
distribution of implicit and explicit attitudes. This implies that it must have been
the content of the behaviors that induced the negative implicit attitudes and
caused the positive explicit impression to negatively affect the implicit attitudes
of the recipient. In Study 3, we reduced the extremity and increased the
believability of the overall impression by making the whole set of behaviors
consistent and possible to have come from a single person in the hopes that this
would eliminate the effect of the behaviors / explicit attitudes on the implicit
attitudes of the person forming the impression.
Study 3
In Studies 1 and 2, the source was communicating his / her impression of the
target (i.e., Bush or ‘Bob’) to a recipient. However, it could be that a person’s
implicit attitudes are most evident in his / her behavior when he / she is
interacting with the target of the attitude. For instance, in Dovidio, Kawakami,
and Gaertner (2002) and in the original validation of the IAT (Greenwald et al.,
1998), the participant interacted with a member of the out-group, and those
interactions were correlated with his / her responses on the implicit measures.
Therefore, the goal of this study was to see if the implicit attitudes of the source
50
were revealed in an interaction with the target of the attitude, and if the behavior
guided by these implicit attitudes would influence the attitudes of an observer of
the interaction.
In addition, it was of interest whether the manipulation of implicit and explicit
attitudes used in Study 2 would result in the same, unpredicted pattern in which
the explicit categorization of behaviors incongruently affected the implicit
attitudes of the source independent of the primes. As mentioned in the discussion
of Study 2, one possible explanation for the reaction to the behaviors was the
extremity implied by all of the behaviors together. To lessen this possibility, and
to ensure that the same, college-aged person (the confederate) could have
realistically performed all of the behaviors, we selected a subset from the original
list of behaviors to use as stimuli in this study.
Methods
Phase I: Participants
Twenty-one undergraduate students from Indiana University participated in
the first phase of the experiment for partial course credit. One participants’ data
had to be dropped because he / she failed to complete the IAT.
Phase I: Procedure
After agreeing to participate in the experiment, the participants were told that
the study was about forming impressions of people, and that we were interested in
how the manner in which one learns about a person affects the way he or she
interacts with them. There were told participants were recruited from another
study in which friends and acquaintances were asked to share experiences they
51
had with these participants, and they would learn about one of these participants.
They were told we took the stories collected from the participant’s friends and
acquaintances and condensed them into short descriptions of the behaviors, and
then made another list of behaviors that we judged to be uncharacteristic of the
participant. We told them they would learn about the participant by categorizing
these behaviors as characteristic or uncharacteristic, and after answering some
questions and completing a short computer task (the IAT), they would have a
short discussion with him.
Once they heard the instructions and had any questions answered, the
participants sat in front of a computer to form an impression of ‘Matt,’ the target.
The procedure was the same as that in Study 2, but instead of the picture of ‘Bob’,
there was a picture of the confederate, and the set of behaviors was reduced from
Study 2 (see Appendix D).
After forming an impression of Matt, the source rated the confederate on the
same six semantic differential scales (except they were 7-point instead of 9-point)
and completed the same IAT as in Study 2. For the IAT in this study the single
picture of ‘Bob’ was replaced with five pictures of the confederate (including the
one used in the impression formation task). The source was then asked to identify
the 10 subliminal primes from a list of 20 positive and 20 negative words (or to
guess, if they could not identify any) as a manipulation check. They were then
told they were going to meet Matt, and led to a room with two chairs and a video
camera.
52
The confederate, who was blind to the participants’ condition, was sitting in
one of the chairs, and the experimenter thanked them both for their participation
(to maintain the cover story). The experimenter then started the camera while
explaining that they would have 3 minutes to discuss the topic of, “Things to Do
in Town.” This topic was chosen to be personally interesting to the participants
without requiring the confederate to reveal too much personal information that
might bias the participants’ reactions or contradict the impression they formed of
him. The confederate was trained to engage in the conversation in a consistent
manner that was not too friendly or unfriendly while keeping the conversation as
natural as possible.
After the conversation, the confederate was thanked and asked to stay in the
room, and the participant was taken to another room where they were given a
process debriefing in which the suspicions of the participant were gradually
probed while also informing the participant about the nature and purpose of the
experiment. During this time, the confederate rated the participant and the
interaction on six Likert scales (see Appendix E). After it was explained that
‘Matt’ was a confederate and that the behaviors they read about him were
fabricated, they were given an opportunity to talk to Matt and ask any questions
they might have. After they were done, they were thanked for their participation
and allowed to leave.
Phase II: Participants
53
Thirty-eight people from Bloomington, Indiana (mostly undergraduate
students at Indiana University) received eight dollars in compensation for their
participation in the second part of the study.
Phase II: Procedure
After they agreed to participate in the experiment, the experimenter informed
them that we were interested in how different parts of an interaction influence the
way a person forms an impression. The recipients formed an impression of the
target by watching one of the recorded interactions between the confederate and
the participant. To ensure that the impression they were forming of the
confederate was based primarily on the behavior of the participant, one half of the
frame was replaced with a picture of the confederate (the same picture used in the
impression formation procedure in the first part of the study), so the recipients
could only see the behavior of the participant (see Figure 9).
Once they had watched the video interaction, the participants completed the
same IAT and answered the same explicit questions as the participants in the first
phase of the experiment. After completing these, they were asked if they felt they
were able to form an impression of the target, and then were debriefed on the
purpose of the experiment.
Results
Sources’ Awareness of Primes
The sources were given a randomized list of 40 words and asked to circle the
10 that were primes. Therefore, 10 of the words were correct, and 10 others were
of the same valence. The probability of correctly circling 6 or more of the words
54
by chance is < 0.05, and 7 of the 20 sources did so, indicating they were able to
see the primes. However, in work using evaluative conditioning (e.g., Olson &
Fazio, 2001) the primes were clearly perceptible, so it is possible the primes
would still have had the desired effect. Except where noted, the removal of these
seven sources did not meaningfully change the results.
Manipulation of Sources’ Implicit and Explicit Attitudes
As in Study 2, the effect of the primes and behaviors on the source’s IAT
scores and explicit ratings of ‘Matt’ was examined. To replicate the results of
Rydell and colleagues (2006), the primes should independently lead to changes in
the implicit attitudes and the behaviors should independently lead to changes in
explicit attitudes. To replicate the results of Study 2, the behaviors should lead to
congruent changes in explicit attitudes and incongruent changes in implicit
attitudes. To test this, a 2 (primes; positive / negative) x 2 (behaviors; positive /
negative) ANOVA was conducted on the IAT scores and the explicit ratings.
The six 7-point semantic differential scales were again tested for internal
reliability, and they were again found to be highly consistent (Cronbach’s alpha =
0.971), and a principal component analysis revealed all of the questions loaded on
a single factor that accounted for 87.62% of the variance, so a single explicit
rating was created from the average of the six scales (with appropriate questions
reverse-scored). The average rating across all subjects was 4.175 (SD = 2.373),
slightly but not significantly negative. The error variances were significantly
different across conditions (Levene’s F(3,16) = 11.564, p < 0.001), so as before,
the analysis was done on the ranks of the explicit ratings. The 2 x 2 ANOVA
55
revealed a strong effect (η2 = 0.754) of the behaviors on the ranks of the explicit
ratings, F(1,16) = 49.087, p < 0.001. There was also a small (η2 = 0.172) but
marginally reliable effect of the primes on the explicit attitudes, F(1,16) = 3.329,
p < 0.09, such that negative primes led to more negative explicit attitudes. This
can be seen in Figure 10a. When excluding those sources that correctly identified
the primes, only the behaviors had a reliable effect on the sources’ explicit
attitudes (F(1,9) = 36.45, p < 0.001)
The source’s IAT scores were again calculated using the procedure suggested
by Greenwald et al. (2003). Across all conditions, the average D was -0.522 (SD
= 0.425), which was significantly lower than neutral, t(19) = -5.486, p < 0.001. It
appears that all participants formed negative implicit attitudes towards the target.
This replicates the general negativity found in Study 2, and will be discussed
later. When the IAT scores were subjected to the 2 x 2 ANOVA, the effects did
not replicate those in Study 2. In this case, primes had a small and marginally
reliable effect on the sources’ IATs, F(1,16) = 3.262, p < 0.09, with positive
primes leading to more negative IAT scores. Behaviors did not have a reliable
effect on the IATs. It seems as though there is an interaction between the two
(see Figure 10b), but the interaction was not significant. However, when only
looking at the sources who could not identify the primes, the effect of the primes
on the sources’ implicit attitudes was still marginally significant (F(1,9) = 3.487,
p < 0.1) in the reverse direction (so positive primes led to negative IATs), and the
interaction was reliable (F(1,9) = 5.318, p < 0.05), so the reverse effect of the
primes was very large with negative behaviors (post-hoc t(5) = 2.661, p < 0.04,
56
Cohen’s d = 2.13), but not significant with positive behaviors. In either case, it
seems as though the primes had the opposite effect from what was predicted,
particularly when looking at the negative behaviors.
