Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Journal of Personality and Social Psychology Copyright 1998 by the American Psychological Association, Inc.

1998, Vol. 74, No. 3, 590-605 0022-3514/98/S3.00

Terror Management and Aggression: Evidence That Mortality Salience


Motivates Aggression Against Worldview-Threatening Others
Holly A. McGregor Joel D. Lieberman
University of Rochester University of Nevada, Las Vegas

Jeff Greenberg Sheldon Solomon


University of Arizona Skidmore College

Jamie Arndt Linda Simon


University of Arizona University of Arizona

Tom Pyszczynski
University of Colorado, Colorado Springs

The hypothesis that mortality salience (MS) motivates aggression against worldview-threatening
others was tested in 4 studies. In Study 1, the experimenters induced participants to write about
either their own death or a control topic, presented them with a target who either disparaged their
political views or did not, and gave them the opportunity to choose the amount of hot sauce the
target would have to consume. As predicted, MS participants allocated a particularly large amount
of hot sauce to the worldview-threatening target. In Studies 2 and 3, the authors found that following
MS induction, the opportunity to express a negative attitude toward the critical target eliminated
aggression and the opportunity to aggress against the target eliminated derogation. This suggests
that derogation and aggression are two alternative modes of responding to MS that serve the same
psychological function. Finally, Study 4 showed that MS did not encourage aggression against a
person who allocated unpleasant juice to the participant, supporting the specificity of MS-induced
aggression to worldview-threatening others.

Previous terror management research has shown that inducing existence of those with different conceptions of reality. Solo-
people to think about their own mortality increases negative mon, Greenberg, and Pyszczynski (1991) have argued that ef-
evaluations of those who do not share their cultural worldviews forts to defend one's worldview motivated by terror management
(for a recent review, see Greenberg, Solomon, & Pyszczynski, needs may also play a significant role in aggression against
1997). Derogation of dissimilar others presumably serves to worldview-threatening others. However, the measures used in
defend against the threat to their own worldview posed by the studies to date have been self-report evaluations of targets that
participants knew the targets would not see, and thus have not
involved actual harm to the target. If terror management pro-
cesses do indeed play a role in aggression, mortality salience
Holly A. McGregor, Department of Clinical and Social Sciences in (MS) should elicit aggression against those who threaten one's
Psychology, University of Rochester; Joel D. Lieberman, Department of worldview. The purposes of the present research were to (a)
Criminal Justice, University of Nevada, Las Vegas; Jeff Greenberg, Jamie test the hypothesis that MS increases aggression against those
Arndt, and Linda Simon, Department of Psychology, University of Ari-
who threaten one's worldview and (b) investigate the inter-
zona; Sheldon Solomon, Department of Psychology, Skidmore College;
Tom Pyszczynski, Department of Psychology, University of Colorado, changeability of derogation and aggression as alternative modes
Colorado Springs. of responding to MS and worldview threat.
Holly A. McGregor and Joel D. Lieberman share equal responsibility
for this work, which was partially supported by National Science Foun-
dation Grants SBR 9212798, 9312456, and 9601474. We wish to thank Terror Management Theory
Melanie Elliott for her assistance with this project.
Terror management theory begins with the assumption that
Correspondence concerning this article should be addressed to either
humans are similar to other animals in that we share a common
Holly A. McGregor, Department of Clinical and Social Sciences in Psy-
chology, Meliora Hall, University of Rochester, Rochester, New \brk evolutionary heritage, which includes a strong biological pro-
14627, or Joel D. Lieberman, Department of Criminal Justice, University clivity for self-preservation or continued existence. Humans are
of Nevada, 4505 Maryland Parkway, Box 455009, Las Vegas, Nevada different from other species, however, in having unique cognitive
89154-3007. Electronic mail may be sent to mcgregor@scp.rochester. abilities that, among other things, make them aware of iheir own
edu or jdl@nevada.edu. existence and the possibility that their existence may end when

590
TERROR MANAGEMENT AND AGGRESSION 591

they die. This awareness, when juxtaposed with the drive to stay Wars and persecutions are, at bottom, expressions of rivalry be-
alive, creates the potential for paralyzing terror. According to tween contending claims to immortality and ultimate spiritual
terror management theory, this potential for terror is managed power. Religious victimization is a one-sided version of that process
by a dual component cultural anxiety buffer, consisting of (a) with the specific psychological functions of rinding a target for
death anxiety. . . . (Lifton, 1983, p. 315)
a personalized version of a cultural worldview, which gives
subjective reality a sense of permanence, order, and meaning; a Of course, no single study or small set of experiments is going
set of standards that enable individuals to attain a sense of to definitively establish a role of terror management in past
personal value; and, if these standards are met, a promise of or ongoing historical events. However, if terror management
either literal or symbolic immortality; and (b) self-esteem, the processes do play a role in actual aggression, then MS should
belief that one is living up to the standards and values prescribed encourage aggression against a person who threatens one's cul-
by the cultural worldview. Because self-esteem and the cultural tural worldview.
worldview from which it is derived are so important in the
But how would aggression against a worldview threatener
management of existential terror, a large part of human social
serve a terror management function? Whereas derogation of
behavior is oriented toward their maintenance and defense.
worldview threateners or their arguments allows one to ratio-
Although treated by the individual as absolute reality, the nally dismiss the criticisms of one's worldview, aggression does
cultural worldview is a fragile social construction in need of not. There are, however, three interrelated ways in which aggres-
constant validation from others. Consequently, the existence of sion may serve a terror management function and therefore be
others who share one's worldview bolsters faith in that a likely response to worldview threat under conditions of MS.
worldview, thus increasing its effectiveness as an anxiety buffer, First, it may occur as a simple lashing out at the threatener.
and the existence of others who do not share one's worldview After mortality is salient, people need faith in their worldviews
threatens one's faith in it, thus reducing its effectiveness as an and so are likely to be displeased with anyone who threatens
anxiety buffer. As a result, people generally respond favorably that faith; aggression may be a primitive and particularly danger-
to those who share their worldviews and negatively to those ous way to express this displeasure.
who do not. Accordingly, a series of studies have shown that Such a response may be particularly likely because, as recent
MS leads to especially positive evaluations of those who validate research indicates, MS effects occur primarily when people are
one's worldview and especially negative evaluations of those in an experiential mode of thinking rather than a rational, ana-
who challenge it (e.g., Greenberg et al., 1990; Rosenblatt, lytic mode. In three studies, Simon et al. (1997) found that MS
Greenberg, Solomon, Pyszczynski, & Lyon, 1989). led to derogation of a worldview threatener if experimental
These effects have been replicated in other laboratories in the conditions encouraged an experiential mode in participants but
United States (e.g., Nelson, Moore, Olivetti, & Scott, 1997), not if the conditions encouraged a rational and analytic mode.
Canada (e.g., Baldwin & Wesley, 1996), Israel (e.g., Florian & This suggests that terror management responses to MS are expe-
Mikulincet, 1997), Germany (e.g., Ochsmann & Mathy, 1994; riential (from Epstein's, 1994, perspective: intuitive, primitive,
Ochsmann & Reichelt, 1994), and the Netherlands (e.g., M. emanating from the unconscious) rather than rational. Certainly
Dechesne, personal communication, March 1997). The effects aggression often seems to qualify as such an experiential re-
appear to be unique to thoughts of mortality; they have not been sponse. Indeed, when responding experientially, aggression may
found in response to thoughts of failure, dental pain, one's next serve as a defense even though it does not rationally refute the
important exam, giving a speech in public, or worries about life attack; just as a drunk person who is outwitted may resort to
after college, or in response to a laboratory failure experience violence, an MS participant whose worldview is under attack
(Greenberg, Pyszczynski, Solomon, Simon, & Breus, 1994; may do so when the opportunity is there.
Greenberg, Simon, Harmon-Jones, Solomon, Pyszczynski, & Finally, to the extent that aggression asserts one's convictions
Lyon, 1995; Greenberg et al., 1997). regarding one's worldview, it may be a way to defend one's
All of the studies to date investigating the effects of MS on worldview despite not logically addressing criticisms. As an
reactions to others have involved only pencil- and-paper mea- example, after hearing one's country berated, in lieu of refuting
sures in which participants rate target individuals on intelli- the arguments, one can go and dig a flag out of the closet
gence, likability, and so forth. However, Becker (1971, 1973, and display it proudly in the front of one's house. Similarly,
1975) and others (e.g., Harrington, 1969; Lifton, 1983) have aggressing against another who attacks one's views can serve
proposed that the fear of death has greatly contributed to many to reassert one's own faith in one's worldview, the criticisms
of the egregious historical examples of violence and war. Simi- be damned! Thus, given what we know so far about terror
larly, Solomon et al. (1991) posited that reactions to the threat management processes, there are good reasons to believe that
posed by an alternative conception of reality include not only mortality salience will encourage aggression against a
derogation but aggression as well. Indeed, historically, many worldview threatener. Our first study was designed to test this
have tried to neutralize a threat to their worldview by attempting hypothesis.
to annihilate those who are different. What more extreme way
to retaliate against a threatening worldview and assert one's
Measuring Aggression
own convictions than to destroy or intentionally hurt those who
hold them? Consider, for example, the recent crisis in Bosnia, Aggression is commonly denned by social psychologists as
in which the Serbs and the Bosnians were effectively destroying behavior with intent to harm the individual who is its object
each other because of their cultural, ideological, and religious (e.g., Baron & Richardson, 1994; Berkowitz, 1993; Geen,
differences: 1990). Several different laboratory measures of aggression have
592 MCGREGOR E T AL.

