When Self-Affirmations Reduce Defensiveness: Timing Is Key

Abstract

Research on self-affirmation has shown that simple reminders of self-integrity reduce people's tendency to respond defensively to threat. Recent research has suggested it is irrelevant whether the self-affirmation exercise takes place before or after the threat or the individual's defensive response to it, supposedly because the meaning of threats is continuously reprocessed. However, four experiments revealed that affirmations may be effective only when introduced prior to the initiation of a defensive response. Affirmations introduced before threatening feedback reduced defensive responding; affirming after a threat was effective in reducing defensiveness only if the defensive conclusion had yet to be reached. Even though threats may activate a defensive motivation, the authors' results suggest that defensive responses may not be spontaneous and may be prompted only when suggested by the dependent measures themselves. This explains why some affirmations positioned after threats are effective in reducing defensiveness. Implications for self-affirmation theory are discussed.

Full text

Article
Personality and Social
Psychology Bulletin
36(7) 947 –959
© 2010 by the Society for Personality and
Social Psychology, Inc
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0146167210369557
http://pspb.sagepub.com
When Self-Affirmations Reduce
Defensiveness: Timing Is Key
Clayton R. Critcher1, David Dunning2, and David A. Armor3
Abstract
Research on self-affirmation has shown that simple reminders of self-integrity reduce people’s tendency to respond defensively
to threat. Recent research has suggested it is irrelevant whether the self-affirmation exercise takes place before or after the
threat or the individual’s defensive response to it, supposedly because the meaning of threats is continuously reprocessed.
However, four experiments revealed that affirmations may be effective only when introduced prior to the initiation of a
defensive response. Affirmations introduced before threatening feedback reduced defensive responding; affirming after a
threat was effective in reducing defensiveness only if the defensive conclusion had yet to be reached. Even though threats may
activate a defensive motivation, the authors’ results suggest that defensive responses may not be spontaneous and may be
prompted only when suggested by the dependent measures themselves. This explains why some affirmations positioned after
threats are effective in reducing defensiveness. Implications for self-affirmation theory are discussed.
Keywords
self-affirmation, defensiveness, threat, self-integrity, dissonance
Received March 5, 2009; revision accepted December 18, 2009
Whether from bosses, spouses, or some other critical source,
people occasionally receive threatening feedback about
their competence and character. Researchers have identified
an eclectic array of defensive strategies that people use to
dampen the impact of unfavorable information on their
self-integrity, thereby allowing people to maintain unreal-
istically positive illusions about themselves and their
place in the world (Alicke, Klotz, Breitenbecher, Yurak, &
Vredenburg, 1995; Dunning, 2003; Taylor & Brown, 1988).
For example, people engage in downward social compari-
sons (Spencer, Fein, & Lomore, 2001; Taylor & Lobel,
1989), view their own successes as unique and shortcom-
ings as commonplace (Campbell, 1986; Marks, 1984), and
dissociate their own specific deficits from broader implica-
tions for the self (Beauregard & Dunning, 2001; Wentura &
Greve, 2003). Although these defensive strategies might
help one brush off a crush’s rejection, a less-than-flattering
teaching evaluation, or a personal insult heard through the
grapevine, these same defensive strategies may also prevent
people from attaining accurate assessments of their own
strengths and shortcomings (Dunning, 2005; Dunning,
Heath, & Suls, 2004; Sedikides, Green, & Pinter, 2004). It
would at times be useful to be able to “switch off” one’s
defensive shield so that threatening information and ideas
could be more objectively considered (Critcher, Helzer, &
Dunning, in press; Radcliffe & Klein, 2002).
The psychology of self-affirmation suggests that even
simple reminders of self-worth may be sufficient to flip this
switch—to reduce the normal tendency to respond to threat
defensively—so that people can incorporate useful but
potentially unflattering information about themselves (also
see Trope & Pomerantz, 1998). Self-affirmation theory
(Sherman & Cohen, 2006; Steele, 1988) proposes that any
strategy that restores the integrity of the self after a psychic
assault should alleviate the impact of the new threat and thus
eliminate the need to respond defensively. In other words,
threats to the self need not be dealt with at the site of the
psychic wound but can be healed more indirectly by calling
to mind valued aspects of one’s identity in some other life
domain, even though these identity considerations bear no
relation to the source of the threat.
Consistent with this argument, self-affirmations have
been shown to reduce a number of defensive processes (for a
review, see Sherman & Cohen, 2006). Self-affirmed partici-
pants provide help to others even when that other person’s
success is threatening (Tesser, Martin, & Cornell, 1996),
more objectively and less defensively evaluate the arguments
of an ideological opponent (Cohen, Aronson, & Steele,
1University of California, Berkeley
2Cornell University, Ithaca, NY, USA
3San Diego State University, San Diego, CA, USA
Corresponding Author:
Clayton R. Critcher, University of California, Berkeley, Haas School of
Business, Department of Marketing, Berkeley, CA 94720
Email: clayton.critcher@aya.yale.edu
at CORNELL UNIV on June 24, 2010 http://psp.sagepub.comDownloaded from

948 Personality and Social Psychology Bulletin 36(7)
2000), and are more open to self-relevant information about
risks to their health, such as the link between cancer and
alcohol consumption (Harris & Napper, 2005; Reed &
Aspinwall, 1998).
The number and diversity of these demonstrations serve
both as a sign of the strength of the self-affirmation model
and a possible source of concern. Self-affirmations clearly
curtail defensive responding in many circumstances, but
there may be reason to suspect that several recent claims of
the effects of self-affirmations may be overstated and that
self-affirmations may be subject to a more significant bound-
ary condition than past research has suggested (also see
Sherman et al., 2009). In particular, the defense-inhibiting
power of self-affirmations may depend on a variable that, to
date, has been viewed as relatively unimportant—the timing
of the affirmation in relation to the defensive response.
The importance of timing in self-affirmation processes
can be predicted from the convergence of two theoretical
notions that underlie current understanding of the psychol-
ogy of defense: the substitutability of self-esteem mainte-
nance mechanisms and the notion that people seek to maintain
and not necessarily maximize their positive self-views. From
the substitutability perspective, self-affirmation is not in
itself a special process but rather one tool in an arsenal of
psychic tactics that people use to bolster and maintain their
self-esteem, such as finding worse-off others to compare
themselves against and dispelling dissonant thoughts by
altering belief in those thoughts (for a review, see Tesser,
2000). Importantly, data suggest that these tactics for self-
esteem maintenance are quite interchangeable: One tactic
can often be used to substitute for another (Hart, Shaver, &
Goldberg, 2005; Tesser, 2000).
The second key idea, implicit in the notion of the substi-
tutability of self-maintenance processes, is that there is a
natural limit to people’s positivity strivings. Results from
several independent research programs suggest that people
do not self-enhance at all costs (Markus & Wurf, 1987) but
are motivated to maintain rather than to maximize a positive
self-image (Tesser, 1988) and that their “psychological
immune system” kicks in only when the self experiences
some degree of threat (Aspinwall, 1997; Tesser, 1988). This
self-image maintenance perspective suggests that, once a
positive self-image has been restored—whether through a
self-affirmation or through a defensive process—there will
be no further need to restore the self as there will be no threat
for any subsequent affirmation to undo.
The combination of these two notions suggests two
boundary conditions that may limit the impact of self-
affirmations, one of which has already received support in
the affirmation literature, one of which has on its surface
been consistently contradicted. First, if affirmations target
defensiveness in response to threat, then affirmations should
affect only the responding of threatened participants.
Several studies support this threat–repair stipulation: that
affirmations do halt defensive responding only of those for
whom a valued self-relevant domain has been threatened
(e.g., Harris & Napper, 2005).
The second condition, building on the first, centers on the
timing of the self-affirmation exercise. If affirmations
mostly alleviate threats to the self, then affirmations should
not affect those who have already engaged in an act of
defensiveness—that is, on those who have already removed
the threat. In short, from these two perspectives, there is no
reason to expect that affirmations would have a retroactive
effect on defensive responses previously made. Affirmations
should reduce any remaining inclination to be defensive but
should not undo products of past defensiveness. This was the
key hypothesis we explored in this research: Do affirma-
tional interventions have an impact once a person has had a
chance to respond defensively? According to our perspec-
tive, they should not.
Does Timing Matter? Competing Perspectives
On close inspection, our timing hypothesis may have been
implied in Steele’s (1988) formulation of self-affirmation
theory. In a speculative (but, we think, prescient) para-
graph, Steele wrote, “Self-affirming thoughts may be an
effective means of reducing thought-distorting defense
mechanisms such as denial and rationalization,” but these
effects “may depend, at least partially, on what other
thoughts about the self are salient at the time the informa-
tion is processed [italics added]” (p. 290). This statement
clearly implies that affirmations may be expected to be
effective only in prospect—that affirming thoughts must be
in place as one thinks through the implications of a threat
to the self.
But much evidence appears to speak against the tim-
ing hypothesis. For example, in a recent meta-analysis,
McQueen and Klein (2006) found that the effect of affirma-
tions placed after the threat was equivalent to that of those
placed before the threat. Data such as these have led other
researchers to modify Steele’s (1988) argument to justify
why affirmations should be able to undo defensive responses.
Cohen et al. (2000), for example, argued that affirmations
could do more than prevent defensive responding, suggest-
ing that affirmations could also actively undo defensive
conclusions that had already been drawn. In differentiating
the defensiveness-reducing effects of their pre-threat affir-
mations from the defensiveness-reducing effects of their
post-threat affirmation, Cohen et al. stated, “The affirma-
tion may reduce on-line defensiveness processing, at the
time of encoding . . . or [a post-threat affirmation] may
attenuate memory-based defensive processing” (p. 1162).
Cohen and colleagues posited that threats to the self are
continuously reprocessed and that self-affirmations enable
people to reconsider and reverse defensive conclusions
even after they have been drawn.
at CORNELL UNIV on June 24, 2010 http://psp.sagepub.comDownloaded from