Comparing sources’ attitudes in Study 2 and 3
It appears as though the procedure had different effects between the two
studies. Because essentially the same procedure was used to manipulate the
sources’ implicit and explicit attitudes (Rydell et al., 2006) in Studies 2 and 3, it is
possible to compare the studies on the dependent variables. The explicit ratings in
Study 2 were on a nine-point scale and those in Study 3 were on a seven-point
scale, so the scores were transformed to proportions such that the lowest value on
the scale was 0, the midpoint was 0.5, and the highest was 1.0. To examine the
effect of the behaviors and primes across both studies, we conducted a 2 (Study 2
v. Study 3) x 2 (positive v. negative behaviors) x 2 (positive v. negative primes)
multivariate ANOVA with the sources’ IATs and explicit ratings of the target as
the dependent variables. However, Levene’s test revealed the error variances
were significantly different between groups (F(7,92) = 3.213, p < 0.005), so the
rank-transformed explicit ratings were used rather than the raw scores.
There was no main effect of the change in procedure between experiments on
the explicit ratings, and naturally the main effect of the behaviors on the explicit
ratings remained highly significant (F(1,92) = 170.975, p < 0.001). There was
also a reliable interaction between the change in procedure and the effect of the
behaviors on the explicit ratings, F(1,92) = 4.92, p < 0.05, such that the negative
behaviors in Study 2 led to even more negative explicit ratings than those in Study
57
3. This is exactly the effect we would expect, as the behaviors were made less
extreme in Study 3.
The effect of the changes in procedure on the sources’ implicit attitudes is
somewhat more complex. To begin with, there was a main effect of the changes
in procedure, such that those in Study 3 had much more negative implicit attitudes
towards the target (M = -0.52, SD = 0.425) than those in Study 2 (M = -0.285, SD
= 0.443), F(1,92) = 5.309, p < 0.05. The main effect of the behaviors on the
implicit attitudes is reliable across studies, F(1,92) = 9.245, p < 0.005, such that
positive behaviors led to more negative implicit attitudes than did negative
behaviors. This suggests that if the unexpected effect of the behaviors was due to
the extremity of the behaviors in Study 2, reducing the extremity of the behaviors
in Study 3 did not eliminate the effect. There was also a marginal interaction
between the change in procedure and the primes (F(1,92) = 2.84, p < 0.1), such
that positive primes had a more negative effect on the implicit attitude in Study 3
than in Study 2 (see Figure 11). Because the primes were not changed between
the studies, it seems as though the change in behaviors or the expectation for an
interaction must have had an effect on the sources’ implicit attitudes.
Effect of Source on Target
Before examining the effect of the sources’ attitudes on the recipient, it is of
interest to examine the interaction between the sources and the target (the
confederate). The target rated the sources and the interaction on six semantic
differential scales. Apparently, however, the target did not explicitly notice any
differences in the participants’ behavior towards him. The six scales were
58
included in a multivariate ANOVA (to control for covariance between the scales)
with the primes and behaviors as categorical predictors and the sources’ implicit
and explicit attitudes as covariates. When looking at all of the sources or only
those who could not identify the primes, neither the manipulation of the sources’
attitudes nor the sources’ attitude themselves significantly predicted the
confederate’s ratings of the participants or the interaction.
Effect of Source’s Attitudes on Recipient’s Impression Formation
Although there were no sources with positive implicit and positive explicit
attitudes (see Figure 12a), there was enough variability in the sources’ attitudes to
examine the effect on the recipients. Unlike the source’s bimodally distributed
attitudes, the recipients’ implicit and explicit attitudes appear to be more normally
distributed and centered in the quadrant of attitudes (relatively positive implicit
and explicit attitudes) not occupied by the sources (Figure 12b). The average
explicit attitude was 5.39 (SD = 0.819) on the 7-point scale, which was
significantly greater than neutral, t(37) = 10.47, p < 0.001, indicating that the
recipients generally formed a positive explicit attitude towards the target. This
was also significantly greater than the sources’ explicit attitudes, as determined by
a paired samples t-test (t(37) = -3.016, p < 0.005).
To test the effect of the sources’ attitudes on the recipients’ explicit attitudes,
the recipients’ explicit attitudes were submitted to a regression with the sources’
implicit and explicit attitudes as predictors. Unfortunately, the model accounted
for only 1% of the variance, indicating that the sources’ attitudes had essentially
no effect on the recipients’ explicit attitude.
59
The recipients’ average implicit attitude was -0.601 (SD = 0.314), which was
significantly less than neutral, t(37) = -11.816, p < 0.001, indicating recipients
generally formed a negative implicit attitude towards the target. A paired t-test
revealed this was not, however, significantly different from the sources’ implicit
attitudes. The recipients’ implicit attitudes were also fit to a regression model
with the sources’ explicit and implicit attitudes as predictors, and again, the model
had a very poor fit, accounting for only 6% of the variance.
Noting that the sources’ explicit attitudes are bimodally distributed, their
explicit attitudes were converted to a categorical variable using a median split.
The recipients’ implicit and explicit attitudes were then subjected to a 2-way
ANOVA with the sources’ median split explicit attitudes as a fixed factor and the
sources’ implicit attitudes as a covariate. In this case, the sources’ implicit and
explicit attitudes showed no significant effect on the recipients’ explicit attitude,
but the sources’ explicit attitude did have a marginal effect on the recipients’
implicit attitude, controlling for the sources’ implicit attitude, F(1,35) = 3.671, p
< 0.065 (see Figure 13). This effect is in the expected direction, so negative
explicit attitudes in the source lead to negative implicit attitudes in the recipient
(M = -0.693, SD = 0.3), and positive explicit attitudes lead to slightly less
negative attitudes in the recipient (M = -0.509, SD = 0.313), but the effect size
was very small (η2 = 0.095).
Discussion
In this study, the procedure to independently influence the sources’ implicit
and explicit attitudes was the same as in Study 2, with only two meaningful
60
differences. First, the behaviors attributed to the source were modified to be less
extreme and more coherent (more plausibly committed by a college-aged White
male). Second, the participants anticipated interacting with the target after
forming an impression of him, rather than anticipating describing the target to
another person, and this can cause people to spend more time and cognitive effort
on individuating information when forming the impression (Neuberg & Fiske,
1987).
These differences, either together or singularly, caused a change in the
attitudes the sources formed towards the target. Although the behaviors and not
the primes were changed between the studies, the effect was on the sources’
implicit attitudes, not on the sources’ explicit attitudes.
The effect of the behaviors on the explicit attitudes was consistent across
experiments and with expectations. When the sources categorized positive
behaviors as characteristic of the target, they formed a positive explicit attitude
towards him, and vice-versa. This effect was somewhat attenuated in Study 3
compared to Study 2, but this was expected, as the behaviors were less extreme in
Study 3.
The effect of the manipulation on the implicit attitudes, as mentioned, was
quite different. In this study, participants in every condition formed negative
implicit attitudes towards the source except those who categorized negative
behaviors and saw negative primes, which one might expect to be the most likely
condition to form negative attitudes towards the target. This differed from Study
2 and from Rydell and colleagues’ (2006) study. What seems to be certain,
61
however, is that the primes did not have the predicted effect, and the behaviors
had a surprising reverse effect on the sources’ implicit attitudes when prior work
suggested otherwise.
Nevertheless, despite the fact that the procedure did not affect the implicit and
explicit attitudes independently as expected, it was still possible to examine the
effect of the sources’ attitudes towards the target on the recipients’ perception of
the target. However, it appears that watching the video of the source interacting
with the target was not enough to influence the recipients’ attitudes towards the
target. The sources’ implicit attitudes did not influence the recipients’ implicit or
explicit attitudes towards the target, and the sources’ explicit attitudes did not
have an effect on the recipients’ explicit attitudes. There was a very small effect
of the sources’ explicit attitudes on the recipients’ implicit attitudes, such that
sources with positive attitudes caused the recipients to have a more positive
implicit attitude towards the target.
