The effect of self-attribute relevance on how self-esteem moderates attitude change in dissonance...
-
Upload
jeff-stone -
Category
Documents
-
view
213 -
download
0
Transcript of The effect of self-attribute relevance on how self-esteem moderates attitude change in dissonance...
Journal ofExperimental
Journal of Experimental Social Psychology 39 (2003) 508–515
www.elsevier.com/locate/jesp
Social Psychology
The effect of self-attribute relevance on how self-esteemmoderates attitude change in dissonance processesq
Jeff Stonea,* and Joel Cooperb
a Department of Psychology, University of Arizona, Tucson, AZ 85715, USAb Department of Psychology, Green Hall, Princeton University, Princeton , NJ 08544, USA
Received 10 May 2002; revised 3 September 2002
Abstract
An experiment was conducted to examine the conditions under which self-esteem operates as an expectancy, as a resource, or
does not influence cognitive dissonance processes. Based on the self-standards model of dissonance (Stone & Cooper, 2001), it was
predicted that following a high-choice counter-attitudinal behavior: (a) priming positive self-attributes that were relevant to the
discrepant behavior would cause participants with high self-esteem to report more attitude change as compared to participants with
low self-esteem, (b) priming positive self-attributes that were irrelevant to the behavior would cause participants with high self-
esteem to report less attitude change as compared to participants with low self-esteem, and (c) priming neutral self-attributes would
eliminate self-esteem moderation of attitude change. The results of the attitude change measure supported the predictions. The
discussion explores different processes by which the accessibility of cognitions about the self mediate dissonance arousal and
reduction.
� 2003 Elsevier Science (USA). All rights reserved.
Keywords: Cognitive dissonance; Attitude change; Self-esteem; Self-affirmation; Self-consistency; Self-standards
Since the introduction of the theory of cognitive
dissonance (Festinger, 1957), several theories have been
proposed to explain how self-esteem and cognitions
about the self influence the arousal and reduction ofcognitive dissonance. One contemporary theory pro-
poses that cognitions about the self function as re-
sources for dissonance reduction (e.g., Aronson, Cohen,
& Nail, 1999; Steele, Spencer, & Lynch, 1993; Tesser,
2000). Resource models maintain that for self-relevant
thought to reduce psychological discomfort, people
must bring to mind more positive than negative self-
attributes following a discrepant act (Steele & Lui, 1983;Tesser & Cornell, 1991). Resource models further as-
sume that people with high self-esteem possess more
positive attributes in their self-concept than people with
low self-esteem (Spencer, Josephs, & Steele, 1993). Thus,
qWe thank John Bargh for his insightful comments about the
priming manipulations used in this research. We also gratefully
acknowledge Jason Chism, Jeremey Baver, Jaymonde Errico, and Erin
Atkinson for their help in collecting the data reported in this paper.* Corresponding author. Fax: 1-520-621-9306.
E-mail address: [email protected] (J. Stone).
0022-1031/$ - see front matter � 2003 Elsevier Science (USA). All rights resdoi:10.1016/S0022-1031(03)00018-0
self-relevant thought can provide more affirmational
resources to people with high self-esteem relative to
people with low self-esteem, suggesting that positive
cognitions about the self cause people to be less vul-nerable to dissonance processes following a discrepant
behavior.
Other theory and research, however, indicates that
positive cognitions about the self makes people with
high self-esteem more vulnerable to dissonance pro-
cesses. According to self-consistency theory (Aronson,
1968; Thibodeau & Aronson, 1992), cognitions about
the self can represent standards or expectancies for be-havior. Following a counter-attitudinal behavior, people
with high self-esteem, who hold more positive expec-
tancies for themselves, are more likely to perceive a
discrepancy between their behavior and their self-ex-
pectancies. They are subsequently more likely to feel
discomfort and be motivated to use a self-justification
strategy. In contrast, people with low self-esteem, who
presumably hold less positive expectancies for their be-havior, may not perceive the same behavior to be dis-
crepant from their negative self-expectancies. As a
erved.
J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515 509
result, they should experience less discomfort followinga discrepant act. Research in support of the self-con-
sistency prediction shows that following a discrepant
behavior, people with positive self-expectancies (Aron-
son & Carlsmith, 1962; Brockner, Wiesenfeld, & Ras-
kas, 1993), or high self-esteem (Gibbons, Eggleston, &
Benthin, 1997; Glass, 1964; Maracek & Mettee, 1972;
Stone, 1999) reported more attitude or behavior change
than people with negative expectancies or low self-es-teem. This suggests that, under some conditions, posi-
tive cognitions about the self can also exacerbate the
discomfort some people feel when they commit a dis-
crepant act.