been used over the years, including verbal assault (e.g., De- exam. They were then asked to read an essay, supposedly written
Charms & Wilkins, 1963; Wheeler & Caggiula, 1966), direct earlier in the session, that was derogatory toward either conser-
physical aggression involving noninjurious acts such as hitting vatives or liberals. Participants then allocated a quantity of very
with a Styrofoam sword (Diener, 1976; Diener, Dineen, Endre- spicy hot sauce to the author of the essay, knowing that the
sen, Beaman, & Fraser, 1975), and administering electric shock author did not like spicy foods and would have to consume the
(e.g., Berkowitz, 1964; Buss, 1961; Hammock & Richardson, entire sample of hot sauce. We predicted that, relative to the
1992a, 1992b; Taylor, 1967). Each of these measures has yielded other groups, MS participants who read an essay criticizing
some evidence of validity (e.g., Bernstein, Richardson, & Ham- their political stance would assign the author an especially large
mock, 1987; Diener et al., 1975; Shemberg, Leventhal, & All- quantity of hot sauce.
man, 1968), but each has certain ethical or practical limitations.
For our purposes, we wanted a face-valid measure of aggres- Method
sion that clearly indicated an intent to physically harm the target.
Probably the best existing measure of this nature is the amount Participants. Thirty-eight male and 36 female American introduc-
tory psychology students at the University of Arizona participated in
and intensity of electric shock that participants administer to
partial fulfillment of a course requirement. Participants were recruited
targets (e.g., Buss, 1961; Diener et al., 1975). However; the on the basis of a pretest measure identifying them as having either liberal
association of the shock paradigm with the Milgram obedience or conservative political views. The measure consisted of a single 9-
research presents problems with approval from university hu- point bipolar scale that ranged from extremely conservative (1) to ex-
man subjects committees and with participants linking the study tremely liberal ( 9 ) . Participants who were moderately conservative ( 2 -
to the well-known Milgram research. Also, as Hammock and 4) and moderately liberal ( 6 - 8 ) were selected. Three participants were
Richardson (1992b), among others, have noted, in the shock excluded after reporting suspicion about the procedures, and one partici-
paradigms, the intensity of shock could be associated with inten- pant was excluded because she had seen the news report about the
tions besides hurting the other (e.g., being a good teacher). Denny's chef's aggressive use of hot sauce and thought about it during
the experiment.
Because of these concerns, we sought to devise a new means
Materials and procedure. Participants were randomly assigned to
to measure intent to physically aggress against another. The key
conditions in a 2 (MS vs. control) x 2 (worldview threaten vs.
was to provide the participant with an opportunity to engage in worldview consistent) between-subjects design and were run in groups
a behavior that could cause physical harm to another individual. of 4, consisting of either all liberal or all conservative participants. The
We consequently devised a context in which participants deter- experimenter was blind to the participant's political stance, as well as
mined the amount of hot sauce to be consumed by another to experimental conditions.
person who does not like spicy foods. Because prior terror management research has shown that MS has a
Thanks in part to our extensive personal research with South- greater effect when participants are in an experiential mode (Simon et
western cuisine, it was clear to us that hot sauces can indeed al., 1997), the experimenters were trained to address participants in a
inflict pain. There are also a number of well-publicized incidents relaxed and casual manner. Upon arrival in the laboratory, participants
were led to believe that they would be taking part in two experiments
of hot sauce being used in malevolent efforts to harm others,
investigating the relationship between personality and attitudes.
including child abuse (e.g., Arizona Daily Wildcat, 1995;
Bogus Study ] materials and procedure. The first experiment was
Brooks, 1992; Columbus (Ohio) Dispatch, 1992; Donahue, described as an examination of the relationship between personality
1989; Martin, 1994; Milne, 1995; "N.Y. foster mother," 1993; traits and how people form impressions of others. Participants were told
Serrano, 1989; "Thirty-nine," 1992). In one case, a cook at a that they would write a short paragraph describing their opinion of
New Hampshire Denny's restaurant spiked the food of two state politics in the United States and exchange those essays with another
troopers with Tabasco sauce after they had stopped there to eat person. The participants were then led to individual cubicles and given
breakfast. One of the officers reported that his mouth was a sheet of paper that reiterated these instructions. After telling the partici-
burned, and the other said that the food caused his stomach to pants they would have 5 min to work on their paragraphs and not to be
become upset. The cook was subsequently arrested for assault. concerned with spelling or grammar, the experimenter left the cubicle.
A witness, who claimed to observe the incident, reported that After 5 min, the experimenter returned, collected the paragraphs,
the cook did not like police officers and had acted intentionally handed each participant a packet that consisted of three questionnaires,
to harm the troopers {Arizona Daily Wildcat, 1995). A similar and told them to work through the entire packet in order. Tb encourage
an experiential mode of processing, the experimenter told participants
aggressive act was portrayed in the recent film Mrs. Doubtfire,
to give their first, natural response when answering the questions. The
in a scene in which the character played by Robin Williams put first page consisted of a filler personality inventory, and the second
an excessive amount of cayenne pepper on the food being served consisted of the MS manipulation (Rosenblatt et al., 1989). MS partici-
to a man who was romantically interested in his ex-wife. These pants were instructed to "Please briefly describe the emotions that the
incidents indicate that spicy food has indeed been used in both thought of your own death arouses in you" and to "Jot down as specifi-
real world acts and media portrayals of aggression. cally as you can, what you think will happen to you as you physically
die, and once you are physically dead." Control participants were in-
structed to ' 'Please briefly describe the emotions that the thought of your
Study 1 next important exam arouses in you'' and to ' 'Jot down as specifically as
you can, what you think will happen to you as you physically take your
Tb assess the hypothesis that MS increases aggression against next exam, and once you are physically taking your next exam.'' This
those who threaten one's worldview, we used the amount of hot control condition concerns a future aversive event other than death and
sauce given to a fellow participant as the dependent measure of has been used successfully in past terror management research (e.g.,
aggression. Politically conservative and liberal participants were Greenberg et al., 1995). Participants then completed the Positive and
induced to think about their own death or their next important Negative Affect Schedule (PANAS-X; Watson & Clark, 1991).
TERROR MANAGEMENT AND AGGRESSION 593
After completion of the personality packet, the experimenter returned to complete, which consisted of six rating scales to evaluate preferences
to the cubicle and handed each participant an essay that was purportedly for different tastes and textures. The items evaluated were sweet, sour,
written by another participant but was in fact a bogus essay used to creamy, salty, spicy, and dry. Participants evaluated these items using a
manipulate worldview threat. The writer of the essay was identified only 21-point rating scale that ranged from no liking at all (1) to extreme
by a three-digit number. Consequendy, although participants believed liking (21), These were included so that participants later could be
the writer of the essay was present in the same session, they did not presented with a bogus taste preference inventory purportedly completed
know who it was. One essay was designed to conflict with the worldview by the person to whom they were giving the hot sauce.
of liberal, but not conservative, participants: Although participants were led to believe that two groups existed,
everyone was actually put into the same group, which sampled the dry
Don't even get me started. Liberals are the cause of so many prob- food and allocated the hot sauce. After completing die taste preference
lems in this country, it1 s not funny. Not only that, but they get in inventory, the experimenter returned and informed participants that they
the way of decent Americans who are trying to solve all those had been randomly assigned to the dry food group and that another
problems that they created in the first place. The bleeding heart participant had selected a sample of cracker for them to eat. At this
stance they take, of trying to help everyone is a joke and incredibly point the participant was handed an envelope which contained a saltine
stupid. How can they help the world when they can't even help cracker and was instructed to consume the entire cracker and then to
themselves?! Do Liberals put any thought into what they believe? evaluate it using a 9-point scale that ranged from no liking at all (1)
1 don't think so. If they did, they would realize that they are ruining to extreme liking (9). The experimenter then left the room.
die country. Thankfully there are people in power that agree with After several minutes, the experimenter returned carrying a tray with
me who can, and will control the whiny Liberals, and put them in hot sauce, a Styrofoam cup with a lid, a plastic spoon, a wooden spoon,
their place. The best place for a Liberal is out of my sight. They and a cup of water. Participants were informed that they would be
make me sick. preparing a sample of hot sauce to give to another person and that, to
avoid confusion, they would be giving the hot sauce to the person whose
Thus, liberal participants who read this essay had their worldview threat- paragraph they read earlier. The experimenter then mentioned that be-
ened, but conservative participants did not. The other essay was designed cause people are often curious as to the taste preferences of the other
to conflict with the worldview of conservatives, but not that of liberals: participants they would be shown the taste preference inventory of the
person to whom they would be giving me hot sauce. Participants were
Don't even get me started. Conservatives are the cause of so many handed a bogus taste preference inventory indicating that the target
problems in diis country, it's not funny. Not only that, but they get disliked spicy food. For the item "spicy—like hot salsa," the target
in the way of decent Americans who are trying to solve all those indicated a liking of 3 on the 21-point scale (where 1 indicated no liking
problems that they created in the first place. The cold-hearted stance at all).
they take, of trying to help only themselves is a joke and incredibly Participants were instructed to use the plastic spoon to place a quantity
stupid. They are too busy thinking of themselves, and don't care of hot sauce into a 4-oz Styrofoam cup and seal it with the lid that was
about anyone else. Do Conservatives put any thought into what provided.1 The participants were told that all quantities of hot sauce
they believe? I don't think so. Tf they did, they would realize that were useful and to put in as much or as little hot sauce as they wanted.
they are ruining the country. Thankfully there are people in power However, it was made clear to the participants that the person who
that agree with me who can, and will control the selfish Conserva- received the hot sauce would have to consume the entire quantity of it.
tives, and put them in their place. The best place for a Conservative To be sure the participants were aware of the intensity of the hot sauce
is out of my sight. They make me sick. (which was quite hot), they were instructed to taste a sample of it with
the wooden spoon prior to allocating it to the target Participants were
Consequently, regardless of whether participants were liberal or conser- told that because the cups were covered, the experimenter would not
vative, half read an essay that threatened their worldview and half did know how much hot sauce had been allocated. They were again reminded
not. All participants were asked to evaluate the essay using an "impres- that they would be giving it to the person whose essay they read and
sion formation" form designed to allow them to evaluate the essay in were instructed to write that person's number on the Styrofoam cup.
a nonjudgmental way. This was important because of the possibility that Thus, it was clear who would receive the hot sauce they prepared and
a negative evaluative judgment about the essay at this point might pro- that their behavior would be anonymous.
vide them with an opportunity to defend their worldview, as in previous Before leaving the cubicle, the experimenter handed the participants
MS studies (e.g., Greenberg et al., 1990), thereby possibly negating the a checklist to remind them of each step involved in allocating the hot
need to aggress against the author. (We addressed this possibility directly sauce. At the bottom of the checklist, the participants were asked to
in Study 2.) The impression formation form consisted of five questions. indicate the number of spoonfuls allocated. To maintain a sense of ano-
Participants were asked to guess the college major, the sex, and the age nymity, the participants were asked to place this form inside an envelope.
of me author, what part of the country the author was from, and whether At this point the experimenter left the cubicle, and the participants
the author was reserved or outgoing. After completing this evaluation, allocated the hot sauce to the target.
participants were informed that the first study was completed. Participants were then asked to indicate the type of food they had
Bogus Study 2 materials and procedure. The second study was de-
scribed as an examination of the relationship between personality and
1
food preferences. Participants were informed that they would be tasting The selection of hot sauce used in the experiments proved to be a
and giving their impression of a food sample, that in different experi- challenging endeavor. Because die hot sauce allocation had to be
ments different types of foods were being tested, and that dry foods and weighed, it was critical to use a hot sauce that had an even consistency,
spicy foods were being examined that day. In addition, they were told that would be quite hot, and that would ordinarily be applied in a volume
that because the experimenter needed to be blind to certain specifics greater than a drop or two. As a result of these restrictions, we created
about the sample type and quantity of food tasted, it was necessary for our own hot sauce by combining several available commercial products
participants to administer the food samples to each other. This provided in the following proportions: 5 parts Heinz chili sauce and 3 parts
a plausible explanation as to why the participants, rather than the experi- Tapatio salsa picante hot sauce, produced by the Empacadora Company.
menter, were allocating the food to each other. The final hot sauce had a consistency similar to cocktail sauce and was
The experimenter then gave participants a taste preference inventory indeed quite hot.
594 MCGREGOR ET AL.

