Terror Management Theory and Self ...

[ Pobierz całość w formacie PDF ]
Journal of Personality and Social Psychology
0022-3514/09/$12.00 DOI: 10.1037/a0015091
Terror Management Theory and Self-Esteem Revisited: The Roles of
Implicit and Explicit Self-Esteem in Mortality Salience Effects
Brandon J. Schmeichel
Texas A&M University
Matthew T. Gailliot
Florida State University
Emily-Ana Filardo and Ian McGregor
York University
Seth Gitter and Roy F. Baumeister
Florida State University
Three studies tested the roles of implicit and/or explicit self-esteem in reactions to mortality salience. In
Study 1, writing about death versus a control topic increased worldview defense among participants low
in implicit self-esteem but not among those high in implicit self-esteem. In Study 2, a manipulation to
boost implicit self-esteem reduced the effect of mortality salience on worldview defense. In Study 3,
mortality salience increased the endorsement of positive personality descriptions but only among
participants with the combination of low implicit and high explicit self-esteem. These findings indicate
that high implicit self-esteem confers resilience against the psychological threat of death, and therefore
the findings provide direct support for a fundamental tenet of terror management theory regarding the
anxiety-buffering role of self-esteem.
Keywords:
death, mortality salience, self-esteem, terror management
Many aspects of life involve avoiding death either physically or
psychologically. Terror management theory (TMT) proposes that
the purpose of self-esteem is to buffer against the psychological
threat of death (Greenberg, Pyszczynski, & Solomon, 1986; Pyszc-
zynski, Greenberg, Solomon, Arndt, & Schimel, 2004; see Becker,
1973). According to the theory, high self-esteem (i.e., the extent to
which one holds favorable views of oneself) reflects the successful
participation in and internalization of a meaningful cultural world-
view. By becoming a valued member of society, linked with
like-minded others through shared ideals and beliefs, an individual
can manage the potentially paralyzing terror of death and live life
with relative equanimity. Conversely, low self-esteem reflects an
ineffectual buffer against death anxiety due to failure to meet
cultural standards of value. This view of self-esteem forms one of
the core pillars of TMT.
Although research has observed that reminders of death (i.e.,
mortality salience) tend to increase esteem-seeking behavior, the
relationship between dispositional self-esteem and reactions to
mortality salience has yet to be clearly established. Whereas some
studies have found attenuated reactions to mortality salience
among people with high (versus low) self-esteem (e.g., Harmon-
Jones et al., 1997), other studies have found the exact opposite
pattern (e.g., Baldwin & Wesley, 1996). The current work sought
to resolve this inconsistency by examining the link between mor-
tality salience and both explicit (relatively conscious and reflec-
tive) and implicit (relatively unconscious and spontaneous) forms
of self-esteem.
Self-Esteem and Mortality Salience
TMT posits that self-esteem buffers against death-related
thought and anxiety. Consistent with this view, research has dem-
onstrated that thinking about death (vs. other aversive topics such
as dental pain or personal failure) increases the need for self-
esteem, as evidenced by increased self-esteem striving and a
tendency toward self-enhancement (for a review, see Pyszczynski
et al., 2004). This evidence only partially supports the idea that
self-esteem buffers against death, however. More complete evi-
dence would consist of showing not only that mortality salience
increases defensive reactions and self-serving biases (Solomon,
Greenberg, & Pyszczynski, 2004) but also that self-esteem mod-
erates reactions to mortality salience. If self-esteem buffers against
the psychological threat of death, then people with high self-
esteem should be less defensive in response to mortality salience
than are people with low self-esteem.
Consistent with the anxiety-buffering hypothesis of self-esteem
is evidence that positive personality feedback ameliorates several
responses to mortality salience, including self-reported anxiety
(Greenberg, Solomon, et al., 1992), worldview defense (Arndt &
Greenberg, 1999), and the tendency to deny the possibility of an
early death (Greenberg et al., 1993). Positive personality feedback
often increases self-esteem, but little direct evidence exists to
support the view that increased self-esteem explains the ameliora-
tion of terror management defenses. Some evidence indicates that
Brandon J. Schmeichel, Department of Psychology, Texas A&M Uni-
versity; Matthew T. Gailliot, Seth Gitter, and Roy F. Baumeister, Depart-
ment of Psychology, Florida State University; Emily-Ana Filardo and Ian
McGregor, Department of Psychology, York University, Toronto, Ontario,
Canada.