Critcher et al. 949
Sherman, Nelson, and Steele (2000) advanced a similar
argument to explain the effectiveness of their post-threat affir-
mation, “given that [the threat] is continually reprocessed and
reformulated until the participant is asked to report on the
attitude” (p. 1056).
The reprocessing hypothesis is thus consistent with exist-
ing data but remains stubbornly inconsistent with the theo-
retical principles motivating our conclusion that timing
matters. In contrast to the reprocessing-based claim that the
timing of the self-affirmation is irrelevant (Cohen et al.,
2000; Sherman et al., 2000), we hypothesize that self-
affirmations will work only in prospect—that is, only if the
affirmation is in place before the threat itself or before defen-
sive response has been initiated. If the defensive reaction has
already taken place, self-affirmation may reduce any remain-
ing motivation to respond defensively, but the self-affirmation
should not “undo” a defensive reaction that has already been
crafted, according to the logic of substitutability and of self-
image maintenance.
How, then, can this timing hypothesis be reconciled with
previous research showing that post-threat affirmations can
be effective in curtailing a defensive response? We propose
that the critical moderator is the timing of the affirmation in
relation to a threat response rather than the timing of that
affirmation in relation to the threat itself. To date, research
on the effects of post-threat affirmations may have obscured
this distinction. We argue that post-threat affirmations are
at times successful in reducing defensiveness not because
threats are continually reprocessed but because many
post-threat affirmations are actually positioned before the
response to the threat commences. In particular, we suggest
that post-threat affirmations are often successful only because
the defensive processing measured in affirmation studies
does not start until participants are presented with the defen-
siveness measure itself.
Overview of Studies
Four studies examine whether the timing of self-affirmations,
in relation to the initiation of defensive responding, matters.
In Study 1, we examined whether self-affirmation exercises
at times reduce defensive responding more effectively when
they come before rather than after a threat, establishing that
timing does matter. In Studies 2a and 2b, we examined the
timing of affirmation exercises more carefully to see if an
affirmation is effective after the threat has occurred but
before the person has had a chance to respond to it. According
to our analysis, post-threat affirmations can be effective if
people have yet to initiate a defensive response. However, if
we can prompt people to respond defensively at an earlier
moment before confronting an affirmation intervention, that
intervention should lose its power to reduce defensiveness.
Study 3 examined the same issues in the realm of cognitive
dissonance.
Study 1
Participants were asked to take one of two versions of a test
purporting to measure “integrative orientation ability.” We
constructed a particularly difficult version of the test meant
to inspire threat. Participants were randomly assigned to
self-affirm just before the test (pre-threat affirmation), just
after scoring their own test (post-threat affirmation), or not at
all (no affirmation). We predicted that the affirmation would
be effective in reducing defensiveness in the pre-threat con-
dition, relative to the no affirmation condition, but not in the
post-threat affirmation condition. If, however, timing does
not matter, then both pre- and post-threat affirmations should
be equally effective in leading participants to less defen-
sively accept the implications of their performance in the
hard test condition.
As a supplemental goal, we also examined the impact of
self-affirmation for individuals not threatened. We included
another version of the test that was more moderately demand-
ing and was meant to be nonthreatening (easy test). We pre-
dicted that self-affirmation would have no impact on people’s
defensive responding in this condition, given the view that
people more enthusiastically seek to maintain their self-esteem
once threatened than they do to maximize their self-esteem at
all times. An impact of affirmation on the threatened, but not
on the nonthreatened, would support one of the premises on
which our timing hypothesis was based.
Method
Participants and design. In exchange for either class credit
or $5, 184 Yale University undergraduates took part in the
study. The experiment used a 3 (affirmation: pre-test affir-
mation, post-test affirmation, or no affirmation) ! 2 (test dif-
ficulty: hard or easy) factorial design.
Procedure: Test of integrative orientation ability. All partici-
pants were given a test that ostensibly evaluated people’s
ability “to think creatively” and “to find unusual solutions to
problems.” We modified the items on Mednick’s (1962)
Remote Associates Test to create two versions—one that was
quite difficult (and thus threatening) and one that was much
easier (less threatening). Each item on these 15-item tests
provides participants with three words and asks the respon-
dent to generate a fourth word that relates to each of the other
three words. A representative item from the easy test is
“Chocolate—Fortune—Tin”; a representative item from the
hard test is “Soap—Shoe—Tissue.”1 After a 4-min testing
period, participants were given correct solutions and were
asked to score their own tests.
Procedure: Self-affirmation. Participants assigned to one of
the self-affirmation conditions completed a self-affirmation
task adapted from previous research (Cohen et al., 2000;
Shira & Martin, 2005). In this task, participants were given a
list of eight domains (e.g., religious fulfillment, physical