tasted and the type of food they had given the target. Finally, participants Table 1
answered three other questions using 21-point rating scales. The ques- Study I: Cell Means for the Two-Way Interaction of Mortality
tions were "To what extent did you use the TASTE PREFERENCE INVENTORY Salience X Target on Weight (in Grams), and Self-Report
when giving out the food sample to the other person?," on a scale
Estimates (in Spoonfuls) of Hot Sauce Allocated
ranging from not at all (1) to completely (21); "How useful do you
think die TASTE PREFERENCE INVENTORY was when giving out the food Mortality ;salient Exam control
sample?,'' on a scale ranging from not at all useful (1) to extremely
useful (21); and "Using the scale below, indicate the extent to which Target M SD M SD
the person you gave die food sample to liked that kind of food,'' on a
scale that was the same as on the taste preference inventory and ranged Weight
from no liking at all (1) to extreme liking (21).
Worldview threatening 26.31 21.87 15.20 8.39
Participants then left their cubicles and were debriefed. During die Worldview consistent 11.86 8.33 17.56 11.34
debriefing, special care was taken to ensure that the participants did not
leave with any negative feeling that may have come from either reflection Self-report estimates
on their own mortality or die knowledge that they may have caused
Worldview threatening 2.11 1.88 1.39 0.89
someone pain and distress. We assured participants that no one had
Worldview consistent 1.09 0.80 1.61 1.09
tasted the hot sauce they had administered. In addition, we explained
to them that they were simply responding to the situation that we put
them in and that in no way should they feel as though they were an
aggressive or otherwise bad person for giving the hot sauce to the target.
exam salient worldview-consistent conditions. This comparison
No participant objected in any way to the procedure or indicated distress,
was not significant, t(60) = .49, p > .62, and indicates that in
and many said they were glad they had participated.
the exam condition, type of essay had no effect on amount of
hot sauce allocated. Thus, as predicted, MS worldview threat
Results participants allocated a significantly greater amount of hot sauce
to a woridview-threatening target than did participants in the
The primary analyses to be reported are 2 (MS vs. exam salient)
other conditions.
X 2 (woridview-threaterring target vs. worldview-consistent target)
A secondary dependent measure was participants' own esti-
between-subjects analyses of variance (ANOWs). 2
To assure that participants were aware of the target's dislike
of hot sauce, the final questionnaire assessed the extent to which 2
Initial analysis of the primary dependent measure in all four studies
participants recalled and used that information.3 The target os- revealed heterogeneity of variance (all F s ^ > 4.01, ps < .05). This
tensibly had rated hot sauce as a 3 on a 21-point scale. The heterogeneity resulted from high variance within the MS woridview-
mean response for recall of the target's rating of liking hot threatening conditions. Violation of this assumption of ANOVA is not
sauce, which did not differ by condition (Fs < 2.82), was 3.70. considered a serious problem when cell /is are equal (Kirk, 1995).
Mean responses for the questions assessing usefulness of the Therefore, in the primary analyses in Studies 1 and 2, we randomly
dropped participants from the data set so that cell sizes were equal across
taste inventory and the extent to which they used it (which did
conditions (8 participants were dropped in Study l,and 1 participant was
not differ by condition; Fs < 1), were 17.3 and 16.2. Thus, dropped in Study 2). Although heterogeneity of vari ance was a problem
participants in all conditions were equally aware that the target in Studies 3 and 4, cell sizes were equal, and consequently, no partici-
did not like spicy food and used this rating in deciding how pants were dropped. An alternative solution for dealing with the hetero-
much hot sauce to allocate. geneity of variance problem is to conduct a square root transformation
MS and aggression. The main dependent measure of aggres- on the entire data set (Kirk, 1995). The results of that transformation
sion was the weight in grams of hot sauce allocated to the were very similar to those obtained by randomly dropping participants.
targets, determined by weighing the hot sauce in the Styrofoam For Study 1, when all participants were included in the analyses, the
cups on an OHaus Cent-O-Gram triple beam balance. An critical two-way interaction between MS and target was marginally sig-
nificant for both die raw and square root transformed means, both Fs( 1,
ANOVA revealed a marginally significant main effect of target,
68) > 3.85, ps < .057, and the outcome of the a priori contrasts
F(l, 60) = 3.13, p < .09, reflecting the general tendency for paralleled those reported in the text. For Study 2, a one-way ANOVA
participants to allocate more hot sauce to the woridview-threat- based on the entire data set yielded a marginally significant effect for
ening target than the worldview-consistent target. This effect the raw means, F(2, 49) = 3.12, p = .053, and a significant effect on
was qualified, however, by the expected MS X Target interaction, the square root transformed means, F(2, 49) = 3.42, p < .05. Again,
F(l, 60) = 6.06, p < .02. Means are displayed in Table 1. the results of the orthogonal contrasts parallel those reported in the text.
A planned orthogonal contrast provided strong support for However, throughout the body of the text and in the tables, we present
our main hypothesis: MS participants allocated significantly results based on equal n analyses, because these analyses allowed us to
more hot sauce to the author of the woridview-threatening essay use and display the raw means while minimizing the heterogeneity prob-
than did participants in the other three conditions, r(60) = lem. Ft should also be noted that initial analyses for all studies mat
included gender and political orientation as independent variables re-
2.90, p < .01. A second orthogonal contrast compared the MS
vealed only one significant effect, and that effect did not qualify the
worldview-consistent condition with the two exam conditions. primary results. Specifically, in Study 1, there was an MS X Gender
Previous research suggests that worldview-supportive targets are interaction, F ( l , 47) = 4.71, p < .04, on hot sauce allocation, which
reacted to especially positively following MS induction. resulted from the fact that, collapsing across targets, MS males allocated
Whereas the means were consistent with this possibility, this an especially large amount of hot sauce.
contrast was not significant, r(60) - 1,08, p > .28. A third 3
Unfortunately, in Study 1, these measures were not included until
contrast compared the exam salient woridview-threatening and roughly halfway through the first study (n = 30).
TERROR MANAGEMENT AND AGGRESSION 595