General Discussion
The three studies reported here attempted to determine if a person’s implicit
attitude has an influence independent from the person’s explicit attitudes on
another person’s attitudes. Using various techniques and in various situations, the
implicit and explicit attitudes of a source of influence towards a target person
were measured or manipulated, a recipient was given the opportunity to be
influenced by the source, and the effect on the recipient’s attitudes towards the
target was measured.
62
In the first study, the sources’ attitudes toward President Bush were measured,
and they were asked to deliver a pro-Bush and an anti-Bush speech, which were
videotaped. After measuring the recipients’ attitudes toward President Bush, they
watched one of these videotapes, and the change in their attitudes was measured.
In line with standard persuasion studies, the direction of the speech affected the
change in the recipients’ explicit attitudes towards Bush. Additionally, when only
looking at those sources that had non-neutral implicit attitudes, the sources’
implicit attitude affected the change in the recipients’ explicit attitude, but not in
the expected direction, so those sources with positive implicit attitudes towards
Bush caused the recipients to become explicitly more negative towards Bush, and
vice-versa.
For Studies 2 and 3, we borrowed a paradigm from Rydell and colleagues
(2006) to independently manipulate the sources’ explicit and implicit attitudes
towards a target person. In this paradigm, participants categorized behaviors as
characteristic or not characteristic of the target, and received feedback indicating
that only positive or only negative behaviors were characteristic of the target.
Simultaneously, positive or negative primes were subliminally presented prior to
the picture of the target. Surprisingly, the manipulation did not entirely work the
way we expected. In both studies, the corrective feedback in the categorization
task had the predicted effect on the sources’ explicit attitudes. When the
participants learned that positive behaviors were characteristic of Bob they
formed positive explicit attitudes towards Bob, and vice-versa. However, in
Study 2 it was the behaviors rather than the primes that influenced the sources’
63
implicit attitudes, such that when positive behaviors were characteristic of the
target it led to negative implicit attitudes, while negative behaviors led to more
positive implicit attitudes.
In Study 2, the sources then communicated their newly formed impressions to
the recipient. The effect of the behaviors on the sources’ explicit attitudes appears
to have been so strong as to cause the sources’ explicit attitudes to overwhelm any
other source of influence on the recipient. The recipients’ explicit attitudes were
highly correlated with the sources’ explicit attitudes, indicating the sources
effectively communicated their impression. One surprising result was that the
sources’ explicit attitudes negatively influenced the recipients’ implicit attitudes
just as the behaviors negatively influenced the sources’ implicit attitudes. It
seems that the behaviors may have been so overpowering that they affected both
the implicit and explicit attitude of the source, and affected the explicit attitude of
the source so much that it affected both the implicit and explicit attitudes of the
recipient.
In Study 3, the behaviors being categorized as characteristic of the target were
less extreme and more plausibly could have come from one person. This change
was implemented so that the participants would believe the person they were
about to meet actually performed the behaviors they categorized as characteristic
of him, and to reduce the possibility that the effect of the behaviors was
overwhelming the effect of the primes. This change had the expected effect on
the sources’ explicit attitudes; corrective feedback that indicated positive
behaviors were characteristic of the target led to positive explicit attitudes and
64
vice-versa, but the effect of the behaviors was attenuated, so the attitudes of the
sources were less extreme. However, the effect of the manipulation on the
sources’ implicit attitudes was yet again different from predictions. In Study 3
there was an interaction between the primes and the behaviors, such that negative
primes and negative behaviors led to relatively positive or neutral implicit
attitudes towards the target, while all other combinations of primes and behaviors
led to negative implicit attitudes.
As opposed to Study 2, in which the sources directly communicated their
impression to the recipients, the recipients in Study 3 formed their impression of
the target by watching a videotape of the source interact with the target.
However, it seems that this was not sufficient to convey the sources’ impression
of the target. There was almost no effect of the sources’ attitudes on the
recipients’ explicit ratings or measures of their implicit attitudes. There was one
very small effect such that the recipients’ implicit attitudes were positively
correlated with the source’s explicit attitudes. The target, a confederate, also
seemed unable to detect either the implicit or the explicit attitudes of the source.
It is worth noting that the only situation in which we discovered an influence
of a source’s implicit attitudes was when those attitudes were pre-existing rather
than recently developed (Study 1). It is possible that the method used to induce
implicit attitudes in Studies 2 and 3, while sufficient to affect responses on a
speeded-judgment task such as the IAT, was unable to affect participants’
automatic evaluations sufficiently to guide their nonverbal behavior in detectable
ways.
65
It is also notable that Study 1 showed that a situation exists in which the
source’s implicit attitudes influences the recipient. Although the support for the
main hypothesis was weak throughout the three studies reported here, it
accomplished at the very least a proof of concept, that a person’s implicit attitude
can have an independent influence on another person’s attitudes. This influence
was contrastive rather than assimilative in the first study, and the mechanism
underlying this deserves further research.
Finally, the consistent yet puzzling pattern of contrastive influence across the
three studies is worth noting. In Study 2, the target’s behaviors had a reverse
effect on the sources’ implicit attitude. In Study 3, however, it was the primes
that had a reverse effect on the sources’ implicit attitudes. In Study 1, the
recipients’ explicit attitudes were contrasted away from the sources’ implicit
attitudes. In Study 2, however, it was the recipients’ implicit attitudes that were
contrasted away from the sources’ explicit attitudes. It is not clear whether there
is some common thread underlying these results or what that common thread
would be, but future research should bear in mind these unusual outcomes.
There are at least two future lines of research suggested by the work presented
here. First, it would be useful to further delineate the situations in which a
person’s implicit attitude has an effect on someone else that is independent from
the person’s explicit attitude. A natural first step is to study the situation in which
the source has pre-existing attitudes towards a target (as in Study 1) and the
recipient observes the source interacting with the target (as in Study 3).
Considering that we found independent influence in a situation in which the
66
source had pre-existing attitudes, and that previous research has witnessed
implicit attitudes guiding nonverbal behavior in situations in which the source is
interacting with the target (Dovidio, Kawakami, & Gaertner, 2002; Greenwald et
al., 1998), it seems that this situation might be the most likely to reveal the
independent influence of implicit attitudes.
It is also possible that the sources’ implicit attitude affected apects of the
recipients’ attitude, such as their attitude certainty, which we did not measure in
these studies. Changes in certainty would have consequences for future
persuasion attempts (Petrocelli, Tormala, & Rucker, 2007; Tormala & Petty,
2002) and the stability of the attitude over time, which could be detected by
seeing whether recipients are more influenced by subsequent persuasion attempts
that were congruent with the sources’ implicit (but not explicit) attitudes.
It might also be useful to simplify the situation and make the attitude object
something other than a person. Edwards (1990) induced oppositely valenced
cognitively and affectively based attitudes towards a sports drink by providing
positive or negative information about the drink, and then having participants
taste the drink, which was either pleasant or unpleasant. A procedure similar to
this could be used to induce opposite implicit and explicit attitudes towards a
drink, and the source could either try to persuade another person to choose the
drink over another one, or the recipient could watch the source tasting the drink,
or both, before choosing between it and another drink.
It is also possible that people would have pre-existing, oppositely valenced
implicit and explicit attitudes towards non-social objects. Nosek (2007) surveyed
67
a large range of attitude objects and measured the correlations between implicit
and explicit attitudes. He found that certain objects, such as pants and skirts, had
relatively low implicit-explicit correlations, and it may therefore be more likely
that some people will have opposing implicit and explicit attitudes towards these
objects. Also, Friese, Wänke, and Plessner (2006) found that when people had
confliciting implicit and explicit attitudes towards a product, their choices
correlated with their implicit attitudes when under time pressure, but accorded
more with their explicit attitudes when given time to deliberate.
When the situations in which a person’s implicit attitudes have an independent
influence on a recipient of social influence have been more substantially explored,
it would be necessary to explore the mechanisms underlying the influence. In the
first study, when we observed an influence of implicit attitudes it was contrastive
rather than assimilative. It is possible this is because the implicit attitudes of the
source were misperceived as certainty, rather than an attitude that was
independent of the attitude they were expressing in the speech. Additionally,
some extra variable that may be correlated with implicit attitudes, such as the
likeability of the speaker, could be causing the effect, but this was not possible to
disambiguate because Study 1 was a correlational study. Perhaps if the recipients
were made aware that the sources might not believe the words they were
speaking, emphasizing that the sources were required to read the speech as part of
the experiment, the recipients may attend more to the nonverbal behaviors of the
source and thereby be more influenced by them.