950 Personality and Social Psychology Bulletin 36(7)
health) and asked to rank the domains in order of personal
importance. Because affirming on the threatened domain can
produce a backfiring defensiveness-exaggerating effect
(Blanton, Cooper, Skurnik, & Aronson, 1997; Sivanthan,
Molden, Galinsky, & Ku, 2008), none of the domains related
to academic or intellectual achievement. Participants then
wrote a paragraph about their most valued domain’s impor-
tance in their lives. In place of the self-affirmation task, con-
trol participants received a list of exotic-sounding jelly beans
and were asked to rank them by how tasty they believed the
flavors would be.2 Pre-test affirmation participants com-
pleted the affirmation just before taking the test. Post-test
affirmation participants completed the exercise just after
scoring their test but before the measures of defensiveness.
To control for the delay the post-test affirmation caused
between the test and the critical measures, both pre-test affir-
mation and control participants completed the jelly bean task
after scoring their own test.
Procedure: Measures of defensiveness. Participants estimated
the average score, out of 15, that they thought students at
Yale University would achieve on the test. By minimizing
their estimates of the average score, participants could defen-
sively paint their own performance in a better light (Klein &
Kunda, 1989). Second, we obtained an assessment of how
much participants incorporated feedback into their self-
concepts. To make this measure less transparent, we pre-
sented it as part of a seemingly unrelated study in which
participants were asked to rate themselves and an acquain-
tance on a variety of traits. Within this “second study,” par-
ticipants were asked to rate their own creativity on a scale
from 1 (not at all) to 9 (completely). As the instructions to
the integrative orientation ability test had indicated that the
test was a measure of creative thinking, higher self-ratings
of creativity evidenced greater defensiveness.
Results
Manipulation check. Test scores were submitted to a 3
(affirmation: pre-test affirmation, no affirmation, or post-test
affirmation) ! 2 (test version: hard or easy) ANOVA. As
expected, participants performed significantly worse on the
hard test (M = 2.8) than on the easy test (M = 10.9), F(1, 178) =
596.71, p " .001. Neither the main effect of affirmation nor
the interaction with test condition was significant, F(2, 178) =
1.89, p # .15, and F(2, 178) = 1.59, p # .20, respectively, sug-
gesting that being affirmed did not affect one’s performance.
Influence of threat and timing. To maximize power in
detecting differences in defensiveness, we standardized each
of the two measures of defensiveness, reverse scored the per-
ceived average score, and then summed the two measures.
According to the threat–repair stipulation, an ANOVA on the
defensiveness index should reveal a threat (easy or hard test)
by affirmation (pre-test affirmation, post-test affirmation, or
no affirmation) interaction, with any defensive-reducing
effects of pre-test (and possibly post-test) affirmations
expected only in the threat (hard-test) conditions. As depicted
in Figure 1, the effect of the affirmation manipulation did
depend on whether participants took the hard or easy test,
F(2, 173) = 4.50, p = .01. As predicted, the influence of affir-
mation was significant for those who took the hard (threaten-
ing) test, F(2, 96) = 5.86, p = .004, but not for those who took
the easy one, F(2, 77) = 1.05, p # .35.
To distinguish between the timing and reprocessing
hypotheses, we then placed responses in the hard test condi-
tion under closer scrutiny. We conducted two contrasts to
determine whether affirmations were effective only when
administered before the threat (pre-threat: –2; post-threat and
control: +1) or whether affirmations were effective regard-
less of timing (pre-threat, post-threat: –1; control: +2). These
contrasts suggested that affirmations were effective only if
encountered before the threat, t(96) = 3.02, p = .003.3 The
competing contrast inspired by the reprocessing hypothesis
was not significant (t " 1). Follow-up comparisons found
that pre-threat affirmations marginally reduced defensiveness
compared to the control condition, t(96) = 1.83, p = .07, but
crucially were more effective in reducing defensiveness than
were post-threat affirmations, t(96) = 3.42, p = .001. Post-
threat affirmed participants unexpectedly displayed margin-
ally more defensiveness than those in the control condition,
t(96) = 1.70, p = .09. There was no hint of this pattern in
future studies, so we hesitate to speculate on this effect.
Discussion
The results of Study 1 are consistent with the timing hypoth-
esis. Only participants who had been affirmed prior to taking
the difficult test (and thus prior to the receipt of critical
Figure 1. Defensiveness as a function of affirmation condition
and test difficulty (Study 1)
Values are z score composites and thus have no absolute meaning; 0 is not
“no defensiveness.” Comparisons within the easy and hard test conditions
are meaningful comparisons of defensiveness; comparisons between these
conditions are not. Instead, between-test-condition differences reflect a
mix of defensiveness and actual informational differences.

Critcher et al. 951
feedback) showed a reduction in defensive responding.
Contrary to the notion that defensive conclusions are contin-
ually reassessed, self-affirmations did not undo defensive
conclusions that, presumably, had already been formed. An
act of defensiveness does not appear to be merely a temporary
brace that bolsters the damaged self until a self-affirmation’s
calming force can allow for dispassionate reprocessing. Instead,
affirmations, in restoring self-integrity, appear to obviate the
need for further acts of defensiveness while leaving already
formulated defensive conclusions firmly in place.
Studies 2a and 2b
Our timing hypothesis does not preclude the possibility that
post-threat affirmations will, at times, be effective in prevent-
ing defensiveness. A post-threat affirmation could be effec-
tive in eliminating defensive processes if (and perhaps only
if) these processes are not executed until the participant is
given a chance to respond defensively. That is, the timing of
the threat per se does not matter as much as the timing of the
individual’s response to that threat. An affirmation taking
place after a threat can be effective if it precedes the indi-
vidual’s response to that threat. However, according to the
reprocessing hypothesis, a post-threat affirmation should
still be effective even if it were introduced after the individ-
ual has responded to a threat.
Past work on self-affirmation has shown that post-threat
affirmations can be effective in quelling defensive responding
(e.g., McQueen & Klein, 2006). It is possible that the specific
defensive strategies under investigation in these past studies
may not have been spontaneous. That is, people may not have
initiated their defensive responding until given the post-threat
questionnaire that allowed them the opportunity to be defen-
sive. Thus, there may have been gaps in time between the
threat and participants’ reactions to it in which self-affirmation
could still be effective. If this is the case, then previous studies
using seemingly effective post-threat affirmations may not
have provided evidence of reprocessing at all but may instead
have demonstrated the effect of an affirmation on a defensive
response that had yet to begin.
To more directly test the effects of the timing of a self-
affirmation in relation to a threat response and not just in
relation to the threat itself, it was necessary to experimen-
tally disentangle participants’ awareness of a defensive
opportunity from the time at which they could provide a
defensive response. Studies 2a and 2b sought to do just that.
In Study 2a, we created conditions in which defensive reac-
tions to threat would likely not be spontaneously initiated
(giving post-threat affirmations a chance to be effective) by
making opportunities to respond defensively nonobvious
and by limiting the time between the experience of threat and
the administration of the post-threat affirmation.
Then, in Study 2b, we not only replicated these conditions
but also added a foreshadowing condition that informed
participants, before completing a self-affirmation exercise, of
defensive opportunities we were going to present them on a
follow-up questionnaire. This foreshadowing was intended to
unconfound the presentation of the defensive measures (and
thus the suggestion of particular strategies for ego repair)
from the elicitation of the defensive response. We hypothe-
sized that post-threat affirmations would reduce defensive-
ness when these affirmations were introduced before the threat
response (Study 2a) but not when possibilities for defensive
processing had already been brought to the attention of par-
ticipants (Study 2b). Although Study 1 used threats to intel-
lectual abilities, Studies 2a and 2b used a threat to participants’
ability to maintain and foster personal relationships (Baumeister
& Leary, 1995).
Study 2a
In Study 2a, we changed the nature of the threat to better
control when participants might start a defensive response
to threat. In Study 1, it seemed likely that participants in the
high-threat (hard-test) conditions knew they were doing
poorly for the entire duration of the test-taking experience
(they were not able to come up with free-response answers),
and thus defensiveness may have set in early. To prevent
this, in Studies 2a and 2b participants were given a test that
did not offer performance-related cues, and evaluative feed-
back was withheld until after the test. Second, we chose
indirect, nonobvious measures of defensiveness that partici-
pants were unlikely to engage in until made aware of them
in the questionnaire.
For Study 2a, we predicted that affirmation, relative to a
control condition, would quell defensive responding regard-
less of whether participants affirmed before or after receiv-
ing the threatening feedback. If this result emerged, we could
test in Study 2b whether the post-threat affirmation was
effective because defensive processing had yet to be engaged.
Method
Participants and design. In exchange for extra credit in psy-
chology and human development courses, 76 Cornell
University undergraduates took part in our study. Participants
were randomly assigned to one of three high-threat affir-
mation conditions: a pre-threat affirmation (administered
before the test), a post-threat affirmation (administered after
both the test and feedback but before measures of defensive-
ness), or a no affirmation control.
Procedure: Test of interpersonal perception ability. All par-
ticipants completed the 15-item version of the Interpersonal
Perception Task (IPT-15; Costanzo & Archer, 1989). The
IPT-15 is a video-based test that participants were told
would evaluate their skill at “accurately perceiving verbal
and nonverbal interpersonal cues,” which was said to be a
crucial skill in “fostering and maintaining interpersonal
at CORNELL UNIV on June 24, 2010 http://psp.sagepub.comDownloaded from