mates of how many spoonfuls of hot sauce they had allocated studies have shown that MS engenders more positive evaluations
to the target. An ANOVA on this measure revealed the expected of those who uphold important aspects of the cultural worldview
significant MS x Target interaction, F( 1, 60) = 4.04, p < .05. (e.g., Greenberg et al., 1990); one might therefore have ex-
Means are displayed in Table 1. pected to find significantly less aggression (i.e., smaller alloca-
A planned orthogonal contrast comparing the estimates of tions of hot sauce) toward the worldview-consistent target in
hot sauce allocated by MS participants exposed to a worldview- the MS than in the exam salient control conditions. Whereas
threatening essay to estimates of participants in the other three the means were in the expected direction, the contrast was not
conditions was significant, f (60) = 2.09, p < .05. Thus, partici- significant. Although some other MS studies have also failed to
pants' self-reports of hot sauce allocation produced the same find a significant increase in rating of the pro-worldview target
pattern as their actual hot sauce allocation. A second orthogonal (e.g., Greenberg et al., 1992), the nature of the worldview-
contrast compared the MS worldview-consistent condition with consistent target used in this study may have played a role.
the two exam conditions and was not significant, f(60) = 1.08, Although the essays were designed to either threaten or not
p > .28. This indicates that MS did not lead to especially low threaten participants' worldviews, they were not designed to
self-reports of hot sauce allocated to the worldview-consistent directly support them. For example, liberal participants in the
target. The third contrast compared the two exam conditions threat condition read an essay claiming that liberals were ruining
and was also not significant, t(59) = .50, p > .61. the country, whereas liberal participants in the consistent condi-
Affect measures. The PANAS-X was administered directly tion read an essay claiming that conservatives were ruining the
after the MS manipulation. Two-way ANOVAs on each of the country. Although this latter essay does not attack the partici-
13 subscales revealed no effects on affect (Fs < 1.5). To further pant's worldview, it does not necessarily support it either, and
investigate the role of affect in MS effects, we examined within- thus there is no guarantee that participants would agree with
cell correlations between each of the subscales of the PANAS- the sentiments expressed in it. Thus, if participants had read
X and both weight and self-report measures of hot sauce alloca- an essay directly supporting their worldview (e.g., if liberal
tion. With the exception of the hostility subscale, none of the participants read an essay stating that liberals were wonderful
within-cell correlations approached significance (allps > ,10). people and their beliefs were the solution to the country's prob-
To the extent that hostility is indicative of aggressive propensi- lems), the tendency to allocate less hot sauce to the worldview-
ties, it is not surprising that within both the MS worldview- consistent author may have been stronger.
threatening condition and the exam salient worldview-threaten- The aggression measure. Given that hot sauce allocation to
ing condition, hostility was positively correlated with the weight a person who dislikes spicy foods is a new measure of aggres-
measure of hot sauce allocation, r(14) = .58, p < .05, and sion, it is important to consider the extent to which it validly
r( 14) = .51, p < .05, respectively. Within the MS worldview- taps the aggression construct. The fact that the measure yielded
threatening condition, self-report hostility was also positively theoretically predicted results suggests that it has construct va-
correlated with self-report estimates of hot sauce allocation, lidity. Furthermore, given prior evidence of a relationship be-
r(14) = .52, p < .05, but this correlation did not reach signifi- tween hostility and aggression (see, e.g., Geen, 1990), the fact
cance in the exam salient worldview-threatening condition, MS that hostility and hot sauce allocation were significantly corre-
had no effect on the hostility subscale (F < 1); therefore, this lated in both MS and control worldview-threatening conditions
measure probably reflected a preexisting individual difference provides additional evidence of its validity as an index of ag-
in level of hostility. Thus the correlation might best be interpre- gression. Further evidence of the validity of this measure in-
ted as indicating that people high in hostility, at either a trait or cludes reports in the debriefings by a number of participants
state level, were more aggressive toward the worldview threat- that they intended to create discomfort in the recipient and the
ener but not toward the worldview-consistent target.4 Whether newspaper accounts of actual incidents of aggression using hot
the hostility differences of participants who came into the study sauce that were cited in the introduction.
reflected differences in trait hostility or just their current state To gain additional information about the use of hot sauce as
cannot be determined from the data. a means of aggressing, using an experimenter unaware of our
purpose or the findings of Study 1, we also asked a separate
sample from the same participant pool to taste the hot sauce,
Discussion
rate it, and, on viewing a cup with 26.3 g of the hot sauce
The results of Study 1 clearly supported the hypothesis that (the mean amount allocated in Study 1 in the MS worldview-
MS leads to aggression against worldview-threatening others.
Specifically, participants in the MS worldview-threatening con-
4
dition allocated significantly more hot sauce to targets than did Fbr all four studies, we conducted a variety of medianonal analyses
participants in the three other conditions, and a self-report mea- to examine the possibility that affective responses to MS mediate its
sure of hot sauce allocation yielded a similar pattern of results. effects on aggression. As in previous studies, no evidence of such af-
A priori contrasts indicated that these findings did not simply fective mediation was found. Details of these analyses are available on
request from the authors. Whereas these and previous data suggest that
result from participants having their worldview threatened.
these effects are not mediated by affect, recent evidence indicates that
There was no significant difference in the amount of hot sauce they are mediated by a delayed increase in the accessibility of death-
given to the two targets in the exam control conditions. Rather, related thoughts. Further, death-related thoughts have been shown not
it was the juxtaposition of MS and worldview threat that led to to be accompanied by increases in arousal or affect (Arndt & Greenberg,
a significant increase in aggressive behavior. 1996; Arndt, Greenberg, Solomon, Pyszczynski, & Simon, 1997; for an
Treatment of the worldview-consistent target. Some prior overview of the relevant evidence, see Greenberg et al., 1997).
596 MCGREGOR E T AL.