Thank you to Vanessa Delgado and Cindy Harmon-Jones for helping to
conduct this research.
Correspondence concerning this article should be addressed to Brandon
J. Schmeichel, Department of Psychology, Texas A&M University, Col-
lege Station, TX 77843-4235. E-mail: schmeichel@tamu.edu
1077
2009, Vol. 96, No. 5, 1077–1087
© 2009 American Psychological Association
1078
SCHMEICHEL ET AL.
self-affirmation reduces defensiveness and death-thought accessi-
bility after mortality salience (Landau & Greenberg, 2006;
Schmeichel & Martens, 2005), but self-affirmation and self-esteem
are not synonymous, and evidence suggests that the effects of
self-affirmation are not due to changes in (explicit) self-esteem
(Schmeichel & Martens, 2005). Thus, although most of the evi-
dence provides reason to conclude that self-esteem moderates
reactions to mortality salience, this conclusion may be questioned
because most of the previous research has not afforded a direct test
of the anxiety-buffering hypothesis.
The few studies that have directly examined whether self-
esteem moderates mortality salience effects have produced con-
tradictory findings. We identified six studies in which mortality
salience caused defensive reactions primarily among participants
lower in explicit self-esteem (Gailliot, Schmeichel, & Maner,
2007; Goldenberg & Shackelford, 2005; Greenberg et al., 1993;
Harmon-Jones et al., 1997; Kashima, Halloran, Yuki, & Kashima,
2004; Taubman-Ben-Ari & Findler, 2005), consistent with the
anxiety-buffering hypothesis. We found another six studies, how-
ever, that observed the opposite pattern, whereby mortality sa-
lience produced defensive reactions primarily among participants
higher in explicit self-esteem (Baldwin &Wesley, 1996; Landau &
Greenberg, 2006; McGregor, Gailliot, Vasquez, & Nash, 2007;
Taubman-Ben-Ari & Findler, 2006).
In summary, the anxiety-buffering hypothesis of self-esteem as
proposed by TMT has received some support, but in fact some
evidence has directly contradicted the anxiety-buffering view.
Therefore, it appears as though a major pillar of TMT rests on an
uncertain foundation. Might this pillar of TMT require revision?
self-affirmation (thinking about one’s cherished values) alters how
individuals react to mortality salience (Landau & Greenberg, 2006;
Schmeichel & Martens, 2005). Self-affirmation has been shown to
increase implicit but not explicit self-esteem (Koole, Smeets, van
Knippenberg, & Dijksterhuis, 1999), and so it is possible that
self-affirmation alters responses to mortality salience by increasing
implicit self-esteem. We therefore predicted that high (vs. low)
implicit self-esteem would moderate responses to mortality sa-
lience by making them less defensive.
As discussed above, past work has revealed conflicting results
concerning the relationship between explicit self-esteem and reac-
tions to mortality salience. We expected—on the basis of previous
evidence suggesting that high explicit self-esteem, in combination
with low implicit self-esteem, reflects inner insecurity and a pre-
disposition toward defensiveness, rather than a soothing entrench-
ment in society (e.g., Jordan, Spencer, Zanna, Hoshino-Browne, &
Correll, 2003; McGregor & Marigold, 2003; Ziegler-Hill, 2006)—
that high explicit self-esteem, especially in conjunction with low
implicit self-esteem, would predispose individuals to react defen-
sively to the thought of their own death.
In the present research, Study 1 assessed whether mortality
salience causes most worldview defense among individuals with
low scores on a measure of dispositional implicit self-esteem that
was based on participants’ fondness for the letters in their own
names (Nuttin, 1985, 1987). Study 2 assessed whether an experi-
mental manipulation to boost implicit self-esteem decreases world-
view defense after mortality salience. Study 3 assessed whether
preferences for positive personality feedback following mortality
salience are most evident among participants with the combination
of high explicit self-esteem and low implicit self-esteem, as as-
sessed by an implicit association test (IAT; Greenwald & Farnham,
2000).