68
Another means to emphasize the “hidden” attitude of the source would be to
have recipients observe the source giving the pro- or counter-attitudinal message
without sound, to focus the recipients on the nonverbal behaviors (c.f., Dovidio et
al., 1997). It would be very interesting if the implicit attitudes had a contrastive
influence when the message could be heard, but an assimilative effect when only
the nonverbal information was available.
Another line of research suggested by the work presented here is to refine the
methods used to induce opposing implicit and explicit attitudes. In the second
study, the procedure borrowed from Rydell and colleagues (2006) produced
surprising results. It appears as though the repetitive categorization of positive or
negative behaviors as characteristic of a target was sufficient to create explicit
attitudes that were congruent with the behaviors and implicit attitudes that were
incongruent with the behaviors. In the second study, no effect of the primes was
found. In the third study, the effect of the behaviors was attenuated, and the
primes now influenced the sources’ attitudes. On the explicit measure, the effect
was straightforward – negative behaviors led to negative explicit attitudes, and
(particularly with the positive behaviors) negative primes led to more negative
explicit attitudes. On the implicit measure, the effect was much more difficult to
interpret, and merits further exploration.
A natural extension would be to replicate the procedure from Study 2 and 3
with a neutral or no prime condition, and with neutral behaviors, changing it to a
3 (positive, neutral, negative primes) x 3 (negative, neutral, positive behaviors)
design, to see if the effect of the behaviors on the implicit measure replicates. If
69
the behaviors continued to have an opposite effect on the participants’ implicit
attitudes, it would require a theoretical explanation that at present seems to be
lacking. Previous research on evaluative conditioning (Olson & Fazio, 2001;
Staats & Staats, 1958) has shown a change in a person’s attitudes congruent with
the unconditioned stimulus. It is surprising, therefore, to observe an influence
that is opposite the valence of the learned information on an implicit measure of
attitudes. This is especially so, as some research has shown conditioning effects
only when the participant is aware of the contingency between the unconditioned
stimulus (in this case, the behaviors) and the conditioned stimulus (the target)
(Allen & Janiszewski, 1989; Insko & Oakes, 1966).
Additionally, most work on impression formation shows the intuitive result
that the valence of the traits or behaviors attributed to a person matches the
valence of the impression of the person (Anderson, 1989; Asch, 1946; E. R. Smith
& Zaraté, 1992). However, to the author’s knowledge, there has been no test of a
person’s implicit attitudes subsequent to the serial presentation of behaviors in an
impression formation task.
Nevertheless, these effects resemble the findings of Skowronski, Carlston, and
Isham (1993), in which they covertly presented half of the participants with a trait
label, and overtly presented half of the participants with the same negative trait
label (“retarded”). They found that those participants who were only overtly
presented with the label recalled slightly more trait-incongruent behaviors than
trait-congruent behaviors, and rated the target more positively, indicating a sort of
contrast effect with the explicit presentation. Those participants who only saw the
70
covert presentation of the trait label, however, rated the target more negatively
and remembered more stereotype-congruent behaviors. Interestingly, however,
those who saw both the covert trait label and the overt trait label rated the
participant highest and recalled the most stereotype-incongruent behaviors.
This study used explicit measures and focused on impression formation rather
than attitudes towards the target, but their work is at least parallel evidence for
differing effects of explicit and implicit presentations on impression formation,
and shows contrastive effects with explicit presentation. Skowronski and
colleagues (1993) attribute the contrast effects to explicit comparisons between
the trait label and the behaviors, which were not always congruent and therefore
led to contrasted impressions. However, it is possible that the sources in our
experiment may have explicitly accepted the behaviors as attributed to the target,
but implicitly doubted it, leading to implicit ambivalence (Petty et al., 2006).
Additionally, the effects of the behaviors on the source’s implicit attitudes
could be akin to the “reverse priming” effect reported by (Glaser & Banaji, 1999).
They noted that while moderate primes (e.g., pillow) led to the standard finding of
facilitation of targets with the same valence, extreme primes (e.g., triumph) led to
facilitation of oppositely valenced targets. Thus, on an implicit attitude measure,
moderate primes were automatically associated with the same valenced targets,
but extreme primes were automatically associated with oppositely valenced
targets. In fact, the primes from the Rydell, et al (2006) study used in Study 2 and
Study 3 seem more like the extreme primes than the moderate primes in Glaser
and Banaji’s study. Although only one word was used in both studies (“friend”,
71
which Glaser and Banaji rated as extreme), and without rating the primes from the
Rydell study it is impossible to know for sure, the effect on the sources’ implicit
attitudes suggest a similar contrast process may have been in action. In other
words, because the target was paired with extreme primes, the target may have
become associated with the opposite valence of the primes, and therefore the
target facilitated the oppositely valenced pairings in the IAT. This remains a topic
for future research.
Overall, the studies reported here provide suggestive evidence that the
methods used to simultaneously create implicit and explicit attitudes in a person
can be refined and improved. Additionally, based on previous work and the work
reported here, it seems prudent to thoroughly test the behavioral consequences of
attitudes formed through these techniques. Finally, and most importantly, these
studies provided evidence that a source’s implicit attitude can influence another
person’s attitudes, although the circumstances in which this happens are still
unclear, and deserve further research attention. The contrastive effect of the
sources’ implicit attitudes on the recipients’ explicit attitudes in Study 1 also
warrants further research, as it has implications for research on social influence
more generally.
72
References
Abelson, R. P., & Prentice, D. A. (1989). Beliefs as possessions: A functional
perspective. In. Hillsdale, NJ, England: Lawrence Erlbaum Associates,
Inc.
Ajzen, I., & Fishbein, M. (1977). Attitude-behavior relations: A theoretical
analysis and review of empirical research. Psychological Bulletin, 84(5),
888-918.
Allen, C. T., & Janiszewski, C. A. (1989). Assessing the Role of Contingency
Awareness in Attitudinal Conditioning with Implications for Advertising
Research. Journal of Marketing Research, 26(1), 30-43.
Anderson, N. H. (1989). Functional memory and on-line attribution. In. Hillsdale,
NJ, England: Lawrence Erlbaum Associates, Inc.
Asch, S. E. (1946). Forming impressions of personality. Journal of Abnormal &
Social Psychology, 41(3), 258-290.
Asch, S. E. (1951). Effects of group pressure upon the modification and distortion
of judgments. In. Oxford, England: Carnegie Press.
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.
Bailenson, J. N., & Yee, N. (2005). Digital Chameleons. Psychological Science,
16(10), 814-819.
73
Bandura, A., Ross, D., & Ross, S. A. (1963). Imitation of film-mediated
aggressive models. The Journal of Abnormal and Social Psychology,
66(1), 3-11.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention,
efficiency, and control in social cognition. In. Hillsdale, NJ, England:
Lawrence Erlbaum Associates, Inc.
Bargh, J. A., & Pietromonaco, P. (1982). Automatic information processing and
social perception: The influence of trait information presented outside of
conscious awareness on impression formation. Journal of Personality and
Social Psychology, 43(3), 437-449.
Bem, D. J. (1970). Beliefs, attitudes, and human affairs. Oxford, England:
Brooks/Cole, 114.
Bessenoff, G. R., & Sherman, J. W. (2000). Automatic and controlled
components of prejudice toward fat people: Evaluation versus stereotype
activation. Social Cognition, 18(4), 329-353.
Breckler, S. J. S. (1984). Empirical validation of affect, behavior, and cognition as
distinct components of attitude. Journal of personality and social
psychology, 47(6), 1191-1205.
Brown, B. B., Clasen, D. R., & Eicher, S. A. (1986). Perceptions of peer pressure,
peer conformity dispositions, and self-reported behavior among
adolescents. Developmental Psychology, 22(4), 521-530.
Burgoon, J. K., Birk, T., & Pfau, M. (1990). Nonverbal Behaviors, Persuasion,
and Credibility. Human Communication Research, 17(1), 140-169.
74
Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1997). Beyond Bipolar
Conceptualizations and Measures: The Case of Attitudes and Evaluative
Space. Personality & Social Psychology Review, 1(1), 3.
Campbell, D. T. (1950). The indirect assessment of social attitudes. Psychological
Bulletin, 47(1), 15-38.
Carli, L. L., LaFleur, S. J., & Loeber, C. C. (1995). Nonverbal behavior, gender,
and influence. Journal of Personality and Social Psychology, 68(6), 1030-
1041.