threatening condition), to indicate how much pain or discomfort leagues (e.g., Zillmann, 1979) clearly demonstrates that excita-
would be caused by consuming that amount of hot sauce. The tion transfer occurs within a precise, highly limited time frame.
hot sauce was rated quite hot, 7.2 (n = 10) on a 9-point scale Thus, the existing evidence regarding the MS treatment used in
with endpoints of not at all hot (1) and extremely hot (9). this research casts serious doubt on explanations of these find-
Similarly, the mean amount allocated in the MS worldview- ings that are based on negative affect or arousal.
threatening condition was viewed as likely to cause considerable
pain or physical discomfort, with a mean of 7.8 (n — 10) on a
Study 2
9-point scale with endpoints of not much discomfort (1) and
extreme discomfort ( 9 ) . Thus, the hot sauce allocation measure In Study 1, participants were asked to evaluate the target
seems a reasonable way to assess aggression that may be useful using an "impression formation" form, which contained en-
in other research on aggression. tirely nonevaluative questions about the target. This was done
Affect and arousal as alternative explanations. The terror because of the possibility that if participants had the opportunity
management explanation for the findings of Study 1 is that to report a negative evaluation of the target before allocating
MS intensifies participants' need to defend their worldview; hot sauce, they might take advantage of it to defend their
therefore they lash out at the worldview threatener. One alterna- worldview. Of course, in previous terror management studies
tive explanation of the present findings would be that MS created (e.g., Greenberg et al., 1990), worldview defense has been
negative affect, which then triggered aggression (e.g., Berko- shown to occur in this way. If worldview defense occurred at
witz, 1993; Geen, 1990). However, the affect measures did not this point, it is possible that participants would not also need
reveal any evidence that MS created more negative affect than to defend their worldview by later aggressing against the targets.
the exam control, and correlational analyses did not reveal any Consequently, one factor that might reduce aggression toward
evidence that the effects of MS were mediated by affect. worldview-violating targets is the opportunity for participants
Another possibility is that MS generated physiological to first derogate the worldview threateners. Similarly, if partici-
arousal, which then intensified the negative reaction to the pants are able to defend their worldview by aggressing against
worldview threatener through either the enhancement of domi- the target, they may not subsequently need to derogate them. A
nant responses or excitation transfer (Zillmann, 1979). Al- second study was conducted to examine these possibilities as
though conceivable, we believe these alternatives are highly un- well as to replicate the primary findings of Study 1.
likely for a variety of reasons. First, using measures of skin It is important to examine these hypotheses, because doing
conductance, blood pulse volume, and heart rate, two prior stud- so should shed some light on the nature of terror management
ies failed to find any evidence of increased physiological arousal and responses to MS. The existing theory doesn't precisely
in response to this exact MS treatment (Arndt & Greenberg, specify whether in a given instance MS will motivate a variety
1996; Rosenblatt et al., 1989). Second, if MS created arousal, of negative responses to a worldview threatener or whether one
it would most likely be labeled as affect and lead to elevated response will serve to sufficiently bolster the worldview, thereby
reports of negative affect; this did not happen in the present eliminating the need for other defensive responses. Although
research—or in over 30 prior MS studies (for an overview of this is likely to depend in part on the strength of the threat, our
this entire body of work, see Greenberg et al., 1997). hunch is that for most threats, terror management processes
Third, in two prior MS studies, whereas MS led to increased would function much like dissonance reduction. Evidence from
worldview defense but no negative affect, comparison condi- dissonance research indicates that once people use one mode
tions did not increase worldview defense but did generate nega- of dissonance reduction, they are less likely to use other modes
tive affect (Greenberg et al., 1995). If this negative affect in- (e.g., Gotz-Marchand, Gotz, & Irle, 1974; Scheier & Carver,
volved arousal, this work suggests that neither negative affect 1980; Simon, Greenberg, & Brehm, 1995). Therefore, we hy-
nor arousal is sufficient to produce worldview defense. Fourth, pothesized that the use of aggression would reduce or eliminate
prior research clearly shows that subtle MS treatments lead to the use of derogation, and vice versa. Of course, this would
higher levels of worldview defense than more impactful ones occur only if aggression does indeed serve the same defensive
and that MS effects are stronger after a delay than they are function that derogation does. Thus, the results of this study
immediately after the induction (e.g., Greenberg et al., 1994). may also address the issue raised in the introduction concerning
Presumably, if there were some as yet undetected arousal re- the function of aggression by MS participants in response to
sulting from MS treatments, arousal and dominant response worldview threat.
amplification would be higher with stronger treatments and di- In this study, all participants read a worldview-threatening
rectly after the treatment than with subtler treatments and after essay and were given the opportunity to both allocate hot sauce
a delay (see Footnote 4 ) . and critically evaluate the author. Because Study 1 established
Although it is true that excitation transfer effects rely on a that MS does not encourage aggression against a target who
short delay following a substantially elevated level of arousal, does not threaten the worldview, and previous work has shown
such effects are also unlikely to be involved in this and other that MS does not encourage derogation against such targets,
MS research. In addition to the fact that our treatment clearly only the worldview-threatening target was used in Study 2. Parti-
does not produce a large initial increase in arousal, the plausibil- cipants thought of their own mortality or dental pain and were
ity of the excitation transfer explanation is further reduced be- given an opportunity to aggress against a worldview threatener.
cause MS effects have been found in studies that varied consid- Half the MS participants were given the opportunity to allocate
erably in the length of time between the MS treatment and the the hot sauce prior to evaluating the target, and half were given
dependent variables, whereas the work of Zillmann and col- the opportunity to evaluate the target prior to allocating the hot
TERROR MANAGEMENT AND AGGRESSION 597

sauce. The dental pain control participants allocated hot sauce using a 9-point rating scale with endpoints of not at all (I) and totally
first and then evaluated the target. Because there was no reason (9). Participants then sealed these forms in a blank envelope.
to expect high levels of aggression or derogation in the control After completing either the evaluative or nonevaluative impression
condition, order of dependent variables was not manipulated in formation form, all participants were asked to complete the taste prefer-
ence inventory. As in Study 1, participants were shown a bogus version
this condition.
of the taste preference inventory, indicating that the target did not like
Consistent with the findings of Study 1, we expected that MS hot sauce. The participants were then asked to allocate hot sauce to the
participants who were not first asked to evaluate the target would target and to give a self-report of the number of spoonfuls of hot sauce
allocate a greater amount of hot sauce to a worldview-threaten- administered. Following this, manipulation checks were delivered. Parti-
ing target than would other participants. However, such in- cipants were then given a second impression formation form. Participants
creased aggression should be reduced or eliminated among MS in the MS-evaluate-first condition, who had previously completed the
participants who were first allowed to defend their worldview. evaluative impression formation form, were given the nonevaluative im-
Similarly, we expected that MS-evaLuate-first participants pression formation form. Participants in the MS-evaluate-last and con-
would express more negative attitudes than the other participants trol conditions, who previously had completed the nonevaluative impres-
toward the target, but that such derogation would be reduced sion formation form, were presented with the evaluative impression
formation form. Finally, all participants were debriefed, thanked, and
or eliminated when MS participants were first given an opportu-
dismissed.
nity to aggress against the target.