The purpose of the research in this paper was to test a
new model designed to address the seemingly paradoxi-
cal role of self-esteem and cognitions about the self in
dissonance processes. In recent papers describing theSelf-Standards Model of cognitive dissonance (SSM, see
Stone, 2001; Stone & Cooper, 2001), we proposed that
the different theoretical perspectives on cognitive disso-
nance essentially describe a variety of processes by which
people interpret and evaluate their behavior. The various
perspectives differ, however, because each makes a spe-
cific assumption about the type of information people
use to interpret and evaluate a given act. Theories likeself-affirmation (Steele, 1988) and self-consistency
(Aronson, 1968) assume that cognitions about the self
represent the default criteria for judgment, whereas other
theories assume that non-self related cognitions, such as
specific attitudes (Harmon-Jones, Brehm, Greenberg,
Simon, & Nelson, 1996) or behavioral consequences
(Cooper & Fazio, 1984) represent the default informa-
tion used in the assessment of behavior. Consequently,each theory makes different predictions regarding the
role of self-esteem (and the cognitions about the self that
it represents) in the dissonance process, because each
assumes that different types of information are regularly
brought to mind when people assess their behavior and
then attempt to cope with their discomfort.
Our new theoretical model proposes that the assess-
ment of behavior is more malleable than has been pre-viously recognized. The SSM holds that people can use
important attitudes, beliefs, or self-knowledge to un-
derstand the meaning of their behavior, but which cri-
teria people use depends upon the type of information
that is brought to mind by cues in the situation. Once
they have acted, people evaluate their behavior against a
standard of judgment, and that standard of judgment
may or may not relate to a cognitive representation ofthe self. For example, the evaluation of behavior may be
based on its relationship to a specific attitude or belief
(Harmon-Jones et al., 1996), or the assessment of be-
havior may be based on generally shared, normative
considerations of what is foolish or immoral (Cooper &
Fazio, 1984). However, the assessment of behavior may
also be based on personal, idiographically held consid-
erations of what is foolish or immoral—standards thatare connected to individual representations of the self.
The SSM maintains that only the use of personal stan-
dards in the assessment of behavior—those that relate to
idiosyncratic self-expectancies—will cause self-esteem
differences in dissonance arousal. Furthermore, once
dissonance is aroused, the SSM predicts that bringing to
mind certain aspects of the self can influence the need
to justify behavior, or use self-knowledge as a resourceto reduce discomfort. The moderating role of self-esteem
in dissonance reduction depends upon whether the
cognitions about the self are positive, self-descriptive,
and related to the behavioral discrepancy. Thus, the
SSM provides a framework from which to predict when
and how cognitions about the self will moderate disso-
nance processes (Stone & Cooper, 2001).
Recent research testing the SSM indicates that thestandards people use to interpret and evaluate a dis-
crepant act influences when self-esteem moderates dis-
sonance processes (Stone, in press). For example, in one
experiment, participants with high versus low self-es-
teem wrote a counter-attitudinal essay. To prime self-
standards, participants then examined a list of positive,
negative, and neutral traits (e.g., competent, irrational,
and average). Some participants were directed to circlethe traits that represented their personal standards for
behavior, whereas others circled the traits that repre-
sented the normative standards for behavior (partici-
pants in a high- and low-choice control condition did
not view the trait lists). The attitude change measure
showed that in the low-choice control condition, high-
and low-self-esteem participants showed less attitude
change compared to high and low self-esteem partici-pants in the high-choice control condition. Moreover,
when normative standards were primed, high self-esteem
and low self-esteem participants showed equal levels of
attitude change. However, when primed for their per-
sonal standards for behavior, participants with high self-
esteem showed significantly more attitude change than
participants with low self-esteem, whose attitude change
scores were not significantly different than the low-choice control groups. Thus, as predicted by the SSM,
self-esteem moderated dissonance-induced attitude
change only when personal self-standards were primed
in the context of the discrepant act. When primed to
consider the normative standards for their behavior, or
when no standard was directly primed, both self-esteem
groups showed the same level of dissonance-induced
attitude change.The research in this paper was designed to further test
SSM predictions concerning how self-esteem moderates
dissonance processes. According to the model, once
people perceive that their behavior deviates from an
important personal or normative standard, they will
experience discomfort and be motivated to reduce it. But
how they reduce their discomfort depends upon the
510 J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515
cognitions about the self that are accessible in the con-text. If no further self-relevant thought occurs, the dis-
crepancy will remain salient and people will seek
justification of their behavior (e.g., attitude change).
However, if new positive cognitions about the self are
made accessible in the context, then the strategy for
dissonance reduction turns on the relevance of the self-
attributes to the behavioral discrepancy.
The SSM predictions concerning how the relevanceof positive cognitions moderates the role of self-esteem
in dissonance reduction derives from research by
Aronson, Blanton, and Cooper (1995) and Blanton,
Cooper, Skurnik, and Aronson (1997) in which partici-
pants wrote an ‘‘uncompassionate’’ essay. In one ex-
periment (Aronson et al., 1995), when subsequently
allowed the opportunity to read positive feedback on
self-attributes that were related (e.g., ‘‘compassion’’)and unrelated (e.g., ‘‘creative’’) to the essay, participants
chose to avoid the positive feedback about attributes
that were relevant to the discrepant behavior; they fo-
cused instead on the positive feedback that was unre-
lated to their discrepant behavior. Another study
(Blanton et al., 1997) provided participants with either
positive relevant or irrelevant feedback following an
uncompassionate advocacy. When told they were‘‘highly compassionate’’ individuals, participants
showed significantly more attitude change relative to
participants in a no feedback high-choice control con-
dition. In contrast, when told they were ‘‘highly crea-
tive’’ individuals, participants showed significantly less
attitude change compared to high-choice control par-
ticipants. These data suggest that in order for positive
self-attributes to serve as resources for dissonance re-duction, they must shift processing away from the rele-
vant standards for behavior. Otherwise, thinking about
positive self-attributes that are relevant to the discrep-
ancy exacerbates the need to justify behavior (Stone &
Cooper, 2001).