952 Personality and Social Psychology Bulletin 36(7)
relationships.” The test comprises 15 scenes that participants
watched on a computer. After each scene, the participants
answered a multiple-choice question about the scene. For
example, in one scene, the test taker watches a short conver-
sation between a man and a woman sitting at a table. Based
on verbal and nonverbal cues, participants must decide whether
the pair are siblings or newly formed friends. Following the
20-min test, the experimenter informed participants that to
protect their anonymity, test takers would score their own
tests. After exchanging participants’ blue pens for red pens to
prevent dishonesty, the experimenter provided each test
taker with an answer key. In actuality, the responses on the
answer key had been randomly generated, creating falsely
negative feedback for the participants. Participants rated on
a 9-point scale whether their performance was much worse
(1) or much better (9) than expected.
Procedure: Self-affirmation. As in the previous study, the
self-affirmation manipulation asked participants to write
about an important value. To avoid having participants affirm
themselves on a threat-relevant domain, the value “helping
others” was replaced with “academic success.” Participants
in the pre-threat affirmation condition were asked to com-
plete the affirmation before reading the test instructions.
Those in the post-threat affirmation condition were asked to
complete the affirmation immediately after scoring the test
but prior to the measures of defensiveness. Participants
assigned to the no affirmation control condition were pre-
sented with the names of different candle scents (e.g., coco-
nut smoothie, baby powder) and were asked to rank order
these scents in terms of preference. Control participants, in
addition to pre-threat affirmation participants, completed the
candle-rating task while post-threat affirmation participants
were completing the self-affirmation.
Procedure: Measures of defensiveness. We chose four mea-
sures of defensiveness that seemed likely to capture nons-
pontaneous processes. First, we asked participants to
evaluate their own intelligence so as to give participants an
opportunity to “compensate” for their poor performance by
exaggerating their ability in another domain (Brown &
Smart, 1991). Second, given that the test was supposed to
tap their ability to foster and maintain interpersonal rela-
tionships, we asked participants to estimate how many
friends they had and how many acquaintances they had.
Given that participants were expected to still be smarting
about their negative score on a test that was supposed to tap
their ability to foster and maintain interpersonal relation-
ships, we reasoned that asking about their personal relation-
ships would provide them with an opportunity to reaffirm
their (previously threatened) ability to form and maintain
many personal relationships. Finally, we told participants
that there was another test that had been matched for diffi-
culty with the version that they had just taken. We asked
them what score they thought they would receive on that
test. For each measure—self-rated intelligence, number of
friends, number of acquaintances, and score on a retest—
higher values reflect greater defensiveness.
Results
Perceptions of test performance. A preliminary analysis pro-
vided assurance that the random-response answer key pro-
vided falsely negative feedback to participants. Participants’
responses on the rating of performance relative to their
expectation showed that they tended to rate their score in the
“worse than expected” side of the scale (M = 3.9 vs. a scale
midpoint of 5.0), t(75) = 6.63, p " .001.
Measures of defensiveness. As in Study 1, we began by
standardizing and summing the measures of defensiveness.
The means by condition are displayed in Figure 2. We con-
ducted two contrasts, one testing whether affirmations would
reduce defensiveness, regardless of timing (pre-threat, post-
threat: –1; control: +2), and one testing whether only the pre-
threat affirmation would be effective (pre-threat: –2; control,
post-threat: +1). As expected, the former was significant,
t(72) = 3.50, p = .001, but not the latter, t(72) = 1.43, p # .15.
In contrast to Study 1, participants in both the pre-threat and
post-threat affirmation conditions responded less defensively
to the negative test score than did participants in the control
condition, t(72) = 2.88, p = .01, pre- versus control; t(72) =
3.26, p = .002, post- versus control. Among affirmed partici-
pants, the pre-threat affirmation was no more effective than
the post-threat affirmation (t " 1).
Study 2b
In Study 2b, we sought to prompt some participants to initiate
defensive responding just before they encountered the affir-
mation manipulation. According to our analysis, if partici-
pants initiate that defensive processing before being given a
chance to self-affirm, the self-affirmation exercise will fail to
Figure 2. Defensiveness as a function of affirmation condition
(Study 2a)
Values are z score composites and thus have no absolute meaning;
0 is not “no defensiveness.”

Critcher et al. 953
reduce the degree of defensive processing participants dis-
play. To do this, we replicated the control and post-threat
affirmation conditions from Study 2a but replaced the pre-
threat affirmation condition with a foreshadowing condi-
tion. Participants in this condition, like participants in the
post-threat affirmation condition, were asked to complete a
self-affirming essay immediately after scoring their own test.
The only difference was that participants in the foreshadow-
ing condition were told, just prior to writing their self-affir-
mation essay, what questions they would be answering after
completing the “writing task.” We presumed this would initi-
ate defensive processing, which would fail to be undone by
the affirmation exercise. By this reasoning, participants would
show just as much defensiveness in the foreshadowing condi-
tion as participants in the no affirmation condition. The defen-
siveness of participants in the original post-threat condition,
however, would be much lower. If, on the other hand, the
reprocessing hypothesis is correct, then the foreshadowing
manipulation should be irrelevant, and those in the foreshad-
owing and post-threat affirmations conditions should display
less defensiveness than those in the control condition.
Method
Participants and design. In exchange for extra credit in their
psychology and human development courses, 84 Cornell
University undergraduates took part in the study. Participants
were randomly assigned to a post-threat affirmation condi-
tion, a foreshadowing condition (in which the dependent
measures were foreshadowed just prior to the post-threat
affirmation), or a no affirmation control condition.
Procedure. The threat and measures of defensiveness were
the same as those used in Study 2a, as was the control condi-
tion and the (nonforeshadowed) post-threat affirmation con-
dition. A third group of participants was randomly assigned
to a foreshadowing condition, which replaced the pre-threat
affirmation condition. Participants in this foreshadowing
condition, like participants in the post-threat affirmation
condition, completed the self-affirmation manipulation after
scoring their own test with the false answer key. How the
foreshadowing condition differed was that the experimenter
delivered the following lines just prior to the affirmation
manipulation:
You will have two tasks remaining today. First, you
will complete a brief writing task. Second, we will ask
you some follow-up questions to the test you com-
pleted today. We will ask you to make some test-specific
judgments like how you think you would score on a
comparably difficult alternate version of this test.
Because the test is related to interpersonal perception
ability, we’ll ask you how many friends and acquain-
tances you have. And also, we’ll have you rate yourself
on a few domains, like intelligence.
It was expected that foreshadowing would put the wheels of
defensive processing in motion and that, once these processes
had been initiated, self-affirmations would be powerless to
stop them. That is, participants in this condition would show
just as much defensiveness as participants in the no affirmation
condition and heightened levels relative to those in the post-
threat condition.
Results
Perceptions of test performance. As in Study 2a, partici-
pants’ ratings of their performance (M = 3.8) were signifi-
cantly below the midpoint (5.0), suggesting that participants
had experienced their performance as falling short of expec-
tations, t(83) = 7.35, p " .001.
Measures of defensiveness. As in Study 2a, we standardized
and averaged the responses to the measures of defensive-
ness. The means by condition are depicted in Figure 3. We
again tested two contrasts: one was consistent with our tim-
ing hypothesis that the post-threat affirmation condition
would be unique in reducing defensiveness (post-threat: –2;
control, foreshadowing: +1). The other contrast examined
the reprocessing-based prediction that the affirmation condi-
tion would equally reduce defensiveness compared to the
control condition (post-threat, foreshadowing: –1; control: +2).
The former contrast was significant, t(79) = 2.12, p = .04, but
not the latter (t " 1). Participants were significantly more
defensive in the foreshadowing condition than in the simple
post-threat affirmation condition, t(79) = 1.97, p = .05. This
comparison is the most direct test of whether foreshadowing
eliminates the effect of post-threat affirmations. Nonaffirmed
participants displayed marginally more defensiveness than
those who were post-threat affirmed, t(79) = 1.75, p = .08,
but no more defensiveness than those affirmed partici-
pants for whom the measures of defensiveness had been
foreshadowed (t " 1).
Figure 3. Defensiveness as a function of affirmation condition
(Study 2b)
Values are z score composites and thus have no absolute meaning;
0 is not “no defensiveness.”