Results
Method
The primary analysis to be reported is a one-way ( M S -
Participants. Twenty-seven male and 31 female introductory psy-
chology students at the University of Arizona participated in partial evaluate-last vs. MS-evaluate-first vs. dental pain) A N O W
fulfillment of a course requirement. Participants were recruited on the The mean response for recall of the target's rating of liking hot
basis of the same pretest used in Study 1 to classify them as either liberal sauce was 2.86 (which did not differ by condition, F — .12).
or conservative. Three participants were excluded from the analyses after For the question assessing the usefulness of the inventory, the
reporting suspicion about the experimental procedures, 1 was excluded mean response was 16.2 (which did not differ by condition, F
because he was confused about the instructions, and 2 were excluded = .12). For the question assessing the extent to which partici-
because they were not American citizens and the essays were designed pants used the inventory, the mean response, which again did
specifically to criticize American conservatives or liberals. not differ by condition, was 15.6 (F = 1.21). As in Study 1,
Materials and procedure. Participants were randomly assigned to participants in all conditions were equally aware that the target
one of three conditions: an MS-evaluate-last condition, where partici- did not like spicy food and used this rating in deciding how
pants administered hot sauce prior to evaluating the worldview-threaten-
much hot sauce to allocate.
ing target; an MS-evaluate-first condition, in which participants ex-
pressed their attitudes toward the worldview-threatening target prior to MS and aggression. An ANOV\ on the weight (in grams)
administering the hot sauce; and a control condition, in which partici- of hot sauce allocated to the target yielded a significant main
pants were asked to write an essay about experiencing dental pain, and effect for condition, F ( 2 , 48) = 3.36, p < .05. Means are
administered hot sauce prior to evaluating the worldview-threatening displayed in Table 2. A planned orthogonal contrast to test the
target. A between-subjects design was used, and participants were run hypothesis that MS-evaluate-last participants would allocate
in groups of 4, consisting of either all liberal or all conservative partici- significantly more hot sauce to targets than participants in the
pants. The experimenter was blind to both participants' political stance other two conditions was significant, r(48) = 2.42, p < .02. A
and experimental conditions. second contrast, comparing MS-evaluate-first participants to
For the most part, the cover story and procedure were identical to dental pain participants, was not significant, /(48) = .91, p >
Study 1. Upon arrival, participants were directed toward cubicles and
.36. Thus, MS participants allocated significantly more hot sauce
presented with a packet of personality questionnaires, which contained
when they were not able to verbally derogate the targets prior
the MS manipulation. The topic of dental pain was used instead of an
important exam, so that thoughts of another aversive event could be to the administration of hot sauce. However, when MS partici-
used as a control condition. pants were able to first express their attitudes toward the target,
Next, participants were given the same bogus essay used in Study 1 the amount of hot sauce allocated was not significantly greater
to threaten their worldviews. However, unlike in Study 1, all participants than for controls.
read the worldview-threatening essay. Participants were then given the
first of two different impression formation forms. MS-evaluate-last
participants and control participants were presented with the nonevalua-
tive impression formation form used in Study 1, which allowed them to Table 2
evaluate the target without engaging in worldview defense. MS-evalu- Study 2: Cell Means for the Weight (in Grams), and Self-
ate-first participants were given an opportunity to evaluate the target in Report Estimates (in Spoonfuls) of Hot Sauce Allocated
a judgmental way, prior to allocating die hot sauce. Participants in this
condition were presented at this point with an evaluative impression Mortality Mortality
formation form (Greenberg et al., 1994), which included three questions salient salient Dental pain
regarding the target and two questions regarding agreement with the evaluate last evaluate first control
essay. Participants were asked, "How much do you like this person?,"
"How intelligent do you think this person is?," "How knowledgeable Measure M SD M SD M SD
do you think this person is?," "How much do you agree with this
Weight 16.98 13.70 11.50 8.69 8.44 4.78
person's opinion?," and "From your perspective, how true do you think
Self-report estimates 1.59 1.18 1.07 1.25 0.74 0.44
this person's opinion is?" Participants responded to these questions
598 MCGREGOR E T AL.

An ANOVA on participants' estimates of how many spoonfuls However, in the two MS conditions, although the correlations
of hot sauce they allocated to the target revealed a marginally between both weight and self-report estimates and the target
significant effect for condition, F(2,48) = 2.96, p < .07. Means ratings were in the negative direction, they did not approach
are displayed in Table 2. A planned orthogonal contrast again significance (ps > .20).
provided support for our main hypothesis: Estimates of hot Affect measure. One-way ANOV^s on each of the 13
sauce made by MS-evaluate-last participants were greater than PANAS-X subscales (administered immediately after the MS
estimates made by the other participants, t(4S) = 2.25, p < manipulation) revealed no effects of condition on affect {Fs <
.03. A second orthogonal contrast compared the MS-evaluate- .90). As in Study 1, we conducted within-cell correlations be-
first participants with dental pain participants and, as expected, tween each of the PANAS-X subscales and both the weight
was not significant, f(48) = .92, p > .36. and self-report measures of hot sauce allocation. In contrast to
Attitude toward and agreement with the target. As in previ- the findings of Study 1, the within-cell correlations between
ous research (e.g., Greenberg et al., 1994) separate composite hostility and weight were not significant, although they were
measures of attitudes toward the author and essay were com- positive (e.g., r( 15) = .30, in the MS-evaluate-last condition).
puted from the evaluative impression form. We predicted that
MS participants who evaluated the target prior to administering Discussion
hot sauce would evaluate the target less positively than other
participants. However; no difference was expected between MS Study 2 replicated the basic findings of Study 1: MS increased
participants who evaluated the target after administering hot aggression against the target when participants were not given
sauce and control participants. We had similar predictions re- an opportunity to derogate the worldview-threatening target
garding agreement with the essay, but we were less confident prior to aggressing. The findings also supported the hypothesis
for this measure because prior research has typically found that MS participants would engage in less aggressive behavior
stronger effects on evaluation of the target than on agreement toward a worldview-threatening target if they were first given
with the essay (indeed, the latter measure has sometimes failed an opportunity to derogate the target. Whereas MS-evaluate-
to yield significant effects, e.g., Greenberg et al., 1994). last participants allocated more hot sauce to targets than did
A one-way ANGVA on the attitude toward the target measure the other participants, when MS participants were able to first
yielded a marginally significant result, F(2, 47) — 2.75, p < derogate the target, they did not allocate significantly more hot
.08 (means are displayed in Table 3). An orthogonal contrast sauce than control participants.
offered support for the hypothesis that MS-evaluate-first parti- In addition, consistent with previous terror management re-
cipants would evaluate the target less positively than other parti- search (e.g., Greenberg et al., 1990), a planned contrast showed
cipants, t(47) = 2.26, p < .03. A second orthogonal contrast, that MS-evaluate-first participants derogated the worldview-
comparing MS-evaluate-last participants and dental pain parti- threatening target relative to the other participants. However,
cipants, was not significant t(47) = .50, p > .57. Thus, MS when MS participants were able to aggress against the target
led to more negative attitudes toward the worldview-threatening prior to this evaluation, there was no difference in liking for the
target, but only when these attitudes were assessed prior to target between MS and control participants. The general pattern
the opportunity to aggress. A one-way ANO\A on the essay of results suggests that aggression against the target and deroga-
agreement composite revealed no significant effect, F(2, 48) = tion of the target serve the same terror management function,
.23, p > .79 (means are displayed in Table 3). because once one mode of defense is used, the other is not.
Within-cell correlations for target evaluation and hoi sauce As noted earlier, other types of psychological defenses have
allocation. Within-cell correlations between attitude and be- also been shown to work in this manner. Tn general, evidence
havior measures were assessed. In the dental pain control condi- concerning dissonance reduction indicates that people will
tion, there was a significant negative correlation between weight choose whatever mode of dissonance reduction is most salient
of hot sauce allocation and target rating, r(14) = —.60, p < and available and that once they have used that mode, they will
.02, as well as between self-report of hot sauce allocation and not use alternative modes (e.g., Gotz-Marchand et al., 1974;
target rating, r(14) = - . 6 1 , p < .02, indicating that the more Scheier & Carver, 1980; Simon et al., 1995). For example,
negatively the target was rated, the more hot sauce was allocated. Simon et al. (1995) found that after writing a counterattitudinal
essay, people either shift their attitude to be more consistent with
the essay or reduce the importance of their action, depending on
Table 3 which is measured first or which is easier to do. Similar findings
Study 2: Cell Means on Composite Measures of Attitude have also been reported for defense of self-esteem (e.g.,
Toward and Agreement With Target's Essay Greenberg & Pyszczynski, 1985).
Given this reasoning, and the reciprocal between-condition
Mortality Mortality
relationship between target derogation and aggression in Study
salient salient Dental pain
evaluate last evaluate first control 2, one might have expected, in the MS conditions, a positive
correlation within-cells between the two measures such that the
Measure M SD M SD M SD more aggression, the less derogation. However, this runs counter
to the intuitive relationship between liking and aggression: The
Target rating 5.47 1.48 4.06 2.23 5.10 1.65
Agreement 3.06 1.79 2.88 2.35 3.35 1.89
less you like someone, the more you aggress against that person.
Indeed, in the control condition, in which mortality concerns
Note. Lower means indicate more negative evaluations of target. were not activated, there was a strong negative correlation be-
TERROR MANAGEMENT AND AGGRESSION 599