Alternative Types of Self-Esteem Examined in the
Current Work
We believe the anxiety-buffering hypothesis of self-esteem pro-
posed by TMT is generally correct, but it is crucial to consider the
type of self-esteem that is examined. One type of self-esteem is
explicit self-esteem, which reflects conscious evaluations of the
self. Another type of self-esteem is implicit self-esteem, which is
relatively inaccessible to conscious awareness and distinct from
explicit self-esteem (Gailliot & Schmeichel, 2006; Greenwald &
Banaji, 1995; Hetts & Pelham, 2001; Hetts, Sakoma, & Pelham,
1999; cf. Nosek & Smyth, 2007). Prior theorizing concerning the
role of self-esteem in TMT seems to have invoked a sort of
experiential or gut-level self-esteem (e.g., see Pyszczynski et al.,
2004), whereas researchers have tended to focus on self-reported
explicit self-esteem, which is prone to self-presentational biases
and cognitive distortion (e.g., Baumeister, Tice, & Hutton, 1989).
Implicit self-esteem is relatively less prone to self-presentational
biases and therefore may reflect a less distorted evaluation of self
(Greenwald & Banaji, 1995; cf. Olson, Fazio, & Hermann, 2007).
High implicit self-esteem may be more likely than high explicit
self-esteem to buffer the effects of mortality salience because high
implicit self-esteem represents a more automatic or spontaneous
positive self-evaluation. To provide a strong, direct test of the
anxiety-buffering hypothesis, we examined how both implicit and
explicit forms of self-esteem influence reactions to mortality sa-
lience.
Indirect evidence supports the idea that implicit self-esteem
buffers against the psychological threat of death. For instance,
Study 1
Study 1 tested the hypothesis that mortality salience increases
defensiveness among individuals with relatively low implicit self-
esteem. Participants wrote about either death or a control topic
(dental pain) and then had the opportunity to engage in worldview
defense by evaluating a pro-U.S. and an anti-U.S. essay and their
authors. Bias in favor of the pro-U.S. essay and author over the
anti-U.S. essay and author was taken to indicate greater worldview
defense. We predicted that, consistent with TMT, participants who
wrote about death would exhibit increased worldview defense,
compared with participants who wrote about dental pain. How-
ever, we predicted that this effect would be most pronounced
among individuals lower in implicit self-esteem.
Method
Participants.
One hundred fifty-seven undergraduate students
(112 women) enrolled in an introductory psychology course sat-
isfied a course requirement by participating. Participants were run
in a classroom setting and were randomly assigned to either a
mortality salience or dental pain condition.
Procedure.
Participants completed the name-letter-ratings
measure of implicit self-esteem (Nuttin, 1985) at the beginning of
the semester. Specifically, participants rated the attractiveness of
each of the 26 letters in the English alphabet on a scale from 1 (
not
TERROR MANAGEMENT AND SELF-ESTEEM
1079
at all beautiful
)to7(
extremely beautiful
). Implicit self-esteem
levels were derived from the extent to which participants rated the
letters in their own initials as being attractive, while controlling for
baseline ratings of those letters (i.e., ratings made by participants
whose initials did not contain those letters; see Kitayama & Kara-
sawa, 1997; Koole, Dijksterhuis, & van Knippenberg, 2001).
Approximately 3 weeks later, participants reported to a labora-
tory to complete the main phase of the experiment. Participants
received a packet that contained written instructions and all the
relevant materials for the study, and they worked through the
packet at their own pace (except for a timed filler task; see below).
Participants first completed the mortality salience induction.
Participants in the mortality salience condition described the emo-
tions that the thought of their own death aroused in them and what
would happen to them as they physically died. Participants in the
dental pain condition answered parallel questions about dental
pain.