954 Personality and Social Psychology Bulletin 36(7)
Discussion
Studies 2a and 2b helped to reconcile our assertions about
the timing of affirmation interventions with past work show-
ing that affirmations can halt defensiveness even after par-
ticipants are presented with a threat. In Study 2a, we showed,
as in past work, that a post-threat affirmation can be success-
ful in reducing defensiveness. However, in Study 2b, with
the same threat and defensiveness measures, the very same
self-affirmation was rendered ineffective—but only in the
condition in which the pathways to defensiveness had been
suggested to participants prior to the self-affirmation. Taken
together, these results suggest that when participants form
motivated, defensive conclusions prior to a self-affirmation,
the affirmation will not undo these conclusions. Post-threat
affirmations, therefore, may well be effective in blocking
subsequent defensive processing but will not undo defensive
conclusions when such processes have already taken place.
Study 3
The results of Studies 2a and 2b suggest that post-threat
affirmations will be effective only if they are initiated
before a threat response. If this is true, then we should be
able to use the same logic to explain the apparent effective-
ness of post-threat affirmations that have been documented
in other domains. In Study 3, we returned to the origins of self-
affirmation theory and examined its impact on cognitive dis-
sonance after writing a counterattitudinal essay. Steele and
Liu (1983) found that if participants self-affirmed after writ-
ing a counterattitudinal essay, they no longer defensively
shifted their attitudes to justify having written the essay.
According to our account, this post-threat affirmation was
effective because attitude change had actually yet to occur,
not because the self-affirmation undid an already drawn
defensive reaction.
We had reasons to suspect that attitude change following
counterattitudinal advocacy is not spontaneous but happens
only upon presentation of an attitude questionnaire. Aronson,
Blanton, and Cooper (1995) found that after writing an
uncompassionate counterattitudinal essay, participants were
willing to accept personality feedback that they were a com-
passionate person, but only if they had first been given a
chance to shift their attitudes. If not first given the chance,
they did not want to accept this feedback, which presumably
would have highlighted the personal standard of compassion
that participants had violated. Had attitude change been spon-
taneous, varying the timing of the presentation of the attitude
measure would have had no effect.
If our hypothesis is correct, then foreshadowing our mea-
sures of defensiveness (the attitude measure) following the
essay but prior to the self-affirmation essay (a dissonance +
foreshadowing condition) the affirmation should be less
effective than if no foreshadowing occurred (a dissonance + no
foreshadowing condition). We added a third no dissonance +
foreshadowing condition both as a no threat comparison
group and to make certain that it was not simply the fore-
shadowing that prompted a defensive response. If the post-
threat affirmation was effective in eliminating defensiveness
but there was simply something about foreshadowing that
led to more attitude change, then this should be true whether
participants had first been made to experience dissonance
or not. Thus, we predicted that those in the dissonance +
foreshadowing condition would display more defensiveness
than either those in the dissonance + no foreshadowing or the
no dissonance + foreshadowing condition.
Method
Participants and design. In exchange for extra credit in their
psychology and human development classes, 76 Cornell
University undergraduates participated. All participants
were asked to write a counterattitudinal essay and then to
complete a self-affirmation exercise. Participants were ran-
domly assigned to a dissonance + no foreshadowing, disso-
nance + foreshadowing, or no dissonance + foreshadowing
condition. Of the participants, 8 refused to write the essay
(4 in the dissonance + no foreshadowing condition, 3 in the
dissonance + foreshadowing, and 1 in the no dissonance +
foreshadowing), and 2 unfortunately astute participants iden-
tified the methodology as a “dissonance paradigm.” These 10
participants were excluded from all analyses reported below,
leaving 66 participants in our final set of analyses.
Procedure. The experimenter, intentionally reading off of a
script, explained that the university’s “Committee of Plans
and Resources” was currently soliciting student feedback on
whether to expend the resources to make all university build-
ings accessible to the physically disabled. It was explained
that loopholes in state law had not required the university to
bring its older buildings into compliance with current stan-
dards. It was said that at this point the committee was having
participants write short, persuasive statements explaining
why they “support or oppose a funding increase to help the
physically disabled.” Participants were told they would write
a short essay that would be sealed in an envelope to be sent
to the committee.
At this point, the experimenter looked up and put the
paper from which she was reading aside to make it appear
that she was going “off script.” For those in the two disso-
nance conditions, the experimenter created the impression
that writing the counterattitudinal essay was a matter of free
choice by stating, “We actually already have enough essays
written in favor of the funding increase. So we are asking
participants if they wouldn’t mind writing an essay against
the funding increases. Is that OK?” For participants who
agreed to write the essay, the experimenter gave them a
form on which they were to write their essay and a letter-
size Department of Psychology envelope. At the top of the
at CORNELL UNIV on June 24, 2010 http://psp.sagepub.comDownloaded from

Critcher et al. 955
form was the university seal and the fictitious title of
the committee, under which was written “On University
Resolution 2007-0138. Position: OPPOSED.” Just under-
neath, the form read,
Thank you for your willingness to offer your opinion
on this important issue affecting our university com-
munity. Please write a strong and convincing essay in
the space below. The university should not increase
spending for services for the physically disabled
because . . .
This was followed by a blank page on which participants
could write their essays. Each participant was told that upon
completing the essay, he or she should fold and place it in
the envelope and then seal and sign the envelope to indicate
to the committee that the essay had not been tampered with.
In the no dissonance + foreshadowing condition, the
ostensibly “off script” instructions provided to participants
in the no dissonance + foreshadowing condition were
intended to dispel any dissonance created by making it clear
that the participant had no choice but to write the essay.
Specifically, participants in this condition were told,
Because we already have enough essays written in
favor of the funding increase, we are just having par-
ticipants write against the funding increase. Also, since
we aren’t giving you a choice, we are putting a star on
your envelope so the committee knows that you were
just assigned this position.
After writing the essay, participants in the two foreshadowing
conditions were told,
We have two tasks left for you to do. The first task is a
writing task. Then, we will have you answer a few
questions relevant to the first part of the experiment,
such as to what extent you support or oppose the fund-
ing increase.
By describing the final questions as relevant only to “the
first part of the experiment,” the foreshadowing manipulation
should not have communicated that there was a connection
between the affirmation and the measures of defensiveness
(Sherman et al., 2009). Participants in the no foreshadowing
condition were not forewarned about the additional attitude-
relevant questions. At this point, all participants completed
the same self-affirmation manipulation used in Studies 2a
and 2b. None of the values participants could affirm related
to helping or showing compassion for others.
Finally, all participants answered to what extent they
agreed with the statement, “The University should allocate
more funds to improving facilities and services for the
physically disabled” on a 17-point scale that ranged from 1
(strongly disagree) to 17 (strongly agree). The degree to
which participants opposed the funding increase (and thus
endorsed the counterattitudinal position) was taken as an
indication of dissonance reduction or defensiveness.
Results
We tested our timing hypothesis by contrasting the attitudes
observed in the dissonance + foreshadowing condition
(weighted –2) with the attitudes seen in the other two conditions
(both dissonance + no foreshadowing and no dissonance +
foreshadowing conditions weighted +1). The resulting statis-
tical contrast was significant, t(63) = 2.80, p = .01 (see Figure 4).
Conceptually replicating Study 2b, those in the dissonance +
foreshadowing condition were less supportive of the physi-
cally disabled (M = 10.6) than those in the dissonance + no
foreshadowing condition (M = 12.3), t(63) = 2.00, p = .05,
and in the no dissonance + foreshadowing group (M = 13.1),
t(63) = 2.83, p = .01. Attitudes in the latter two conditions
did not differ (t " 1). We should note that these results are
inconsistent with the view that the timing of affirmation
does not matter. This view would predict that the three
experimental groups would not differ, in that the affirmation
intervention would have prevented attitude change regard-
less of foreshadowing.
Discussion
Study 3 used a very different type of defensiveness—attitude
change following counterattitudinal advocacy—to test
whether self-affirmations only at times appear to undo acts
of defensiveness because the acts of defensiveness they
blocked had actually yet to occur. A post-threat affirmation
became ineffective once the to-be-measured method of
defensiveness (attitude change) was foreshadowed for
participants just prior to the self-affirmation.
Figure 4. Support for the disabled as a function of dissonance
foreshadowing condition (Study 3)
Lower values reflect greater defensiveness.