tween the two measures (i.e., the more aggression, the less liberal and conservative participants and having targets attack
derogation). In contrast, in the two MS conditions, the correla- one position or the other, we simply had American participants
tions were trivial and nonsignificant.5 Whereas we cannot defin- react to an anti-U.S. essayist, as has been done in past research
itively interpret these correlations, this pattern may have oc- (e.g., Greenberg et al., 1990). The only other difference was
curred because the tendency in the MS conditions to use one that we moved the measure of affect to after participants had
mode of defense or the other may have counteracted the more tasted their food sample to keep the procedure consistent with
general tendency to treat someone poorly to the extent you don't the conditions of Study 4. We expected that MS-evaluate-last
like that person. Consistent with this idea, Simon et al. (1995) participants would exhibit especially high hot sauce allocations
found similarly small nonsignificant within-cell correlations be- and that MS-evaluate-first participants would exhibit especially
tween modes of dissonance reduction. negative attitudes toward the anti-U.S. target.
Remaining issues. The results of Studies 1 and 2 clearly
show that MS increases aggression against worldview-threaten- Method
ing others. The findings of Study 2 also indicate that derogation
of and aggression against such a target serve the same function Participants. Seventeen male and 25 female students enrolled in psy-
in that once participants had engaged in one of the behaviors, chology courses at the University of Arizona participated in partial ful-
they did not engage in the other behavior. However, two issues fillment of a course requirement. RJQT participants were excluded from
the analyses because they reported suspicion about the experimental pro-
were not completely resolved. First, in Study 2, although the
cedure, two were excluded because of experimenter error, and three were
planned orthogonal contrasts provided clear support for the excluded because they were not Americans and the worldview-violating
hypotheses on both the hot sauce and evaluation measures, a essays were targeted specifically at American beliefs.
pairwise (two-tailed) comparison between the MS-evaluate- Materials and procedure. The design of Study 3 was identical to
first and MS-evaluate-last conditions would not have been sig- Study 2, with two exceptions. First, because participants were not se-
nificant (p = .11). Given the importance of the questions of lected on the basis of their political ideology, a different worldview-
whether derogation and aggression serve the same terror man- threatening essay was developed. Participants were asked to write about
agement function and, more generally, whether one form of MS- "the direction the country is heading in" and were later given an essay
motivated defense reduces the need for others, we decided to designed to derogate the United States:
replicate Study 2 in a third study.
That's an easy question to answer—I think this country basically
The second issue concerns whether the effects of MS are totally sucks—it's ugly, unfriendly, and out of control. There is no
specific to targets who threaten one's worldview or would occur real freedom here. It's easy to see why so many other countries
in reaction to any target who somehow impinged negatively on hate us. The only way things can change is if we have a new
the participants. According to terror management theory, MS government. And I don't mean a new president. Somebody should
should encourage negative reactions only to those who impinge get things together to overthrow the government—the president,
on one of the two components of the cultural anxiety buffer: congress and all that. And put in a new type of government that
self-esteem or, as in this research, the worldview. Prior research would be better. America thinks this is a great country but it is not.
suggests that MS effects on evaluations are limited to People who are happy wiui America are stupid.
worldview-threatening targets (e.g., Greenberg et al., 1990; Ro-
The other variation from Study 2 was the timing of the PANAS-X
senblatt et al., 1989). However, because the first two studies did measure. In Studies 1 and 2, die PANAS-X was administered immedi-
not include a noxious target who did not challenge participants' ately after the MS manipulation. However, in the present study it was
worldviews but would have been an appropriate target for ag- delivered immediately after participants consumed and rated their liking
gression, we cannot be certain that the MS effects on aggression of the food sample. This change was made because participants in Stud-
are limited to worldview-threatening targets, as terror manage- ies 3 and 4 were randomly assigned to conditions from the same general
ment theory predicts. sample and run in the same sessions, and in Study 4 the PANAS-X
Therefore, we designed Study 4 to determine whether MS was used as a manipulation check on affective reactions to the manipula-
would encourage aggression against a target who provoked par- tion of the food sample.
ticipants without threatening their worldviews. Unfortunately,
the terror management prediction is a null result: no effect of Results
MS. A potential problem arises, however: How would we then
Because Study 3 was a replication of Study 2, the primary
know that in Study 4, MS would affect aggression against any
analyses to be reported were again one-way (MS-evaluate-last
target? Our economical solution to this problem was to conduct
Studies 3 and 4 simultaneously, with participants randomly as- vs. MS-evaluate-first vs. dental pain) ANOVAs and planned
signed to one of seven conditions, from among the three condi-
tions of Study 3 and the four conditions of Study 4. If M S - 5
As an anonymous reviewer noted, from the perspective of dissonance
evaluate-last participants from Study 3 exhibited increased ag- theory, one might predict that more hot sauce allocation would lead to
gression, then an absence of an effect of MS on aggression more derogation and that more derogation would lead to more hot sauce
against the non-worldview-threatening target in Study 4 would allocation. However, cognitive consistency was apparently not the over-
support the idea that MS effects are indeed specific to riding concern in these studies (responding to the threat was). In addi-
worldview-threatening targets. tion, in this context, consistency between the dependent measures may
not have been a salient concern. After MS and worldview threat, aggres-
Study 3 sion may simply not need to be further justified by subsequent derogation
Study 3 is a replication of Study 2, with the same design and of the target. In addition, just because you don't like someone does not
dependent measures. However, in this study, rather than selecting necessarily mean that you will physically aggress against them.
600 MCGREGOR ET AL.