Next, participants completed the Brief Mood Introspection Scale
(BMIS; Mayer & Gaschke, 1988) as a measure of mood valence
and arousal. Afterward, participants worked on a filler task for 5
min (either completing a crossword puzzle or filling out additional
questionnaires) because the effects of the mortality salience induc-
tion on worldview defense are strongest after a brief delay (Green-
berg, Pyszczynski, Solomon, Simon, & Breus, 1994; Pyszczynski,
Greenberg, & Solomon, 1999). Type of filler task did not influence
the results.
Last, participants completed a measure of worldview defense.
Specifically, participants read two handwritten essays about the
United States that were ostensibly composed by two foreigners
(materials from Greenberg, Simon, Pyszczynski, Solomon, & Cha-
tel, 1992). The order of the two essays was counterbalanced across
participants. One essay was pro-U.S. and praised Americans,
whereas the other essay was anti-U.S. and criticized Americans.
Participants evaluated the truth and validity of the essays and the
likeability, intelligence, and knowledgeability of each essay’s au-
thor on 9-point scales. The summed evaluations for each essay
served as the measures of favorability toward worldview-
consistent and worldview-inconsistent opinions, respectively. In
accord with past research (e.g., Greenberg et al., 1994),
worldview
defense
was defined as the difference between these two measures.
Larger differences indicated greater worldview defense. Partici-
pants were then thanked and debriefed.
.25). Thus, mortality salience triggered worldview defense primar-
ily among participants low in implicit self-esteem.
1
Higher implicit self-esteem was associated with lower world-
view defense scores in the mortality salience condition,
r
(77)
–.32,
p
.05. In the dental pain condition, higher implicit self-
esteem was associated with higher worldview defense scores,
r
(80)
.24,
p
.05. Together, these results suggest that partici-
pants lower in implicit self-esteem fortified their worldviews in
response to threat, whereas those higher in implicit self-esteem
endorsed positive worldview beliefs in a nondefensive manner
(i.e., in the absence of threat).
Mood and arousal.
Analyses indicated that the obtained pat-
tern of results was not attributable to mood valence or arousal. For
instance, the interaction between implicit self-esteem and mortality
salience in predicting worldview defense remained significant
when controlling for mood and arousal.
Study 2
Study 1 indicated that mortality salience increases worldview
defense primarily among people with lower implicit self-esteem.
To provide converging evidence for the role of implicit self-
esteem, Study 2 tested the hypothesis that an intervention to boost
implicit self-esteem attenuates the effect of mortality salience on
worldview defense. By experimentally manipulating implicit self-
esteem in Study 2, we minimized the impact of potential third
variables that influence measures of implicit self-esteem and af-
forded ourselves the opportunity to find causal evidence that
implicit self-esteem reduces mortality salience effects.
In the implicit self-esteem boost condition, participants saw the
word
I
presented subliminally on a computer screen immediately
prior to the presentation of positive traits (e.g.,
handsome, smart
).
In this fashion, participants were led to associate the positive traits
with themselves (De Houwer, Thomas, & Baeyens, 2001; Riketta
& Dauenheimer, 2003). Past research has verified that this and
similar procedures do indeed increase implicit self-esteem (Bac-
cus, Baldwin, & Packer, 2004; Dijksterhuis, 2004). Participants in
the no-boost control condition saw the same positive traits, but the
traits were not preceded by the word
I
. Hence, these participants
viewed the positive traits but were not led to associate the traits
with themselves, and therefore their implicit self-esteem should
not have changed.
Following the mortality salience induction and the implicit
self-esteem manipulation, participants responded to the same es-
says used in Study 1 as a measure of worldview defense. We
predicted that mortality salience would increase worldview de-
fense unless participants received a boost to their implicit self-
esteem, consistent with the idea that high implicit self-esteem
attenuates defensive reactions to mortality salience.
Results and Discussion
Implicit self-esteem and worldview defense.