956 Personality and Social Psychology Bulletin 36(7)
The inclusion of the no dissonance + foreshadowing
condition allowed us to rule out an alternative explanation
that people in the foreshadowing condition simply had
more time to consider their attitude toward the physically
disabled, such that an initially positive reaction may have
eventually given way to a more ambivalent stance. That is,
perhaps with more thought, one would think of other wor-
thy causes that would permit better or more efficient uses
of the university’s resources. This alternative predicts that
foreshadowing would not merely prompt what looks like
defensiveness after dissonance had been evoked but that it
would have the same effect even when no dissonance had
been evoked. The significant difference between the disso-
nance foreshadowing and the no dissonance foreshadowing
conditions speaks against this possibility.4
General Discussion
We reason that if a self-affirmation satisfies a self-esteem
need and this need had been heightened by a recent threat,
then an affirmation should block subsequent acts of defen-
siveness. However, to the extent that motivated conclusions
have already been reached, affirmations should not cause
people to undo or “reprocess” such conclusions. This dis-
tinction (of timing in relation to a threat response rather than
timing in relation to a threat experience) has to date been
overlooked. Self-affirmations and defensiveness are inter-
changeable ways of alleviating threat (Tesser, 2000), and
because the motivation to self-enhance is most active during
threat (Tesser, 1988), an affirmation following defensiveness
is unlikely to influence responding.
The results of the four studies herein confirm these pre-
dictions. We found that self-affirmations administered prior
to the defensive response to threat attenuated subsequent
defensive conclusions. When we used measures of defen-
siveness that assessed presumably direct and spontaneous
reactions to a test (Study 1), a post-threat affirmation did not
influence defensive responding. In circumstances in which
post-threat affirmations blocked more indirect measures of
defensiveness (Studies 2a and 2b) and attitude change in
response to dissonance (Study 3), those affirmations became
ineffective once participants were told before the affirmation
about the possible defensive responses that would later be
assessed. Without this foreshadowing condition, the timing
of a threat response, and the effectiveness of an affirmation in
relation to that response, would be unclear, and we suggest that
this is one reason why the role of timing in affirmation effec-
tiveness has not previously been identified. In other words,
self-affirmation does not appear to occupy a privileged status
among means of self-esteem restoration. After threat, people
move to repair their self-integrity, and either an affirmation or
a defensive response will do—and they will take whatever
comes first. People do not necessarily reconsider their defen-
sive responses in light of affirmation as past researchers have
suggested (Cohen et al., 2000; Sherman et al., 2000).
The Foreshadowing Manipulation
The foreshadowing manipulation was meant to simply alert
participants before the affirmation exercise of the defensive-
ness pathways that would ultimately be available to them
and thus to start that process. But might the foreshadowing
manipulation have instead led participants to take the affir-
mation task less seriously? If so, this might explain why the
affirmation was no longer effective in the foreshadowing
conditions.
To address this alternative, we returned to the actual affir-
mation essays that participants wrote in Studies 2b and 3.
Two coders, blind to conditions and hypotheses, (a) rated the
overall “affirming value” of each essay on a 1 to 7 scale
based on a modified version of an affirmation coding scheme
developed by Creswell et al. (2007) (rs = .83 and .93 for the
two studies) and (b) counted the number of words in each
essay (rs = 1.0). Contrary to an explanation that those in the
foreshadowing condition took the affirmation task less seri-
ously, they wrote essays that were just as affirming (ts = 1.13
and 1.48, ps # .14) and of no shorter length (ts " 1). Also,
logistic regressions found that foreshadowing did not change
the domain participants chose to affirm (ps # .20), and across
all essays coders only twice observed any connection at all
between the threatened domain and the content of the
affirmation—once in each condition. In short, foreshadow-
ing seemed to eliminate the tendency for affirmations to
reduce subsequent defensiveness without changing the way
people completed the affirmation task.
Why Wouldn’t Timing Matter?
A consistent theme in social cognition is that timing of cogni-
tive manipulations does matter (Von Hippel, Sekaquaptewa,
& Vargas, 1995). A conceptual prime has the potential to
affect perception of a stimulus only when experienced before,
but not after, exposure to the target (Srull & Wyer, 1980).
Schemas help to organize ambiguous information when
learned of before exposure to the information but confer no
benefit when learned of following exposure (Bransford &
Johnson, 1973). People’s expectations color their interpreta-
tion of bottom-up experience, but only when the expectation
is put in place before the actual experience (Critcher &
Dunning, 2009).
With all this research in mind, a careful reader may wonder
how much news there is in our data that timing also matters in
the realm of self-affirmation. Such a question is reasonable,
but the extant literature does not necessarily guarantee that
timing would still matter as one moves from the cognitive
(e.g., expectations) to the motivational (e.g., self-affirmation).
Not only does past evidence in self-affirmation, at first look,
appear to suggest that timing would not matter (e.g., Cohen
et al., 2000), there was reason to suspect that the reprocess-
ing hypothesis (that timing would not matter) would
hold water. For cognitive variables, such as expectations, to