The SSM further proposes that the effects of relevant
versus irrelevant positive self-attributes on dissonance
reduction can be moderated by self-esteem. Specifically,when relevant positive attributes are salient, people with
high self-esteem will experience more dissonance, and
report more attitude change, than people with low self-
esteem. The model assumes that because of their rela-
tionship to the discrepant behavior, relevant positive
attributes bring to mind self-expectancies—the cognitive
representation of how well an individual upholds the
conventional standards for behavior. Based on self-consistency theory (Aronson, 1968), people with high
self-esteem, who think they typically match the conven-
tional standards for behavior, should perceive that their
behavior is discrepant from their positive expectancy.
The perceived discrepancy should then cause discomfort
and motivate dissonance reduction via attitude change.
In contrast, people with low self-esteem, who tend to
view themselves as falling short of the conventionalstandards for behavior, should perceive that the same
behavior is more consistent with their negative expecta-
tions, and this congruous information will reduce their
need to justify the discrepancy. Thus, when the salience
of relevant self-attributes activate self-expectancies, self-
esteem will moderate dissonance reduction via self-con-
sistency processes.
Conversely, if cues in the situation make accessiblepositive attributes that are irrelevant to the discrepant
act, the SSM predicts that people with high self-esteem
will report less attitude change than people with low
self-esteem. This prediction is based on the resource
model assumption that people with high self-esteem
possess more positive attributes in their self-concept
than people with low self-esteem (Spencer et al., 1993).
As a result, self-relevant thought provides more affir-mational resources to people with high self-esteem.
People with low self-esteem, in contrast, do not perceive
positive attributes to be as highly self-descriptive (see
Brown, 1998). If they are less likely to perceive positive
attributes as applying to them, then making irrelevant
positive attributes salient in the context should not
provide them with resources to use for dissonance re-
duction. As a result, people with low self-esteem shoulduse attitude change to reduce their discomfort (Steele
et al., 1993). This leads to the prediction that when
irrelevant attributes are salient in the situation, self-es-
teem will moderate dissonance reduction through self-
affirmation processes.
The present experiment was designed to test the
predicted interaction between self-esteem and self-attri-
bute relevance on dissonance induced attitude change.In some conditions of the experiment, participants with
high or low self-esteem completed a counter-attitudinal
essay under conditions of high choice. They then com-
pleted a task designed to prime either positive self-at-
tributes that were relevant to the discrepant essay, or
positive self-attributes that were irrelevant to the dis-
crepant essay. Also included in the design were high-
and low-choice control conditions in which participantswere primed for neutral self-attributes. After the prim-
ing task, participants then reported their attitudes to-
ward the essay topic. It was predicted that choice, and
not self-esteem, would moderate attitude change in the
neutral prime conditions (Cooper & Duncan, 1971), but
that self-esteem would interact with the priming ma-
nipulation to influence attitude change when self-attri-
bute relevance was varied.
Method
Participants. Participants were 155 undergraduates
at the University of Arizona who participated in the
experiment for course credit. All had been pretested for
J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515 511
their level of self-esteem using the Rosenberg self-esteemmeasure (1979) during a mass pretest session of the
participant pool. As in previous research on self-esteem
and dissonance (Maracek & Mettee, 1972; Steele et al.,
1993), self-esteem was treated as a categorical variable in
the design and analysis, and only those with scores
falling in the upper (i.e., scores greater than 34) and
lower (i.e., scores less than 30) 30th percentile of the
Rosenberg scale were recruited. In addition, attitudestoward the essay topic (see below) were measured in the
pretest session using a 10 pt scale with the endpoints
‘‘strongly disagree’’ (1) and ‘‘strongly agree’’ (10). Only
those who reported negative attitudes toward the topic
(those who scored less than or equal to 5 on the scale)
were recruited (Elliot & Devine, 1994). A total of 82
with high self-esteem and 73 with low self-esteem com-
pleted the procedures described below.1
Procedure. Participants were contacted by phone and
invited to participate in two short studies on language
and cognition. They participated in groups of 1–3 but
completed the materials individually in a private cubicle.