Worldview de-
fense was regressed on mortality salience condition, implicit self-
esteem scores, and their centered interaction. Results revealed the
predicted interaction between mortality salience condition and
implicit self-esteem (
–.33),
t
(153)
–3.27,
p
.01. To
decompose the interaction, we assessed the simple effect of mor-
tality salience among participants who were relatively high versus
relatively low in implicit self-esteem (1
SD
above and below the
mean, respectively; Aiken & West, 1991; see Figure 1). Results
indicated that mortality salience increased the difference between
evaluations of the pro- and anti-U.S. essays among participants
low in implicit self-esteem (
.35,
t
3.35,
p
.01), but not
among participants high in implicit self-esteem (
.12,
p
1
We also analyzed responses to the pro-U.S. essay and the anti-U.S.
essay separately. The predicted interaction between mortality salience
condition and implicit self-esteem was not significant for evaluations of the
pro-U.S. essay (
–.11,
t
1) or for evaluations of the anti-U.S. essay
(
.20),
t
(153)
1.77,
p
.09. Thus, although higher implicit self-
esteem was associated with slightly less positive evaluations of the pro-
U.S. essay and somewhat more positive evaluations of the anti-U.S. essay
following mortality salience, only the difference score yielded the pre-
dicted interaction between mortality salience and implicit self-esteem.
1080
SCHMEICHEL ET AL.
18
16
14
12
10
Low Implicit SE (-1 SD)
High Implicit SE (+1 SD)
8
6
4
2
0
D
n
l
P
i
M
r
S
l
e
Figure 1.
Worldview defense as a function of mortality salience condition and implicit self-esteem (Study 1).
Method
bility that subliminal primes were presented during the lexical
decision task (i.e., implicit self-esteem boost). Data from these 2
participants were excluded from all analyses, leaving a final sam-
ple of 90 students.
Participants.
Ninety-two undergraduate students (58 women)
earned credit toward a course requirement by participating. Par-
ticipants were run individually and were randomly assigned to
condition in a 2 (mortality salience vs. dental pain)
2 (implicit
self-esteem boost vs. no boost) between-subjects factorial design.
Procedure.
After providing basic demographic information,
participants wrote about their own death or dental pain. The next
task constituted the manipulation of implicit self-esteem, adapted
from Dijksterhuis (2004). Specifically, participants completed a
lexical decision task on the computer. The task proceeded as
follows: First, a row of
X
s appeared in the center of a computer
screen for 500 ms. The row was immediately followed by the
presentation of the word
I
or the letter
X
presented for 17 ms. The
word or letter was immediately followed by the presentation of 1
of 15 positive words (e.g.,
nice
,
healthy
) or 1 of 15 random letter
strings. Here, participants were to indicate as quickly as possible
whether the stimulus was a word or nonword by pressing one of
two computer keys.
For participants in the boost condition, the word
I
preceded the
presentation of all positive words, and the letter
X
preceded
the presentation of all random letter strings. For participants in the
no-boost condition, the letter
X
preceded the presentation of all
positive words and random letter strings.
Next, participants completed the measure of worldview defense
by rating the same pro-U.S. and anti-U.S. essays used in Study 1
(the order of which was counterbalanced across participants). After
rating the essays, participants proceeded to complete the letter-
rating task used in Study 1 (Nuttin, 1985) so that we could assess
the effectiveness of the implicit self-esteem manipulation.
Last, participants were probed for suspicion, thanked, and de-
briefed. Two participants reported being suspicious of the possi-
Results and Discussion
Manipulation check.
Analyses indicated that the subliminal
boost of implicit self-esteem did indeed increase implicit self-
esteem. A 2 (mortality salience vs. dental pain)
2 (boost vs. no
boost) analysis of variance (ANOVA) indicated the predicted main
effect of boost condition,
F
(1, 86)
3.74,
p
.05 (one-tailed).
Participants in the boost condition rated the letters in their initials
(controlling for baseline ratings of those letters) higher (
M
1.25,
SD
1.40) than did participants in the no-boost condition (
M
0.64,
SD
1.53). None of the other effects approached signifi-
cance (
F
s
1).
Implicit self-esteem and worldview defense.