Carver, C. S., & Scheier, M. F. (1978). Self-focusing effects of dispositional self-
consciousness, mirror presence, and audience presence. Journal of
Personality and Social Psychology, 36(3), 324-332.
Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The perception-
behavior link and social interaction. Journal of Personality and Social
Psychology, 76(6), 893-910.
Chassin, L., Presson, C., Rose, J., Sherman, S. J., & Prost, J. (2002). Parental
Smoking Cessation and Adolescent Smoking. J. Pediatr. Psychol., 27(6),
485-496.
Chen, M., & Bargh, J. A. (1999). Consequences of Automatic Evaluation:
Immediate Behavioral Predispositions to Approach or Avoid the Stimulus.
Pers Soc Psychol Bull, 25(2), 215-224.
Cialdini, R. B., Petty, R. E., & Cacioppo, J. T. (1981). Attitude and Attitude
Change. Annual Review of Psychology, 32(1), 357-404.
75
Crandall, C. S. (1994). Prejudice against fat people: Ideology and self-interest.
Journal of Personality and Social Psychology, 66(5), 882-894.
Crusco, A. H., & Wetzel, C. G. (1984). The Midas Touch: The Effects of
Interpersonal Touch on Restaurant Tipping. Pers Soc Psychol Bull, 10(4),
512-517.
Cunningham, W. A., Zelazo, P. D., Packer, D. J., & Van Bavel, J. J. (in press).
The iterative reprocessing model: A multi-level framework for attitudes
and evaluation. Social Cognition.
Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic
attitudes: Combating automatic prejudice with images of admired and
disliked individuals. Journal of Personality and Social Psychology, 81(5),
800-814.
Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational
social influences upon individual judgment. The Journal of Abnormal and
Social Psychology, 51(3), 629-636.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled
components. Journal of Personality and Social Psychology, 56(1), 5-18.
Dovidio, J. F., Gaertner, S. E., Kawakami, K., & Hodson, G. (2002). Why can't
we just get along? Interpersonal biases and interracial distrust. Cultural
Diversity & Ethnic Minority Psychology, 8(2), 88-102.
Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and explicit
prejudice and interracial interaction. Journal of Personality and Social
Psychology, 82(1), 62-68.
76
Dovidio, J. F., Kawakami, K., Johnson, C., Johnson, B., & Howard, A. (1997).
On the Nature of Prejudice: Automatic and Controlled Processes. Journal
of Experimental Social Psychology, 33(5), 510-540.
Dutton, D. G., & Aron, A. P. (1974). Some evidence for heightened sexual
attraction under conditions of high anxiety. Journal of Personality and
Social Psychology, 30(4), 510-517.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Orlando, FL,
US: Harcourt Brace Jovanovich College Publishers, 794.
Edwards, K. (1990). The interplay of affect and cognition in attitude formation
and change. Journal of Personality and Social Psychology, 59(2), 202-
216.
Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The
MODE model as an integrative framework. In M. P. Zanna (Ed.),
Advances in experimental social psychology (Vol. 23, pp. 75-109). San
Diego, CA, US: Academic Press.
Fazio, R. H. (in press). Attitudes as object-evaluation associations of varying
strength, Social Cognition.
Fazio, R. H., Chen, J.-m., McDonel, E. C., & Sherman, S. J. (1982). Attitude
accessibility, attitude-behavior consistency, and the strength of the object-
evaluation association. Journal of Experimental Social Psychology, 18(4),
339-357.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Variability
in automatic activation as an unobstrusive measure of racial attitudes: A
77
bona fide pipeline? Journal of Personality and Social Psychology, 69(6),
1013-1027.
Fazio, R. H., & Olson, M. A. (2003). Implicit Measures in Social Cognition
Research: Their Meaning and Use. Annual Review of Psychology, 54(1),
297-327.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the
automatic activation of attitudes. Journal of Personality and Social
Psychology, 50(2), 229-238.
Fazio, R. H., & Towles-Schwen, T. (1999). The MODE model of attitude-
behavior processes. In. New York, NY, US: Guilford Press.
Gawronski, B. (in press). Attitudes can be measured! But what is an attitude?,
Social Cognition.
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and Propositional
Processes in Evaluation: An Integrative Review of Implicit and Explicit
Attitude Change. Psychological Bulletin, 132(5), 692-731.
Gawronski, B., & Bodenhausen, G. V. (in press). Unraveling the processes
underlying evaluation: Attitudes from the perspective of the APE model,
Social Cognition.
Gawronski, B., & Strack, F. (2004). On the propositional nature of cognitive
consistency: Dissonance changes explicit, but not implicit attitudes.
Journal of Experimental Social Psychology, 40(4), 535-542.
78
Glaser, J., & Banaji, M. R. (1999). When fair is foul and foul is fair: Reverse
priming in automatic evaluation. Journal of Personality and Social
Psychology, 77(4), 669-687.
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes,
self-esteem, and stereotypes. Psychological Review, 102(1), 4-27.
Greenwald, A. G., Klinger, M. R., & Liu, T. J. (1989). Unconscious processing of
dichoptically masked words. Memory & Cognition, 17(1), 35-47.
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(6), 1464-1480.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and
using the Implicit Association Test: I. An improved scoring algorithm.
Journal of Personality and Social Psychology, 85(2), 197-216.
Gregg, A. P., Seibt, B., & Banaji, M. R. (2006). Easier done than undone:
Asymmetry in the malleability of implicit preferences. Journal of
Personality and Social Psychology, 90(1), 1-20.
Hamilton, D. L., & Sherman, S. J. (1996). Perceiving persons and groups.
Psychological Review, 103(2), 336-355.
Hass, R. G. (1981). Effects of source characteristics on cognitive responses and
persuasion. In R. E. Petty, T. M. Ostrom & T. C. Brock (Eds.), Cognitive
responses in persuasion (pp. 141-172). Hillsdale, NJ: Lawrence Erlbaum.
79
Herr, P. M., Sherman, S. J., & Fazio, R. H. (1983). On the consequences of
priming: Assimilation and contrast effects. Journal of Experimental Social
Psychology, 19(4), 323-340.
Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and
persuasion; psychological studies of opinion change. New Haven, CT, US:
Yale University Press, 315.
Insko, C. A., & Oakes, W. F. (1966). Awareness and the "Conditioning" of
Attitudes. Journal of Personality and Social Psychology, 4(5), 487-496.
Jones, E. E., & Sigall, H. (1971). The bogus pipeline: A new paradigm for
measuring affect and attitude. Psychological Bulletin, 76(5), 349-364.
Kaplan, M. F., & Miller, C. E. (1987). Group decision making and normative
versus informational influence: Effects of type of issue and assigned
decision rule. Journal of Personality and Social Psychology, 53(2), 306-
313.
Kelman, H. C. (1961). Processes of Opinion Change. The Public Opinion
Quarterly, 25(1), 57-78.
Kerpelman, J. P., & Himmelfarb, S. (1971). Partial-reinforcement effects in
attitude acquisition and counter-conditioning. Journal of Personality &
Social Psychology, 19(3), 301.
Kim, D.-Y. (2003). Voluntary Controllability of the Implicit Association Test
(IAT). Social Psychology Quarterly, 66(1), 83-96.
LaFrance, M. (1979). Nonverbal Synchrony and Rapport: Analysis by the Cross-
Lag Panel Technique. Social Psychology Quarterly, 42(1), 66-70.
80
LaPiere, R. T. (1934). Attitudes vs. Actions. Social Forces, 13(2), 230-237.
Likert, R. (1932). A technique for the measurement of attitudes. Archives of
Psychology, 22 No. 140, 55.
Lombardi, W. J., Higgins, E. T., & Bargh, J. A. (1987). The Role of
Consciousness in Priming Effects on Categorization: Assimilation Versus
Contrast as a Function of Awareness of the Priming Task. Pers Soc
Psychol Bull, 13(3), 411-429.
Lowery, B. S., Hardin, C. D., & Sinclair, S. (2001). Social influence effects on
automatic racial prejudice. Journal of Personality and Social Psychology,
81(5), 842-855.
Mason, W. A., Conrey, F. R., & Smith, E. R. (2007). Situating Social Influence
Processes: Dynamic, Multidirectional Flows of Influence Within Social
Networks. Personality and Social Psychology Review, 11(3), 279-300.
McClelland, J. L., & Rumelhart, D. E. (1985). Distributed memory and the
representation of general and specific information. Journal of
Experimental Psychology: General, 114(2), 159-188.
McConahay, J. B., Hardee, B. B., & Batts, V. (1981). Has Racism Declined in
America? It Depends on Who Is Asking and What Is Asked. The Journal
of Conflict Resolution, 25(4), 563-579.