The experimenter (who was unaware of the hypothesis
and level of self-esteem) explained that the first study
was designed to measure ‘‘how people think about po-
litical policy issues.’’ Participants were then assigned toa cubicle and provided with the study packet. The
written instructions in the packet explained that the
study intended to measure the relationship between
cognition and evaluation of political policy by having
participants express their opinions and beliefs about
different campus policy issues. The instructions in-
formed participants that the issue currently under in-
vestigation concerned a proposed decrease in fundingfor handicapped services at the university for the next
academic year (Blanton et al., 1997). Ostensibly, previ-
1 The assumption that participants with high and low self-esteem
would have different underlying self-views was tested in the current
participant sample. In a mass pretest, 1487 introductory psychology
students at the University of Arizona completed the Rosenberg self-
esteem scale (1979) and the Personal Attributes Questionnaire (or
PAQ, Pelham & Swann, 1989). They rated the self-descriptiveness of
the attributes of compassion and creativity on a 10 pt scale ranging
from 1 (exceptionally low) to 10 (exceptionally high). They also used
the same scale to rate the certainty and importance of their self-
descriptions. Participants were classified into high ((N ¼ 836) and lowself-esteem (N ¼ 418) groups using the upper and lower 30th percen-tiles. MANOVA analyses showed that participants with high self-
esteem rated compassion (M ¼ 7:87) and creativity (M ¼ 6:98) assignificantly more self-descriptive compared to participants with low
self-esteem (Ms ¼ 7:62 and 6.56, respectively), F ð2; 1251Þ ¼ 8:04,p < :0003. Furthermore, those with high self-esteem were significantly
more certain of their ratings on these traits (Ms ¼ 7:39 and 6.94,respectively) than participants with low self-esteem (Ms ¼ 6:86 and6.42), F ð2; 1251Þ ¼ 16:60, p < :0001. However, both groups rated the
traits as important to possess (high self-esteem Ms ¼ 7:53 and 6.48,respectively; low self-esteem Ms ¼ 7:34 and 6.56, respectively),F ð2; 1251Þ ¼ 2:02, p < :11. Thus, the self-knowledge structure of
participants with high and low self-esteem differed as assumed by the
SSM.
ous research had shown that a good way to collectopinions and beliefs about a topic was to instruct peo-
ple, no matter how they felt personally, to make argu-
ments on only one side of the issue. At this point, the
instructions for executing the uncompassionate behavior
varied as a function of the choice manipulation.
Choice manipulation. Participants randomly assigned
to the high-choice conditions were told that the decision
to argue for the decrease was up to them, but the re-search currently needed ‘‘strong forceful arguments’’ in
favor of the decrease. Participants in the low-choice
condition were told that in order to complete the study,
they needed to write ‘‘strong forceful arguments’’ in
favor of the decrease.
On the next page of the packet was a letter ostensibly
from the Committee for Undergraduate Education. The
letter stated that the committee might read their essay inorder to gauge student opinion about potential cuts in
the university budget for the next academic year (Coo-
per & Fazio, 1984). The final page provided the space
for them to write their essay and instructed them to
begin with the statement ‘‘The U of A should decrease
funding toward facilities and services for people with
physical disabilities on campus because. . .’’ When theycompleted the essay, they were instructed to place it inan envelop and alert the experimenter.
Priming manipulation. After they completed their
uncompassionate essay, the experimenter announced
that the first study was over and that they would now
complete the second study on language and cognition.
He then introduced a modified scrambled sentence test
designed to prime relevant, irrelevant, or neutral self-
attributes (Bargh, Chen, & Burrows, 1996). In eachcondition, the task consisted of eight blocks of words,
with each block containing one distracter word and the
target words. Participants were told the task consisted of
a group of words that could be unscrambled to form a
grammatically correct sentence. However, one word did
not belong in the word group and should be omitted
from the sentence they would form. The experimenter
then showed them an example at the top of the page.Participants were told their goal was to complete the
sentence scrambles as quickly as possible.
The relevant self-attribute prime condition was de-
signed to make accessible self-cognitions that were rel-
evant to the uncompassionate essay. The target
sentences were ‘‘I am a compassionate person,’’ ‘‘I try to
be thoughtful,’’ ‘‘Helping people is important (to prime
�helpful�),’’ and ‘‘I want to be considerate.’’ In the ir-relevant self-attribute prime condition, the target sen-
tences were ‘‘I am a creative person,’’ ‘‘I try to be
imaginative,’’ ‘‘Teaching people is important (to prime
�intelligent�),’’ and ‘‘I want to be flexible.’’ In the neutralself-attribute prime condition the sentences were ‘‘I am a
punctual person,’’ ‘‘I try to be quiet,’’ ‘‘Believing people
is important (to prime �trusting�),’’ and ‘‘I want to be
Fig. 1. The effects of self-esteem and experimental condition on atti-
tude change. LC, low choice and HC, high choice. Higher scores in-
dicate more change in the direction of support for a decrease in
funding for handicapped services. Means with different superscripts
differ significantly at p < :05 using Fisher�s LSD test.
512 J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515
cautious.’’ All priming conditions contained four fillersentences like ‘‘Sharpen a pencil’’ and ‘‘A phone rings.’’