Critcher et al. 957
influence perception, they must be in place before the stimulus
is literally perceived because they shape the very encoding
and interpretation of a stimulus. Self-affirmations do not
directly color processing or encoding; instead, they exert a
subtractive force, eliminating the influence of a motivation
that would have pushed one toward desired perceptions. If
one thinks of motivated conclusions as being bolstered or
“propped up” by this motivation, then it would seem reason-
able that subtracting out the force could diminish defensive-
ness, even after it is already instigated.
Where this intuition is in error, we suspect, is that even
though motivation may be responsible for producing a per-
ception, it is not the motivation that sustains that percep-
tion. As an illustration, threatened participants in Study 2b
may have been motivated to call to mind scores of “friends”
in defensively downplaying the implications of the threat-
ening feedback. But once the motivation that spurred this
intense search was removed, it does not follow that the
acquaintances who were recalled would suddenly be for-
gotten. That is, taking away the motivation need not lead
one to dismiss what past motivated reasoning has pro-
duced. More generally, subtracting the motivational influ-
ence that produces motivated distortions does not cause
them to go away.
Theoretical and Applied Implications
Our data have at least four broad implications for self-
affirmation theory and for the psychology of defensiveness
more generally. First, because the moderating role of affir-
mations’ timing seems to be pre-defensiveness or post-
defensiveness instead of pre-threat or post-threat, post-threat
affirmations may offer promise as a tool to determine whether
defensiveness occurs spontaneously or merely once prompted
by a dependent measure. For example, a comparison of
Studies 1 and 2 suggests that people may be more likely to
spontaneously engage in direct, as opposed to indirect, meth-
ods of alleviating threat. This is consistent with past research
that suggests that those who have positive, affirming identity
resources (an indirect means to self-repair) may have diffi-
culty spontaneously relying on them unless the experimenter
directs participants’ attention to such self-esteem resources
(Spencer, Josephs, & Steele, 1993). We suspect that one rea-
son why research on the psychology of defense has uncov-
ered such a variety of defensive strategies is not because
people spontaneously use all of them but because they know
how to use them once they are suggested by someone else.
Second, they suggest that defensive conclusions, once
drawn, may often be crystallized rather quickly. In Studies
2b and 3, foreshadowed participants received the self-
affirmation manipulation just seconds after being informed
of what questions (the measures of defensiveness) they
would answer after the writing task (the self-affirmation).
The defensive conclusions that participants presumably
drew in that short time span (e.g., “That test got me wrong;
I have so many friends!”) were not tempered by an imme-
diately subsequent self-affirmation, suggesting a very short
critical period for post-threat affirmations.
Third, and related, is to what extent a defensive response
is cemented. Our data suggest that if a defensive response
occurs spontaneously or is triggered before the motivation to
be defensive is eliminated (whether through self-affirmation
or perhaps the mere passage of time), the defensive conclu-
sions reached through these processes will remain even once
the motivation that spurred them is eliminated. Consistent
with this conclusion, Dunning (2003) suggested that self-
affirming defensive responses likely leave residuals of self-
enhancement that over time combine to produce a better-
than-average view of the self. Implicit in this argument is
that motivated self-enhancement does not reverse itself even
once self-integrity is restored.
Fourth, the crucial role of affirmations’ timing may have
important implications for the use of affirmations in applied
or clinical settings. In general, practitioners are likely to have
the most success with affirmation interventions when intro-
duced just prior to a threat. But this situation may be compli-
cated if one has already defensively downplayed the threat.
Consistent with this notion, Epton and Harris (2008) note
that self-affirmations may be more effective in encouraging
new health-promoting behaviors than in discouraging health-
deteriorating behaviors in which one already engages. If
people are more likely to have defensively justified their bad
habits than their failure to engage in a good habit, our
research may explain this previously observed asymmetry.
But we believe this picture is unnecessarily bleak, for an
affirmation should be able to change a previously justified
behavior when the affirmation is presented prior to a brand
new appeal, one that uses a new persuasion tactic. For example,
Armitage, Harris, Hepton, and Napper (2008) found that
heavy cigarette smokers were more receptive to a novel anti-
smoking message when they were first affirmed. When
people affirm before exposure to this novel appeal, the affir-
mation can truly occur pre-defensiveness, giving it a higher
chance of success.
Conclusion
In much of this article we have highlighted the perils of
defensive responding (e.g., inaccurate self-knowledge, rejec-
tion of constructive feedback), but we do not wish to imply
that defensive processes are always best to be suppressed.
Rationalizing away a romantic interest’s snub, downplaying
the importance of one’s own artistic ineptitude, and believ-
ing one will live forever may help avoid a life plagued
by self-consciousness, low self-esteem, and high anxiety
(Critcher et al., in press). At the same time, doing these
things to excess may lead one to continue to pursue unattain-
able dates, unwisely invest one’s inheritance promoting
one’s own bad artwork, and engage in high-risk behaviors.
Determining the boons and banes of defensiveness processes
at CORNELL UNIV on June 24, 2010 http://psp.sagepub.comDownloaded from

958 Personality and Social Psychology Bulletin 36(7)
is a worthy task for continued research, though the present
studies should aid practitioners in effectively harnessing the
defensive-reducing power of self-affirmations.
Acknowledgments
We thank Jane Risen and Dennis Regan for their comments on a
draft of the article. We thank Jill Fleischer, Katie McCrary, and
Paige Weinger for their assistance with data collection. We thank
Sally Apuzzo and Lisa Colton for coding the affirmation essays in
Studies 2b and 3.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interests with respect
to the authorship and/or publication of this article.
Financial Disclosure/Funding
The authors received the following financial support for the
research and/or authorship of this article: This research was sup-
ported in part by a National Science Foundation Graduate Research
Fellowship and a Yale University Mellon Research Grant awarded
to Clayton Critcher and by National Institute of Mental Health
Grant RO1 56072 and National Science Foundation Grant 0745806
awarded to David Dunning.
Notes
1. The answers are cookie and box.
2. A traditional control for this particular affirmation is to have
participants write about why their least valued domain might be
important to someone else. We worried that, to the extent that
devalued domains might be domains on which one did not feel
particularly competent, this “control” might serve as a threat
and exaggerate any observed effects of the affirmation manip-
ulation. To reduce this concern, we used a content-unrelated
control that seemed unlikely to serve as either a threat or a self-
affirmational resource for any participant.
3. Speaking to the comprehensiveness of the hypothesized con-
trasts, the residual variance is not significant in each study:
F(1, 96) = 3.01, p = .09 (Study 1); F " 1 (Study 2a); F " 1
(Study 2b); F " 1 (Study 3). The marginally significant residual
variance in Study 1 reflected the tendency for the post-threat
affirmation to lead to marginally more defensiveness than when
nonaffirmed and does not speak to the tenability of the compet-
ing reprocessing hypothesis.
4. Note that the inclusion of a no affirmation + dissonance condi-
tion would have allowed us to test whether the foreshadowing
manipulation fully or only partially eliminated the impact of
the affirmation, a more nuanced concern that was not central
to the study’s purpose, though note that this comparison is pos-
sible in Study 2b, which showed that foreshadowing eliminated
the full impact of the affirmation.
References
Alicke, M. D., Klotz, M. L., Breitenbecher, D. L., Yurak, T. J., &
Vredenburg, D. S. (1995). Personal contact, individuation, and
the better-than-average effect. Journal of Personality and Social
Psychology, 68, 804-825.
Armitage, C. J., Harris, P. R., Hepton, G., & Napper, L. (2008).
Self-affirmation increases acceptance of health risk informa-
tion among UK adult smokers with low socioeconomic status.
Psychology of Addictive Behaviors, 22, 88-95.
Aronson, J., Blanton, H., & Cooper, J. (1995). From dissonance
to disidentification: Selectivity on the self-affirmation process.
Journal of Personality and Social Psychology, 6, 986-996.
Aspinwall, L. G. (1997). Future-oriented aspects of social com-
parisons: A framework for studying health-related comparison
activity. In B. P. Bunk & F. X. Gibbons (Eds.), Health, coping,
and well-being: Perspectives from social comparison theory
(pp. 125-165). Mahwah, NJ: Erlbaum.
Baumeister, R. F., & Leary, M. R. (1995). The need to belong:
Desire for interpersonal attachments as a fundamental human
motivation. Psychological Bulletin, 117, 497-529.
Beauregard, K. S., & Dunning, D. (2001). Defining self-worth:
Trait self-esteem moderates the use of self-serving trait
definitions in social judgment. Motivation and Emotion, 25,
135-161.
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
Psychology Bulletin, 23, 684-692.
Bransford, J. D., & Johnson, M. K. (1973). Considerations of some
problems of comprehension. In W. G. Chase (Ed.), Visual infor-
mation processing (pp. 383-438). New York, NY: Academic Press.
Brown, J. D., & Smart, S. A. (1991). The self and social conduct:
Linking self-representations to prosocial behavior. Journal of
Personality and Social Psychology, 60, 368-375.
Campbell, J. D. (1986). Similarity and uniqueness: The effects of
attribute, type, relevance, and individual differences in self-
esteem and depression. Journal of Personality and Social
Psychology, 50, 281-294.
Cohen, G., Aronson, J., & Steele, C. M. (2000). When beliefs yield
to evidence: Reducing biased evaluation by affirming the self.
Personality and Social Psychology Bulletin, 26, 1151-1164.
Costanzo, M., & Archer, D. (1989). Interpreting the expressive
behavior of others: The Interpersonal Perception Task. Journal
of Nonverbal Behavior, 13, 225-245.
Creswell, J. D., Lam, S., Stanton, A. S., Taylor, S. E., Bower, J. E., &
Sherman, D. K. (2007). Does self-affirmation, cognitive pro-
cessing, or discovery of meaning explain the cancer-related
health benefits of expressive writing? Personality and Social
Psychology Bulletin, 33, 238-250.
Critcher, C. R., & Dunning, D. (2009). How chronic self-views
influence (and mislead) self-assessments of performance:
Self-views shape bottom-up experiences with the task. Journal
of Personality and Social Psychology, 97, 931-945.
Critcher, C. R., Helzer, E., & Dunning, D. (in press). Self-enhancement
via redefinition: Defining social concepts to ensure positive
views of self. In M. Alicke & C. Sedikides (Eds.), Handbook of
self-enhancement. New York: Guilford Press.