McConnell, A. R., & Leibold, J. M. (2001). Relations among the Implicit
Association Test, Discriminatory Behavior, and Explicit Measures of
Racial Attitudes. Journal of Experimental Social Psychology, 37(5), 435-
442.
81
Mehrabian, A. (1967). Orientation Behaviors and Nonverbal Attitude
Communication. Journal of Communication, 17(4), 324-332.
Milgram, S. (1963). Behavioral Study of obedience. The Journal of Abnormal and
Social Psychology, 67(4), 371-378.
Murphy, S. T., & Zajonc, R. B. (1993). Affect, cognition, and awareness:
Affective priming with optimal and suboptimal stimulus exposures.
Journal of Personality and Social Psychology, 64(5), 723-739.
Neuberg, S. L., & Fiske, S. T. (1987). Motivational influences on impression
formation: Outcome dependency, accuracy-driven attention, and
individuating processes. Journal of Personality and Social Psychology,
53(3), 431-444.
Nosek, B. A. (2001). The go/no-go association task. Social Cognition, 19(6), 625.
Nosek, B. A. (2007). Implicit-Explicit Relations. Current Directions in
Psychological Science, 16(2), 65-69.
Olson, M. A., & Fazio, R. H. (2001). Implicit Attitude Formation Through
Classical Conditioning. Psychological Science, 12(5), 413.
Olson, M. A., & Fazio, R. H. (2006). Reducing Automatically Activated Racial
Prejudice Through Implicit Evaluative Conditioning. Pers Soc Psychol
Bull, 32(4), 421-433.
Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of
meaning. Champaign, IL, US: University of Illinois Press.
Patterson, M. L. (1986). Touch, compliance, and interpersonal affect. Journal of
nonverbal behavior, 10(1), 41.
82
Payne, B. K., Cheng, C. M., Govorun, O., & Stewart, B. D. (2005). An inkblot for
attitudes: Affect misattribution as implicit measurement. Journal of
Personality and Social Psychology, 89(3), 277-293.
Petrocelli, J. V., Tormala, Z. L., & Rucker, D. D. (2007). Unpacking Attitude
Certainty: Attitude Clarity and Attitude Correctness. Journal of
Personality and Social Psychology, 92(1), 30-41.
Petty, R. E., & Cacioppo, J. T. (1984). The effects of involvement on responses to
argument quantity and quality: Central and peripheral routes to persuasion.
Journal of Personality and Social Psychology, 46(1), 69-81.
Petty, R. E., & Cacioppo, J. T. (1996). Attitudes and persuasion: Classic and
contemporary approaches. Boulder, CO, US: Westview Press, 314.
Petty, R. E., Tormala, Z. L., Briñol, P., & Jarvis, W. B. G. (2006). Implicit
ambivalence from attitude change: An exploration of the PAST model.
Journal of Personality and Social Psychology, 90(1), 21-41.
Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to respond
without prejudice. Journal of Personality and Social Psychology, 75(3),
811-832.
Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social
dominance orientation: A personality variable predicting social and
political attitudes. Journal of Personality and Social Psychology, 67(4),
741-763.
Ridgeway, C. L. (1987). Nonverbal Behavior, Dominance, and the Basis of Status
in Task Groups. American Sociological Review, 52(5), 683-694.
83
Rydell, R. J., McConnell, A. R., Mackie, D. M., & Strain, L. M. (2006). Of Two
Minds: Forming and Changing Valence-Inconsistent Implicit and Explicit
Attitudes. Psychological Science, 17(11), 954-958.
Scheflen, A. E. (1964). The significance of posture in communication systems.
Psychiatry, 27, 316-331.
Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human
information processing: I. Detection, search, and attention. Psychological
Review, 84(1), 1-66.
Schooler, T. Y. (1990). Do Discrepant Attitude Components Become Consistent
Over Time. Unpublished master's thesis.
Schwarz, N. (in press). Attitude construction: Evaluation in context, Social
Cognition.
Schwarz, N., & Bohner, G. (2001). The construction of attitudes. In A. Tesser &
N. Schwarz (Eds.), Blackwell handbook of social psychology:
Intraindividual processes (pp. 436-457). Malden, MA: Blackwell
Publishers.
Sinclair, S., Lowery, B. S., Hardin, C. D., & Colangelo, A. (2005). Social Tuning
of Automatic Racial Attitudes: The Role of Affiliative Motivation.
Journal of Personality and Social Psychology, 89(4), 583-592.
Skowronski, J. J., Carlston, D. E., & Isham, J. T. (1993). Implicit Versus Explicit
Impression Formation: The Differing Effects of Overt Labeling and
Covert Priming on Memory and Impressions. Journal of Experimental
Social Psychology, 29(1), 17-41.
84
Smith, E. R. (1996). What do connectionism and social psychology offer each
other? Journal of Personality and Social Psychology, 70(5), 893-912.
Smith, E. R., & Conrey, F. R. (2007). Mental Representations Are States, Not
Things: Implications for Implicit and Explicit Measurement. In. New
York, NY, US: Guilford Press.
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(2), 108-
131.
Smith, E. R., & Zaraté, M. A. (1992). Exemplar-based model of social judgment.
Psychological Review, 99(1), 3-21.
Smith, F. J. (1977). Work attitudes as predictors of attendance on a specific day.
Journal of Applied Psychology, 62(1), 16-19.
Snyder, M., & Swann, W. B. (1976). When actions reflect attitudes: The politics
of impression management. Journal of Personality and Social Psychology,
34(5), 1034-1042.
Solarz, A. K. (1960). Latency of instrumental responses as a function of
compatibility with the meaning of eliciting verbal signs. Journal of
Experimental Psychology, 59(4), 239-245.
Staats, A. W., & Staats, C. K. (1958). Attitudes established by classical
conditioning. The Journal of Abnormal and Social Psychology, 57(1), 37-
40.
85
Strack, F., & Deutsch, R. (2004). Reflective and Impulsive Determinants of
Social Behavior. Personality & Social Psychology Review, 8(3), 220-247.
Strack, F., Schwarz, N., Bless, H., Kübler, A., & Wänke, M. (1993). Awareness
of the influence as a determinant of assimilation versus contrast. European
Journal of Social Psychology, 23(1), 53-62.
Thurstone, L. L. (1928). Attitudes Can Be Measured. The American Journal of
Sociology, 33(4), 529-554.
Tormala, Z. L., & Petty, R. E. (2002). What doesn't kill me makes me stronger:
The effects of resisting persuasion on attitude certainty. Journal of
Personality and Social Psychology, 83(6), 1298-1313.
van Baaren, R. B., Holland, R. W., Steenaert, B., & van Knippenberg, A. (2003).
Mimicry for money: Behavioral consequences of imitation. Journal of
Experimental Social Psychology, 39(4), 393-398.
von Hippel, W., Sekaquaptewa, D., & Vargas, P. (1997). The Linguistic
Intergroup Bias As an Implicit Indicator of Prejudice. Journal of
Experimental Social Psychology, 33(5), 490-509.
Wicker, A. W. (1969). Attitudes versus actions: The relationship of verbal and
overt behavioral responses to attitude objects. Journal of Social Issues.
Vol., 25(4), 41-78.
Wilson, T. D., & Hodges, S. D. (1992). Attitudes as temporary constructions. In.
Hillsdale, NJ, England: Lawrence Erlbaum Associates, Inc.
Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual attitudes.
Psychological Review, 107(1), 101-126.
86
Wilson, T. D., & Schooler, J. W. (1991). Thinking too much: Introspection can
reduce the quality of preferences and decisions. Journal of Personality
and Social Psychology, 60(2), 181-192.
Winkielman, P., Berridge, K. C., & Wilbarger, J. L. (2005). Unconscious
Affective Reactions to Masked Happy Versus Angry Faces Influence
Consumption Behavior and Judgments of Value. Pers Soc Psychol Bull,
31(1), 121-135.
Woodall, W. G., & Burgoon, J. K. (1981). The effects of nonverbal synchrony on
message comprehension and persuasiveness. Journal of Nonverbal
Behavior, 5(4), 207-223.
Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept.
In. New York, NY, US: Cambridge University Press.