Once participants completed the priming task, they
alerted the experimenter who then collected the primary
dependent measures.2
Dependent measures. The experimenter returned and
claimed that he forgot to give them a questionnaire
during the essay task. He handed participants a ques-
tionnaire with the statement ‘‘The U of A should de-crease funding toward facilities and services for people
with physical disabilities on campus.’’ Participants re-
sponded by circling a number on a 10 pt scale with the
endpoints ‘‘Strongly agree’’ (1) to ‘‘Strongly disagree’’
(10). On the next page, participants were asked to
complete a check of the choice manipulation by indi-
cating their agreement with the statement, ‘‘I felt free to
decline to write an essay for the opinion survey today’’on an 10 pt scale ranging from ‘‘strongly disagree’’ (1) to
‘‘strongly agree’’ (10). The experimenter left participants
alone to complete each questionnaire, and once they
alerted him, he collected the materials and announced
that the study was complete. All participants were then
fully debriefed about the purposes for the study.
Results
Choice manipulation check. To test the effectiveness of
the choice manipulation, a planned interaction contrast
was conducted that compared the perception of choice
reported by participants with high or low self-esteem in
the three high-choice conditions against the low-choice
control condition. The analysis showed a significantmain effect for condition, F ð1; 151Þ ¼ 5:98, p < :01,and no main or interaction effects for self-esteem (both
F ’s < 1). Participants in the high-choice conditions ex-perienced greater choice (M ¼ 7:19) than did partici-pants in the low-choice condition (M ¼ 5:85).Attitude change. The degree of attitude change was
computed by subtracting the attitude scores collected
during the pretest session from those collected after theessay task. The change scores were then subjected to a 2
(Self-esteem)� 4 (Experimental Condition: low-choiceneutral prime, high-choice neutral prime, high-choice
relevant prime, and high-choice irrelevant prime) AN-
OVA. The analysis revealed a significant main effect for
Condition, F ð3; 147Þ ¼ 2:88, p < :04, and a significantSelf-esteem�Condition interaction, F ð3; 147Þ ¼ 3:18,p < :03. Planned comparisons of the attitude changescores displayed in Fig. 1 were conducted to unpack the
2 Based on the trait list provided by Anderson (1968), valence
ratings were assigned to each self-attribute used in the priming
manipulations. With higher scores indicating more positively valenced
traits, the average for the set of relevant traits was 5.07, the average for
the set of irrelevant traits was 4.93, and the average for the set of
neutral attributes was 4.02.
meaning of the omnibus interaction between self-esteem
and the experimental conditions.
As expected, an ANOVA conducted on the attitude
change data in the neutral self-attributes prime condi-
tion revealed a significant main effect for choice,
F ð1; 147Þ ¼ 8:15, p < :005 but no main or interactioneffect for self-esteem, both F s < 1. When neutral self-attributes were primed, high-choice participants tendedto justify their advocacy more (M ¼ 2:64, SD ¼ 2:27)than did low-choice control participants (M ¼ 1:25,SD ¼ 1:85) and the effect was not moderated by self-esteem (Cooper & Duncan, 1971).
A planned interaction contrast between self-esteem
and the relevant versus irrelevant priming conditions on
the attitude change scores showed that attitude change
was significantly moderated by self-esteem when therelevance of the self-attributes was varied, simple inter-
action F ð1; 147Þ ¼ 9:10, p < :005. As seen in Fig. 1, aplanned comparison showed that as predicted, when
relevant attributes were primed, those with low self-es-
teem reported less attitude change (M ¼ 1:42, SD ¼ 1:98)compared to participants with high self-esteem (M ¼2:71, SD ¼ 1:74), F ð1; 147Þ ¼ 3:99, p < :05. In con-trast, when irrelevant positive attributes were primed,participants with high self-esteem reported significantly
less attitude change (M ¼ 1:39, SD ¼ 1:78) compared toparticipants with low self-esteem (M ¼ 2:87, SD ¼ 1:74),F ð1; 147Þ ¼ 5:63, p < :02. Thus, both the expectancyand resource role of self-esteem in dissonance emerged as
a function of the type of self-relevant thought invoked by
the priming manipulations.
Discussion
The results of the experiment provided support for
the hypotheses regarding the role of self-attribute rele-
vance on how self-esteem moderates attitude change
following a counter-attitudinal behavior. As predicted
J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515 513
by the processing assumptions of the Self-StandardsModel (SSM, Stone & Cooper, 2001), self-esteem
moderated attitude change following a counter-attitu-
dinal behavior, but only when cognitions about the self
were primed following the discrepant act. Moreover,
whether self-esteem functioned as a resource or as an
expectancy depended upon the type of self-relevant
thought induced by the priming manipulation. Overall,
the data indicate that the role of cognitions about theself and self-esteem in dissonance is a function of if and
how people think about themselves in the context of a
discrepant behavior.
When self-attributes that were irrelevant to the dis-
crepant act were primed, participants with high self-es-
teem showed less attitude change as compared to
participants with low self-esteem. This pattern supports
the perspective that high self-esteem, and the wealth ofpositive attributes it represents, can serve as a buffer or
resource against the discomfort people typically feel
following a discrepant behavior (Steele et al., 1993).