Critcher et al. 959
Dunning, D. (2003). The zealous self-affirmer: How and why the
self lurks so pervasively behind social judgment. In S. J. Spencer,
S. Fein, M. P. Zanna, & J. M. Olson (Eds.), The Ontario sympo-
sium (Vol. 9, pp. 45-72). Mahwah, NJ: Erlbaum.
Dunning, D. (2005). Self-insight. Roadblocks and detours on the
path to knowing thyself. New York, NY: Psychology Press.
Dunning, D., Heath, C., & Suls, J. (2004). Flawed self-
assessment: Implications for health, education, and the work-
place. Psychological Science in the Public Interest, 5, 69-106.
Epton, T., & Harris, P. R. (2008). Self-affirmation promotes health
behavior change. Health Psychology, 27, 746-752.
Harris, P. R., & Napper, L. (2005). Self-affirmation and the biased
processing of threatening health-risk information. Personality
and Social Psychology Bulletin, 31, 1250-1263.
Hart, J., Shaver, P. R., & Goldenberg, J. L. (2005). Attachment,
self-esteem, worldviews, and terror management: Evidence for
a tripartite security system. Journal of Personality and Social
Psychology, 88, 999-1013.
Klein, W. M., & Kunda, Z. (1989, March). Motivated person per-
ception: Justifying desired conclusions. Paper presented at the
meeting of the Eastern Psychological Association, Boston, MA.
Marks, G. (1984). Thinking one’s abilities are unique and one’s
opinions common. Personality and Social Psychology Bulletin,
10, 203-208.
Markus, H. R., & Wurf, E. (1987). The dynamic self-concept: A
social psychological perspective. Annual Review of Psychology,
38, 299-337.
McQueen, A., & Klein, W. M. P. (2006). Experimental manipula-
tions of self-affirmation: A systematic review. Self and Identity,
5, 289-354.
Mednick, S. A. (1962). The associative basis of the creative pro-
cess. Psychological Review, 26, 220-232.
Radcliffe, N. M., & Klein, W. M. P. (2002). Dispositional, unrealistic,
and comparative optimism: Differential relations with the knowl-
edge and processing of risk information and beliefs about personal
risk. Personality and Social Psychology Bulletin, 28, 836-846.
Reed, M. B., & Aspinwall, L. G. (1998). Self-affirmation reduces
biased processing of health-risk information. Motivation and
Emotion, 22, 99-132.
Sedikides, C., Green, J. D., & Pinter, B. (2004). Self-protective
memory. In D. R. Beike, J. M. Lampinen, & D. A. Behrend
(Eds.), The self and memory (pp. 161-179). Philadelphia, PA:
Psychology Press.
Sherman, D. K., & Cohen, G. L. (2006). The psychology of self-
defense: Self-affirmation theory. In M. P. Zanna (Ed.), Advances
in experimental social psychology (Vol. 38, pp. 183-242). San
Diego, CA: Academic Press.
Sherman, D. K., Cohen, G. L., Nelson, L. D., Nussbaum, A. D.,
Bunyan, D. P., & Garcia, J. (2009). Affirmed yet unaware:
Exploring the role of awareness in self-affirmation. Journal of
Personality and Social Psychology, 97, 745-764.
Sherman, D. A. K., Nelson, L. D., & Steele, C. M. (2000). Do
messages about health risks threaten the self: Increasing the
acceptance of threatening health messages via self-affirmation.
Personality and Social Psychology Bulletin, 26, 1046-1058.
Shira, I., & Martin, L. L. (2005). Stereotyping, self-affirmation, and
the cerebral hemispheres. Personality and Social Psychology
Bulletin, 31, 846-856.
Sivanthan, N., Molden, D. C., Galinsky, A. D., & Ku, G. (2008).
The promise and peril of self-affirmation in de-escalation of
commitment. Organizational Behavior and Human Decision
Processes, 107, 1-14.
Spencer, S. J., Fein, S., & Lomore, C. D. (2001). Maintaining one’s
self-image vis-à-vis others: The role of self-affirmation in the
social evaluation of the self. Motivation and Emotion, 25, 41-65.
Spencer, S. J., Josephs, R. A., & Steele, C. M. (1993). Low self-
esteem: The uphill struggle for self-integrity. In R. F. Baumeister
(Ed.), Self-esteem and the puzzle of low self-regard (pp. 21-36).
New York, NY: Wiley.
Srull, T. K., & Wyer, R. S. (1980). Category accessibility and social
perception: Some implications for the study of person memory
and interpersonal judgments. Journal of Personality and Social
Psychology, 38, 841-856.
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). San Diego,
CA: Academic Press.
Steele, C. M., & Liu, T. J. (1983). Dissonance processes as self-
affirmation. Journal of Personality and Social Psychology,
45, 5-19.
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A
social psychological perspective on mental health. Psychological
Bulletin, 103, 193-210.
Taylor, S. E., & Lobel, M. (1989). Social comparison activ-
ity under threat: Downward evaluation and upward contacts.
Psychological Review, 96, 569-575.
Tesser, A. (1988). Toward a self-evaluation maintenance model of
social behavior. In L. Berkowitz (Ed.), Advances in experimen-
tal social psychology (Vol. 21, pp. 181-227). San Diego, CA:
Academic Press.
Tesser, A. (2000). On the confluence of self-esteem maintenance
mechanisms. Personality and Social Psychology Review, 4,
290-299.
Tesser, A., Martin, L. L., & Cornell, D. P. (1996). On the substi-
tutability of self-protective mechanisms. In P. M. Golwitzer &
J. A. Bargh (Eds.), The psychology of action: Linking cognition and
motivation to behavior (pp. 48-68). New York, NY: Guilford.
Trope, Y., & Pomerantz, E. M. (1998). Resolving conflicts among
self-evaluative motives: Positive experiences as a resource for
overcoming defensiveness. Motivation and Emotion, 22, 53-72.
Von Hippel, W., Sekaquaptewa, D., & Vargas, P. (1995). On the role
of encoding processes in stereotype maintenance. Advances in
Experimental Social Psychology, 27, 177-254.
Wentura, D., & Greve, W. (2003). Who wants to be . . . erudite?
Everyone! Evidence for automatic adaptation of trait definitions.
Social Cognition, 22, 30-53.
at CORNELL UNIV on June 24, 2010 http://psp.sagepub.comDownloaded from