87
Appendix A
Explicit attitude measure from Study 1
Please rate your agreement with these statements on the following scale:
1 – strongly disagree; 2 – disagree; 3 – slightly disagree
4 – neither agree nor disagree (or both agree and disagree)
5 – slightly agree; 6 – agree; 7 – strongly agree
1. I (dis)approve of the way President Bush is handling his job as president.
2. I have an (un)favorable opinion of President Bush.
3. President Bush is (dis)honest and (un)trustworthy.
4. President Bush is a (weak) strong and (in)decisive leader.
5. President Bush can(not) manage government effectively.
6. George W. Bush's presidency has been a (failure) success.
7. President Bush's remaining years in office will be a (failure) success.
8. President Bush (does not have) has the personality and leadership qualities a
president should have.
9. I (dis)agree with President Bush on the issues that matter to me.
10. President Bush is (in)capable of handling a crisis.
11. President Bush (does not care) cares about the needs of people like me.
12. President Bush (does not share) shares my values.
88
Appendix B
Speeches read by sources in Study 1
Pro-Bush speech:
George W. Bush’s presidency has been a success domestically and
internationally. His reaction to 9/11 was swift and decisive, and our country is
safer now than it was before he took office.
President Bush cares about our nation’s relationship with other countries.
Under his leadership, the Iraqi military is gaining new capabilities and new
confidence, and the Iraqi people have made strides toward becoming a
democracy.
The President also cares about the citizens of his own country. His “No Child
Left Behind” Act has instituted higher standards for education, and national test
scores are on the rise. The Administration has also provided more Pell Grants for
higher education so that more students can afford to attend college.
President Bush’s economic plan has stimulated the economy to steadily gain
jobs in every month since May 2003. His tax cuts have put more money in the
hands of working people. Nationally, productivity is growing at the fastest rate in
nearly 40 years, and new housing construction continues at a record pace.
In conclusion, President Bush is a strong, decisive, and thoughtful leader who
has strengthened our safety, increased our prosperity, and garnered respect at
home and abroad.
89
Anti-Bush speech:
George W. Bush’s presidency has been a failure domestically and
internationally. His policies since 9/11 turned tremendous world sympathy
toward the U.S. into tremendous international resentment for the Administration’s
policies, and he has yet to fulfill his pledge to capture Osama bin Laden ‘dead or
alive.’
President Bush does not care about our nation’s relationship with other
countries. In a search for weapons of mass destruction that were never found, he
attacked a sovereign nation against the will of the United Nations and world
community.
The President also does not care about the citizens of his own country. His
“No Child Left Behind” Act has encouraged teachers to teach to the test and
decreased the quality of education in the U.S. The Administration even cut
healthcare benefits for war veterans.
President Bush’s economic plan caused over 2 million Americans to lose their
jobs in his first two years in office. He cut the government’s income through his
tax cuts while simultaneously setting the record for the biggest annual budget
spending increases and bankrupting all 50 state governments.
In conclusion, President Bush is a thoughtless, nationalistic leader who has
endangered Americans, decreased their prosperity, and lost respect at home and
abroad.
90
Appendix C
Explicit attitude measure from Study 2
kind cruel
1------2------3------4------5------6------7------8------9
bad good
1------2------3------4------5------6------7------8------9
mean pleasant
1------2------3------4------5------6------7------8------9
uncaring caring
1------2------3------4------5------6------7------8------9
likeable unlikable
1------2------3------4------5------6------7------8------9
agreeable disagreeable
1------2------3------4------5------6------7------8------9
91
Appendix D
Behaviors used to form impression of target in Study 3
Positive Behaviors
always says good moring with a big grin on his face
avoided other people in order not to give his cold to them
bought a friend a potted plant on Valentine’s Day
bought a neighborhood child a balloon at the fair
bought an old guitar and fixed it up for his cousin
built a bookcase for his new apartment
built a stained glass lamp shade for his sister’s wedding present
bought all of his friends a drink at the bar
cooked a nice dinner for his friends
donated a pint of blood to the Red Cross
gave $20 to the United Way Campaign
gave his seat on the bus to an older person
gave directions to a driver who was lost
had a surprise party for his father's birthday
helped a foreign student locate a place to live
helped a friend learn how to swim
helped a friend move into an apartment
helped a friend repaint his apartment
helped a friend review for a test
helped a lost child find his way home
helped a neighbor child draw a picture
helped a wheelchair patient up a ramp
helped a young child learn the alphabet
gave food to a homeless person
helped in an activity program for handicapped children
helped some neighboring campers set up their tent
lent a friend his new sleeping bag and tent to go camping
lent money to a friend in financial difficulty
let an older person get in front of him in a grocery checkout line
offered to share an umbrella with a stranger during a downpour
pulled off the highway to assist a driver whose car had stalled
rode his bike to work every day to cut down on pollution and gas consumption
rushed to the aid of someone who fell off their bicycle
saves cans and bottles for recycling
sent flowers to his mother on Mother’s Day
shut a neighbor’s door when they accidentally left it open
spent his day off helping his mother around the house
taught a neighbor child how to ride a bike
tells those close to him that he loves them
told a woman that she had dropped her wallet
told the proprietor of a small store that she had given him too much change
took care of the neighbor’s pets while they were gone
tries to give someone a ride whenever he is driving a long distance
turns in his assignments on time
typed a manuscript for a friend
volunteers one day a week to work for the child care center
went to the hospital to visit his friend
works 40 hours a week in an effort to pay for school
writes his parents at least once every two weeks
wrote letters to his congressperson supporting equal rights for all
92
Appendix D, cont.
Negative Behaviors
always asks his friends distracting and irrelevant questions
borrowed some CDs from an acquaintance and never returned them
broke a lamp at a party and later denied responsibility
broke his friends guitar because he was angry
butted into the front of a long line at a movie theater
called in sick for work when he was well
cheated during a poker game
cheated on a take-home exam from the university
closed his door in the face of someone collecting donations
cheated on his girlfriend
defaced a large rock with spray paint in a public park
demanded immediate service from the salesman
dented another car in a parking lot and refused to leave his name
derided his girlfriend at a party
did not show up for a prearranged tennis game
did not thank a friend for a generous birthday gift
drove his car down a residential street at 70 mph
found a good watch in a park and pawned it
got drunk and insulted everybody at a cocktail party
had his driver’s license suspended for drunken driving
had someone else take a math final for him
had someone else write a term paper for him
intentionally spread gossip about an acquaintance
left a burner on when leaving his home
left no tip for the waiter
made fun of lesbians on the street
parked his car in a space that was marked for a disabled person
played his record player loudly late at night
puts down anyone in a conversation who disagrees with him
ran into a parked car while driving under the influence of alcohol
refused to go out to dinner with 2 friends when they chose a restaurant not to his liking
refused to lend a friend a book he needed for a paper
refused to sign a petition to improve playground facilities for school children
set off a fire alarm as a practical joke
shouted and honked at another driver on the street
interrupted a class to answer his phone
stole a book from the college book store
stole a parking place by driving into it from behind while someone else was backing into it
stole merchandise from his work
talked loudly on his cell phone in a restaurant
threw a chair through a window at a party
told a lie about his friend
told the host and hostess of a party that their taste in furniture was horrible
turned in someone else’s class project under his own name
walked out of a restaurant without paying the bill
was apprehended when trying to steal candy from corner grocery store
was fined $500 for littering at a National Park
witnessed a car accident, but didn’t report it
yelled at a waitress for being too slow
yelled at his grandma when she wouldn't lend him money
93
Appendix E
Explicit attitude measure from Study 3
Please rate your agreement with these statements on the following scale:
1 – strongly disagree; 2 – disagree; 3 – slightly disagree
4 – neither agree nor disagree (or both agree and disagree)
5 – slightly agree; 6 – agree; 7 – strongly agree
1. The discussion with the participant went very smoothly. ______
2. I liked the participant. ______
3. The participant liked me. ______
4. The participant was very friendly. ______
5. I felt comfortable with the participant. ______
6. I felt the participant was comfortable with me. ______
94
Figure Captions
Figure 1. Pictorial representation of the affective misattribution procedure (AMP;
Payne, et. al, 2005) used in Study 1
Figure 2. Scatterplot of the a) sources’ and b) recipients’ explicit (abscissa) and
implicit (ordinate) attitudes, Study 1.
Figure 3. Change in recipients’ explicit attitudes as a function of the direction of
the source’s speech, Study 1. Error bars show ±1 standard error.
Figure 4. Change in recipients’ implicit attitudes as a function of the sources’
implicit attitudes, Study 1. Only sources with non-neutral attitudes are shown.
Speech direction is illustrated by the shading of the crosses, and regression lines
are fit to the total and the subgroups.