Participants with low self-esteem, in contrast, were
motivated to change their attitudes, presumably because
the primed self-attributes were less self-descriptive, and
therefore, less capable of serving as an affirmational
(Steele et al., 1993) or affective resource (Tesser, 2000).As predicted by the resource models, the more positive
self-attributes individuals have at their disposal, the less
they need to rely on self-justification to resolve behav-
ioral discrepancies.
The data also show that there are constraints on the
ability of people to use their positive self-attributes as
resources for dissonance reduction. In the high- and
low-choice control conditions in which neutral self-at-tributes were primed, participants with high and low
self-esteem reported the same levels of attitude change,
which was significantly moderated by perceptions of
choice, and not by the self-cognitions underlying their
level of self-esteem. This finding conceptually replicates
previous research on the relative importance of self-es-
teem and perceptions of responsibility for behavioral
outcomes in dissonance (Cooper & Duncan, 1971). Thefact that both self-esteem groups showed significant at-
titude change in the high-choice-neutral prime control
condition is compatible with the assumption that both
groups were focused on the inconsistency between spe-
cific attitudes or beliefs and their behavior (e.g., Fest-
inger, 1957; Harmon-Jones et al., 1996) or on the
aversive consequence of their behavior (Cooper & Fa-
zio, 1984). The lack of self-esteem differences in thecontrol conditions suggests that for self-esteem to
moderate dissonance, something in the context of a
discrepant behavior must make cognitions about the self
accessible. Otherwise, the cover story or procedure itself
may focus people primarily on specific attitudes or be-
liefs, and on the situational antecedents of the act (e.g.,
choice and forseeability), and dissonance processes may
proceed without the influence of idiosyncratic self-knowledge (Stone, in press; Stone & Cooper, 2001).
Furthermore, not just any positive self-relevant
thought will provide resources for dissonance reduction;
when positive self-attributes that were relevant to the
topic of the discrepant essay were primed, participants
with high self-esteem reported more attitude change
compared to participants with low self-esteem. This
pattern of self-esteem moderation reflects self-consis-tency for people with low self-esteem in the process of
dissonance, a phenomenon that has proved difficult to
replicate in past dissonance research (e.g., Ward &
Sandvold, 1963; see Swann, 1990). The data suggest that
when relevant positive attributes were primed following
the discrepant act, they activated the different self-ex-
pectancies for behavior held by participants with high
and low self-esteem (Aronson, 1999). For those withhigh self-esteem, the behavior was perceived as incon-
sistent with their self-expectancies for compassion,
which increased the need to change attitudes. This
supports the SSM assumption that in order for positive
self-attributes to provide resources for dissonance re-
duction, they must shift attention away from the dis-
crepancy; otherwise, the accessibility of positive
cognitions about the self may sustain or enhance dis-comfort and the motivation to change attitudes (Stone
& Cooper, 2001). In contrast, for participants with low
self-esteem, the relevant prime may have activated neg-
ative self-expectancies for compassionate behavior,
which reduced the perception of a discrepancy and the
need for attitude change.
It is important to acknowledge that whereas the data
are consistent with the processing assumptions of theSSM, no mediational data were gathered in the present
study that addresses alternatives to the processes speci-
fied in the model. For example, research indicates that
when self-attributes are primed, they may reduce the
need for attitude change by inducing trivialization of the
behavioral discrepancy (Simon, Greenberg, & Brehm,
1995). However, a trivialization explanation may be
hard pressed to account for the cross-over interactionbetween self-esteem and the relevance of the primed self-
attributes on attitude change. According to Simon et al.
(1995), trivialization only occurs when people are able to
bring to mind highly important cognitions, such as those
related to the self, prior to being offered the opportunity
to change attitudes. In the current study, not only were
the self-attributes chosen because they were rated in
previous research as equally desirable to possess (An-derson, 1968; Blanton et al., 1997), but during the pre-
test session, participants in the current study with high
or low self-esteem also rated the attributes as equally
important to their self-concepts. Thus, the manipulation
brought to mind equally important cognitions in each of
the self-attribute priming conditions, but attitude
change was still moderated by the interplay between
514 J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515
self-esteem and the relevance of the attributes to thebehavioral discrepancy.
It is also important to consider other processes by
which relevant attributes reduce the need for self-justi-
fication among people with low self-esteem. For exam-
ple, when self-expectancies were made accessible by the
relevant prime, it is possible that participants with low
self-esteem showed less need for attitude change because
they were relatively less certain of whether the behaviorwas inconsistent with their self-view of compassion.
However, other research has shown that self-consistency
for people with low self-esteem is more likely to occur
among those who are especially certain of their negative
self-views (e.g., Maracek & Mettee, 1972; Stone, in
press). Another possibility is that the observed self-
consistency effect for people with low self-esteem was a
function of how the attribute primes influenced the wayin which they labeled their arousal (Cooper & Fazio,
1984). The relevant primes may have caused participants
with low self-esteem to perceive a discrepancy between
their expectancies and behavior, but they labeled their
arousal as emotions like dejection or depression, which
reduced the motivation for attitude change, whereas
participants with high self-esteem labeled their negative
emotions as discomfort, tension, or guilt, which moti-vated attitude change to reduce their arousal (Galinsky,
Stone, & Cooper, 2000; Harmon-Jones, 2000; Higgins,
1987). Investigating the mediators of the observed in-
teraction between self-esteem and the accessibility of
self-attributes in the context of a discrepant act is an
important direction for future research.