We predicted and
confirmed that an implicit self-esteem boost reduces worldview
defense following mortality salience. A 2
2 ANOVA on world-
view defense scores revealed a significant interaction between
mortality salience condition and boost condition,
F
(1, 86)
4.65,
p
.05. Simple effects tests revealed that, under mortality sa-
lience, participants who received an implicit self-esteem boost
exhibited a smaller difference between their evaluations of the
pro-U.S. and anti-U.S. essays than did participants in the no-boost
condition,
F
(1, 86)
7.86,
p
.01. In the absence of mortality
salience, however, the boost manipulation did not reliably alter
essay evaluations (
F
1; see Figure 2). Looked at another way,
the simple effect of mortality salience condition approached sig-
nificance in the no-boost condition,
F
(1, 86)
2.06,
p
.16, but
was significant and in the opposite direction in the implicit self-
esteem boost condition,
F
(1, 86)
–3.95,
p
.05. Thus, world-
TERROR MANAGEMENT AND SELF-ESTEEM
1081
18
16
14
12
10
No Boost
Implicit SE Boost
8
6
4
2
0
Dental Pain
Mortality Salience
Figure 2.
Worldview defense as a function of mortality salience condition and implicit self-esteem boost
condition (Study 2).
view defense was reduced by a temporary boost in implicit self-
esteem. Implicit self-esteem appeared to provide a buffer against
the threat of death, thereby attenuating defensiveness.
2
whether mortality salience mainly promotes acceptance of positive
self-views, rejection of negative self-views, or a combination of
both of these.
Study 3
Method
In Study 3, we tested the hypothesis that implicit and explicit
self-esteem combine to influence reactions to mortality salience.
Research has indicated that people with low implicit but high
explicit self-esteem are particularly biased in their self-evaluations
(e.g., Bosson, Brown, Zeigler-Hill, & Swann, 2003; Ziegler-Hill,
2006) and tend to respond to psychological threats in a defensive
and self-serving manner (McGregor & Marigold, 2003, Study 3;
McGregor, Nail, Marigold, & Kang, 2005, Study 1). Accordingly,
we predicted that mortality salience would have its most pro-
nounced effects among participants with low implicit and high
explicit self-esteem.
To provide converging evidence through multiple methods, we
used a different measure of implicit self-esteem from that used in
the previous studies, namely an IAT that measured participants’
speed in associating the self with pleasant versus unpleasant words
(Greenwald & Farnham, 2000). Study 3 also used a different
dependent measure, namely the tendency to prefer highly positive
views of self. Research by Dechesne et al. (2003) found that
mortality salience increases participants’ judgments of the accu-
racy and validity of highly positive personality feedback. We
predicted that mortality salience would increase endorsement of a
positive personality profile primarily among participants with low
implicit and high explicit self-esteem. We also sought to extend the
results of Dechesne et al. (2003) by asking participants to judge the
accuracy and validity of negative personality feedback. Including
both positive and negative personality profiles allowed us to assess
Participants.
Eighty Canadian undergraduates (52 women)
participated in exchange for either Can$5 or credit toward their
introductory psychology course.
Procedure.
Participants were run individually and told the
study examined personality. Participants completed all measures
(except the word-stem completion task; see below) on a computer.
First, participants completed an IAT as a measure of implicit
self-esteem (Greenwald & Farnham, 2000; adapted by Jordan et
al., 2003). This version of the IAT required participants to cate-
gorize words as either being related to or not being related to the
self and as being either pleasant or unpleasant by pressing one of
two computer keys. After practice trials, a critical block of trials
required participants to use one key to categorize words as related
to self or pleasant and another key to categorize words as not
related to self or not pleasant. Then the associations were reversed,
and participants were required to categorize words as either related
to self or unpleasant with one key and to categorize words as either
not related to self or pleasant with another key (i.e., the pairings
2
We again analyzed responses to the pro-U.S. essay and the anti-U.S.
essays separately. The predicted interaction between mortality salience
condition and implicit self-esteem boost condition was not significant for
evaluations of the pro-U.S. essay,
F
(1, 86)
3.26,
p
.08, or for
evaluations of the anti-U.S. essay,
F
(1, 86)
2,
p
.16. As in Study 1,
worldview defense was most evident when the pro-worldview and anti-
worldview evaluations were considered in tandem.
[ Pobierz całość w formacie PDF ]