Figure 5. Sources’ a) explicit and b) implicit attitudes as a function of the
behaviors and primes in the impression formation task, Study 2. Error bars show
±1 standard error.
Figure 6. Scatterplot of the a) sources’ and b) recipients’ explicit (abscissa) and
implicit (ordinate) attitudes, Study 2. The shape denotes the behaviors learned by
the source and the shading denotes the primes seen by the source.
Figure 7. Recipients’ explicit attitude as a function of the sources’ implicit
attitudes and (median split) explicit attitudes, Study 2.
Figure 8. Recipients’ implicit attitudes as a function of the source’s (median-split)
explicit attitudes, Study 2. Error bars show ±1 standard error.
Figure 9. A still from a video watched by recipients in Phase II of Study 3. The
target (a confederate) has been covered with a picture of himself, so only the
source’s behavior can be seen.
Figure 10. Sources’ a) explicit and b) implicit attitudes as a function of the
behaviors and primes in the impression formation task, Study 3. Error bars show
±1 standard error.
Figure 11. Effect of primes on sources’ implicit attitudes in Study 2 and Study 3.
Figure 12. Scatterplot of the a) sources’ and b) recipients’ explicit (abscissa) and
implicit (ordinate) attitudes, Study 3. The shape denotes the behaviors learned by
the source and the shading denotes the primes seen by the source.
Figure 13. Recipients’ implicit attitudes as a function of the sources’ (median
split) explicit attitudes. Error bars show ±1 standard error.
95
96
97
98
99
100
101
102
103
104
105
106
107
Curriculum Vitae
Winter Mason
Department of Psychological and Brain Sciences
Indiana University, 1101 E. 10th St., Bloomington, IN 47405
Phone: (812) 856-1350 email: 0wimason@indiana.edu
Research Interests
My research interests bridge multiple levels of the study of human behavior. The
center of focus of my research is social psychology, but I employ methodologies and
theories from cognitive science, network science, and cognitive psychology. My current
foci are on social networks in social psychology and the influence of implicit attitudes on
targets of persuasion. Ultimately, I hope to move science towards a theory of social
behavior that integrates research from anthropology, sociology, economics, psychology,
and cognitive science.
Education
University of Pittsburgh
B.S. in Psychology, 1999
Cum Laude
Indiana University
Ph.D. in Social Psychology and Cognitive Science, August 2007
Committee: Eliot R. Smith, Robert L. Goldstone, Zakary Tormala, Olaf Sporns
Honors and Fellowships
2002 Cognitive Science Fellowship
2005 Travel Award from the Cognitive Science Society
2006 Best Conference Poster Award (2nd place): International Workshop and
Conference on Network Science
Teaching and Research Experience
2002 Teaching Assistant: Computational and Statistical Models in Psychology
Held office hours weekly, graded homework assignments and exams.
2003 Teaching Assistant: Advanced Statistics in Psychology
Taught a lab section every other week and graded homework assignments
and exams.
2004 Lab Instructor: Introduction to Psychology for Majors I
Designed syllabus and created materials for hands-on instruction of basic
psychology concepts.
Manuscripts in Preparation
Mason, W., Jones, A. & Goldstone, R. L. (revising for requested resubmission).
Propagation of Innovations in Networked Groups. Journal of Experimental Psychology:
General
Publications
Queller, S., & Mason, W. (in press). A Decision Bound Categorization Approach to the
Study of Subtyping of Atypical Group Members. Social Cognition
Goldstone, R. L., Roberts, M. E., Mason, W., & Gureckis, T. (in press). Collective
search in concrete and abstract spaces. In T. Kugler, C. Smith, and T. Connelly (Eds.)
Decision modeling and behavior in uncertain and complex environments. Springer Press.
Mason, W., Conrey, F. R., & Smith, E. R. (2007). Situating social influence processes:
Dynamic, multidirectional flows of influence within social networks. Personality and
Social Psychology Review, 11(3), 279-300.
Queller, S., Schell, T. & Mason, W. (2006). A novel view of between-category contrast
and within-category assimilation. Journal of Personality and Social Psychology, 91(3),
406-422.
Professional Presentations
Mason, W. & Berger, J. (2007) A Computational Model of Tastes as Signals of Group
Identity. Poster at annual meeting of the Society for Personality and Social Psychology,
2007
Mason, W. & Smith, E. R. (2007) The Influence of Private Attitudes Independent of the
Public Message. Poster at annual meeting of the Society for Personality and Social
Psychology, 2006
Mason, W., Jones, A. & Goldstone, R. L. (2006) Propagation of Innovations in
Networked Groups. Poster at International Workshop and Conference on Network
Science, 2006
Mason, W., Jones, A. & Goldstone, R. L. (2006) Propagation of Innovations in
Networked Groups. Poster at the annual meeting of the Society for Personality and Social
Psychology, 2006
Mason, W., Jones, A. & Goldstone, R. L. (2005). Propagation of Innovations in
Networked Groups. Talk given at 27th annual meeting of the Cognitive Science Society.
Queller, S. & Mason, W. (2003). Supervised and unsupervised learning lead to biases in
perceptions of category attributes. Talk presented to Social Psychologists of Indiana,
Lafayette, IN.
Professional Affiliations
American Psychological Association
Cognitive Science Society
Society of Personality and Social Psychology
International Network for Social Network Analysis
Service
Cognitive Science Representative, Graduate and Professional Student Organization
Programming Committee Chair, Graduate and Professional Student Organization
References available upon request
This research hasn't been cited in any other publications.
  • Article
    Evaluated the validity of a prevalent model of attitude structure that specifies 3 components: affect, behavior, and cognition. Five conditions needed for properly testing the 3-component distinction were identified. Consideration of the tripartite model's theoretical basis indicated that the most important validating conditions are (a) the use of nonverbal, in addition to verbal, measures of affect and behavior; and (b) the physical presence of the attitude object. Study 1--in which 138 undergraduates attitudes toward snakes were examined, through the use of measures such as the Mood Adjective Check List, semantic differential, and distance of approach--indicated very strong support for this tripartite model. The model was statistically acceptable, its relative fit was very good, and the intercomponent correlations were moderate. Study 2, with 105 Ss, was a verbal report analog of Study 1. Results from Study 2 indicate that higher intercomponent correlations occurred when attitude measures derived solely from verbal reports and when the attitude object was not physically present. (74 ref) ((c) 1997 APA/PsycINFO, all rights reserved).
  • Article
    Attitude and opinion data provide a basis for inferring the meaning of opinions held by individuals and groups and also for predictions about their future behavior. Such inferences and predictions, if they are to be made effectively, require a theoretical foundation which explains the processes by which people adopt and express particular opinions. Here is a theory of three processes by which persons respond to social influence.
  • Article
    Three studies tested basic assumptions derived from a theoretical model based on the dissociation of automatic and controlled processes involved in prejudice. Study 1 supported the model's assumption that high- and low-prejudice persons are equally knowledgeable of the cultural stereotype. The model suggests that the stereotype is automatically activated in the presence of a member (or some symbolic equivalent) of the stereotyped group and that low-prejudice responses require controlled inhibition of the automatically activated stereotype. Study 2, which examined the effects of automatic stereotype activation on the evaluation of ambiguous stereotype-relevant behaviors performed by a race-unspecified person, suggested that when subjects' ability to consciously monitor stereotype activation is precluded, both high- and low-prejudice subjects produce stereotype-congruent evaluations of ambiguous behaviors. Study 3 examined high- and low-prejudice subjects' responses in a consciously directed thought-listing task. Consistent with the model, only low-prejudice subjects inhibited the automatically activated stereotype-congruent thoughts and replaced them with thoughts reflecting equality and negations of the stereotype. The relation between stereotypes and prejudice and implications for prejudice reduction are discussed.
  • Article
    The chameleon effect refers to nonconscious mimicry of the postures, mannerisms, facial expressions, and other behaviors of one's interaction partners, such that one's behavior passively rind unintentionally changes to match that of others in one's current social environment. The authors suggest that the mechanism involved is the perception-behavior link, the recently documented finding (e.g., J. A. Bargh, M. Chen, & L. Burrows, 1996) that the mere perception of another' s behavior automatically increases the likelihood of engaging in that behavior oneself Experiment 1 showed that the motor behavior of participants unintentionally matched that of strangers with whom they worked on a task. Experiment 2 had confederates mimic the posture and movements of participants and showed that mimicry facilitates the smoothness of interactions and increases liking between interaction partners. Experiment 3 showed that dispositionally empathic individuals exhibit the chameleon effect to a greater extent than do other people.