Conclusions
Over 40 years after its initial publication (Festinger,
1957), the role of cognitions about the self in dissonance
continues to intrigue researchers. The research in this
paper found support for predictions made by both the
self-resource and the self-consistency perspectives on
how self-esteem moderates dissonance induced attitudechange. However, the present research also demon-
strated that cognitions about the self and self-esteem are
not always invoked by a discrepant act. This suggests, as
predicted by the self-standards model (Stone & Cooper,
2001), that people are capable of using their self-
knowledge to interpret and evaluate their behavior, but
whether they use cognitions about the self, specific at-
titudes, or norms to interpret behavior is a function ofwhich criterion is made salient in the situation. How
cognitions about the self influence the arousal and re-
duction of dissonance depends upon the type of self-
attributes and self-standards brought to mind when
people make a counter-attitudinal statement, a difficult
decision, or engage in other established dissonance
arousing behavior, including hypocrisy (Stone, Wie-
gand, Cooper, & Aronson, 1997). The process by whichpeople interpret their behavior, conclude it represents a
discrepancy, experience discomfort, and seek a way to
reduce it, may be more malleable than has been recog-
nized in previous revisions of dissonance theory.
References
Anderson, N. H. (1968). Likableness ratings of 555 personality-trait
words. Journal of Personality and Social Psychology, 9, 272–279.
Aronson, E. (1968). Dissonance theory: Progress and problems. In R.
Abelson, E. Aronson, W. McGuire, T. Newcomb, M. Rosenberg,
& P. Tannenbaum (Eds.), Theories of cognitive consistency: A
sourcebook (pp. 5–27). Chicago, IL: Rand McNally.
Aronson, E. (1999). Dissonance, hypocrisy, and the self-concept. In E.
Harmon-Jones, & J. Mills (Eds.), Cognitive dissonance: Progress on
a pivotal theory in social psychology (pp. 103–126). Washington,
DC: American Psychological Association.
Aronson, E., & Carlsmith, J. M. (1962). Performance expectancy as a
determinant of actual performance. Journal of Abnormal and Social
Psychology, 65, 178–182.
Aronson, J., Blanton, H., & Cooper, J. (1995). From dissonance to
disidentification: Selectivity in the self-affirmation process. Journal
of Personality and Social Psychology, 68, 986–996.
Aronson, J., Cohen, G. L., & Nail, P. R. (1999). Self-affirmation
theory: An update and appraisal. In E. Harmon-Jones, & J. Mills
(Eds.), Cognitive dissonance: Progress on a pivotal theory in social
psychology (pp. 127–148). Washington, DC: American Psycholog-
ical Association.
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social
behavior: Direct effects of trait construct and stereotype activation
on action. Journal of Personality and Social Psychology, 71, 230–
244.
Blanton, H., Cooper, J., Skurnik, I., & Aronson, J. (1997). When bad
things happen to good feedback: Exacerbating the need for self-
justification with self-affirmations. Personality and Social Psychol-
ogy Bulletin, 23, 684–692.
Brockner, J., Wiesenfeld, B. M., & Raskas, D. F. (1993). Self-esteem
and expectancy-value discrepancy: The effects of believing that you
can (or can�t) get what you want. In R. F. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 219–240). New York:
Plenum Press.
Brown, J. B. (1998). The self. New York, NY: McGraw-Hill.
Cooper, J., & Duncan, B. L. (1971). Cognitive dissonance as a function
of self-esteem and logical inconsistency. Journal of Personality, 39,
289–302.
Cooper, J., & Fazio, R. H. (1984). A new look at dissonance theory. In
L. Berkowitz (Ed.), Advances in experimental social psychology:
Vol. 17 (pp. 229–262). Hillsdale, NJ: Erlbaum.
Elliot, A. J., & Devine, P. G. (1994). On the motivational nature of
cognitive dissonance: Dissonance as psychological discomfort.
Journal of Personality and Social Psychology, 67, 382–394.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA:
Stanford University Press.
Galinsky, A. D., Stone, J., & Cooper, J. (2000). The reinstatement of
dissonance and psychological discomfort following failed affirma-
tions. European Journal of Social Psychology, 30, 123–147.
Gibbons, F. X., Eggleston, T. J., & Benthin, A. (1997). Cognitive
reactions to smoking relapse: The reciprocal relation of dissonance
and self-esteem. Journal of Personality and Social Psychology, 72,
184–195.
Glass, D. (1964). Changes in liking as a means of reducing cognitive
discrepancies between self-esteem and aggression. Journal of
Personality, 32, 531–549.
J. Stone, J. Cooper / Journal of Experimental Social Psychology 39 (2003) 508–515 515
Harmon-Jones, E. (2000). An update on cognitive dissonance theory,
with a focus on the self. In A. Tesser, R. B. Felson, & J. Suls
(Eds.), Psychological perspectives on the self: Vol. 1 (pp. 119–144).
Washington, DC: American Psychological Association.
Harmon-Jones, E., Brehm, J. W., Greenberg, J., Simon, L., & Nelson,
D. E. (1996). Is the production of aversive consequences necessary
to create cognitive dissonance? Journal of Personality and Social
Psychology, 70, 5–16.
Higgins, E. T. (1987). Self-discrepancy: A theory relating self and
affect. Psychological Review, 94(3), 319–340.
Maracek, J., & Mettee, D. (1972). Avoidance of continued success as a
function of self-esteem, level of esteem certainty, and responsibility
for success. Journal of Personality and Social Psychology, 22, 98–
107.
Pelham, B. W., & Swann, W. B., Jr. (1989). From self-conceptions to
self-worth: On the sources and structure of global self-esteem.
Journal of Personality and Social Psychology, 57, 672–680.
Rosenberg, M. (1979). Conceiving the self. New York, NY: Basic
Books.
Simon, L., Greenberg, J., & Brehm, J. (1995). Trivialization: The
forgotten mode of dissonance reduction. Journal of Personality and
Social Psychology, 68, 247–260.
Spencer, S. J., Josephs, R. A., & Steele, C. M. (1993). Low self-esteem:
The uphill struggle for self-integrity. In R. Baumeister (Ed.), Self-
esteem: The puzzle of low self-regard (pp. 21–36). New York:
Plenum Press.
Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the
integrity of the self. In L. Berkowitz (Ed.), Advances in
experimental social psychology: Vol. 21 (pp. 261–302). Hillsdale,
NJ: Erlbaum.
Steele, C. M., & Lui, T. J. (1983). Dissonance processes as self-
affirmation. Journal of Personality and Social Psychology, 45, 5–19.
Steele, C. M., Spencer, S. J., & Lynch, M. (1993). Dissonance and
affirmational resources: Resilience against self-image threats.
Journal of Personality and Social Psychology, 64, 885–896.
Stone, J. (1999). What exactly have I done? The role of self-attribute
accessibility in dissonance. In E. Harmon-Jones, & J. Mills (Eds.),
Cognitive dissonance: Progress on a pivotal theory in social
psychology (pp. 175–200). Washington, DC: American Psycholog-
ical Association.
Stone, J. (2001). Behavioral discrepancies and construal processes in
cognitive dissonance. In G. Moskowitz (Ed.), Cognitive social
psychology: The Princeton symposium on the legacy and future of
social cognition (pp. 41–58). Hillsdale, NJ: Erlbaum.
Stone, J. (in press). Self-consistency for low self-esteem in dissonance
processes: The role of self-standards. Personality and Social
Psychology Bulletin.
Stone, J., & Cooper, J. (2001). A self-standards model of cognitive
dissonance. Journal of Experimental Social Psychology, 37, 228–243.
Stone, J., Wiegand, A. W., Cooper, J., & Aronson, E. (1997). When
exemplification fails: Hypocrisy and the motive for self-integrity.
Journal of Personality and Social Psychology, 72, 54–65.
Sorrentino, R. M. (1990). To be adored or to be known? The interplay
of self-enhancement and self-verification. In E. T. Higgins (Ed.),
Motivation and cognition: Vol. 2 (pp. 408–450). New York, NY:
Guilford Press.
Tesser, A. (2000). On the confluence of self-esteem maintenance
mechanisms. Personality and Social Psychology Review, 4, 290–299.
Tesser, A., & Cornell, D. P. (1991). On the confluence of self processes.
Journal of Experimental Social Psychology, 27, 501–526.
Thibodeau, R., & Aronson, E. (1992). Taking a closer look:
Reasserting the role of the self-concept in dissonance theory.
Personality and Social Psychology Bulletin, 18, 591–602.
Ward, W. D., & Sandvold, K. D. (1963). Performance expectancy as a
determinant of actual performance: A partial replication. Journal
of Abnormal and Social Psychology, 67, 293–295.
Further reading
Cooper, J. (1999). Unwanted consequences and the self: In search of
the motivation for dissonance reduction. In E. Harmon-Jones, & J.
Mills (Eds.), Cognitive dissonance 40 years later: Revival with
revisions and controversies (pp. 149–174). Washington, DC: Amer-
ican Psychological Association.
Festinger, L., & Carlsmith, J. M. (1959). Cognitive consequences of
forced compliance. Journal of Abnormal and Social Psychology, 58,
203–210.
Harmon-Jones, E. (1999). Toward an understanding of the motivation
underlying dissonance effects: Is the production of aversive
consequences necessary. In E. Harmon-Jones, & J. Mills (Eds.),
Cognitive dissonance: Progress on a pivotal theory in social
psychology (pp. 71–99). Washington, DC: American Psychological
Association.
Harmon-Jones, E., & Mills, J. (1999). Cognitive dissonance: Progress
on a pivotal theory in social psychology. Washington, DC: American
Psychological Association.