contrasts. As in the previous two studies, participants in all Table 5


conditions were equally aware that the target did not like spicy Study 3: Cell Means on Composite Measures of Attitude
food (Af = 4.00) and used this rating in deciding how much Toward Target and Agreement With Target's Essay
hot sauce to allocate (M = 14.07); neither varied by condition
(F < .40). Mortality Mortality
salient salient Dental pain
MS and aggression. An ANOV*V on the weight (in grams) evaluate last evaluate first control
of hot sauce allocated yielded a significant main effect for condi-
tion, f ( 2 , 39) = 3.56, p < .04. Means are displayed in Table Measure M SD M SD M SD
4. A planned orthogonal contrast to test the hypothesis that M S -
Target rating 4.81 2.35 3.83 1.65 5.29 1.37
evaluate-last participants would allocate more hot sauce than Agreement 4.46 2.74 3.11 1.94 5.29 2.16
participants in the other two groups was significant, f(39) =
2.65,p < .02. However, as expected, a contrast comparing M S - Note. Lower means indicate more negative evaluations of target.
evaluate-first participants to dental pain participants revealed
no hint of a difference, t(39) = .32, p > .15. Thus, the main
findings of Study 2 were replicated. Within-cell correlations for target evaluation and hot sauce
On the secondary measure of estimates of amount of hot allocation. Within-cell correlations were again conducted on
sauce allocated, an ANOVA yielded a marginally significant the attitude and behavior measures. However, none of the within-
effect for condition, F(2, 39) = 3.06, p < .06. Means are cell correlations approached significance (ps > .23).
displayed in Table 4. Planned orthogonal contrasts supported Affect measures. One-way ANOVAs performed on each of
our hypothesis. Estimates of hot sauce made by MS-evaluate- the 13 PANAS-X subscales revealed a significant effect on
last participants were significantly greater than estimates made self-report ratings of attentiveness, F(2, 39) = 8.19, p < .01.
by the other participants, *(39) = 2.47, p < .02, and there The means were 3.46, 3.20, and 2.54, respectively, for M S -
was no difference between MS-evaluate-first and dental pain evaluate-last, MS-evaluate-first, and dental pain conditions.
participants, r(39) = .16, p > .87. This effect suggests that MS participants were more attentive
Attitude toward and agreement with the target. The same than dental pain participants. There was also a significant effect
two composite measures of evaluation of the target and agree- on overall positive mood, F(2, 38) = 3.65, p < .04, which
ment with the target were computed. A one-way ANOVA per- reflects the fact that MS participants reported more positive
formed on the evaluation measure yielded a nonsignificant trend moods than dental pain participants. The means were 2.92, 2.65,
toward an overall effect, F{2, 39) = 2.27, p < .12. Means are and 2.21, respectively, for MS-evaluate-last, MS-evaluate-first,
displayed in Table 5. An orthogonal contrast comparing M S - and dental pain conditions. No significant effects were found on
evaluate-first participants with both MS-evaluate-last and den- the remaining 11 subscales. Of course, the finding of increased
tal pain participants yielded a marginally significant effect, positive mood in response to MS is paradoxical and may reflect
r(39) = 2.02, p < .052. As expected, the contrast comparing a tendency to deny one's distress at thoughts of one's own
MS-evaluate-last and dental pain participants did not approach death; Harmon-Jones, Greenberg, Solomon, and Simon (1995)
significance, r(39) = .69, p > .49. and Rosenblatt et al. (1989) also found some evidence of an
We expected MS-evaluate-first participants to exhibit sig- increase in positive mood in response to MS. Interpretation of
nificantly lower agreement with the essay than participants in these mood findings is complicated by the fact that in Studies
the other groups. In Study 2, the ANOVA and planned compari- 3 and 4, the PANAS-X was completed immediately after the
sons on this measure did not approach significance. However, in participants had tasted and evaluated the food sample; this pro-
the present study, the A N 0 \ A revealed a marginally significant vided a check on participants' affective responses to the drinks
effect, F ( 2 , 39) = 3.18,/? < .053, consistent with this hypothe- their partners assigned them in Study 4 and was included in
sis. Means are displayed in Table 5. When MS-evaluate-first Study 3 to make the procedures between the two studies as
participants were compared with the other two conditions, a parallel as possible. Thus, the effect of the treatment on these
significant difference was found, t(39) - 2.34,p < .03. Finally, affect scales may have been influenced by the food tasting expe-
the contrast comparing MS-evaluate-last and dental pain parti- rience. In addition, interpretation of the role of affect in reac-
cipants did not approach significance, f(39) = .94, p > .36. tions to targets is difficult because at this point in the experiment,
the evaluate-first participants already had had the opportunity
to derogate the target through verbal evaluations, but the evalu-
ate-last participants had not yet had the opportunity to aggress
Table 4
against the target.
Study 3: Cell Means for the Weight (in Grams), and Self-
Report Estimates (in Spoonfuls) of Hot Sauce Allocated As in Studies 1 and 2, within-cell correlations were conducted
between each of the PANAS-X subscales and both the weight
Mortality Mortality and self-report measures of hot sauce allocation. In contrast to
salient salient Dental pain
evaluate last
Study 1, the within-cell correlations between the hostility sub-
evaluate first control
scale and weight were not significant (e.g., r(14) = .002, in
Measure M SD M SD M SD the MS-evaluate-last condition). The only other correlation to
approach significance was between attentiveness and self-report
Weight 28.49 32.08 12.63 9.21 10.29 6.55 of hot sauce allocation in the MS-evaluate-first condition, r( 14)
Self-report estimates 3.09 3.74 1.29 1.09 1.15 0.90
= —.51, p = .06. Thus, the more attentive participants in the
TERROR MANAGEMENT AND AGGRESSION 601
MS-evaluate-first condition were, the less hot sauce they re- First, because the goal of the study was to determine whether MS
ported allocating. would influence reactions to a target who negatively affected participants
but did not threaten their worldview, we replaced the worldview-vio-
lating essay with a neutral essay that was designed to neither threaten
Discussion nor bolster participants' cultural worldviews. The topic of the essay
participants were asked to write was changed from * 'the direction the
The results of Study 3 generally replicated the findings of country is heading i n " to "recreational interests." Thus, all participants
Study 2. Once again, MS-evaluate-last participants allocated were presented with the following bogus essay:
the most hot sauce and MS-evaluate-first participants evaluated
the target most negatively. In this study, a pairwise comparison I think that recreation is very important I like lots of different
of hot sauce allocations between MS-evaluate-last participants activities. I like playing basketball, softball, and soccer. I also like
and MS-evaluate-first participants was significant, 2(39) = tossing a frisbee and swimming. Exercise is important to me so I
2.14, p < .04. However, the findings regarding derogation of use machines at the rec center pretty regularly. I also like going to
the target were a bit weaker than in Study 2. Although the the movies, reading, and hiking. And music, especially live music.
So there are many things I'm into, but I'm not really a fanatic about
planned orthogonal contrasts were supportive in both studies, a
any one kind of recreational activity.
pairwise comparison of hot sauce allocation was not significant
in Study 2, and the overall F for attitude was not significant in Next, we manipulated whether the target provided participants with
Study 3. However, when the data from Studies 2 and 3 were a negative experience. In the first two studies, participants were told
combined, all ANO\S\s were significant, as were all predicted that we were investigating the effects of personality on taste preferences
differences, whether assessed by planned contrasts or pairwise for dry and spicy foods. However, in Studies 3 and 4, we informed
comparisons (all ps < .05). Thus, overall, these two studies participants that we were also interested in tart foods. Instead of the
replicated the findings of Study 1 and also suggest that deroga- cracker sample, participants in Study 4 received a sample of juice.
tion and aggression are two functionally equivalent responses They were told that the juice sample was selected for them by another
participant, and that they ' 'selected this flavor from a variety of flavors
to worldview threat in response to MS.
which ranged from neutral to very tart in taste. So, they could have
given you any one of a number of flavors to taste, and this is the one
Study 4 they chose." All participants were then given a container of a purple
"juice." The taste of the juice was manipulated to be either very sour
As noted earlier, our final study was designed to address the and distasteful or neutral and rather bland.6 Unpleasant tasting juice of
question of whether the effect of MS on aggression is specific this nature has been successfully used to create an unpleasant experience
to those who threaten one's worldview or whether this effect in past research (Harmon-Jones, Brehm, Greenberg, Simon, & Nelson,
would generalize to any appropriate noxious target person. A 1996). Participants were instructed to consume the entire amount of
juice and then rate the taste of the juice on a 9-point scale with end
considerable body of research on aggression demonstrates that
points of neutral (1) and very tart (9) and to rate their liking of the
people are likely to aggress against others who deliver a noxious
juice on a 9-point scale which ranged from no liking at all (I) to extreme
stimulus to them (see, e.g., Geen, 1990, for an overview). There- liking (9).
fore, in Study 4, we had MS and control participants allocate hot Next, the PANAS - X was given to the participants. It was administered
sauce to someone who allocated either neutral or very unpleasant at this point in the experiment, rather than after the MS manipulation,
tasting juice to them. In general, we should have found increased because we wanted to use it to assess the affective impact of the juice
hot sauce allocations to the target who allocated the unpleasant manipulation. Because we were attempting to anger participants in a
juice and increased negative affect in those who drank the un- manner that did not violate their worldview, we expected the bad juice
pleasant juice. The key question, however, was whether MS groups to report higher ratings on the hostility and negative mood sub-
would intensify aggression against this target. If so, it would scales than the neutral juice groups.
suggest that MS encourages aggression against others who im- Because Studies 3 and 4 were run simultaneously within experimental
pinge negatively on participants independent of the implications session, it was necessary to ensure that the time it took participants to
complete the materials for both studies was equivalent. Consequently,
for the individuals' cultural worldview. If not, it would support
even though the recreation essay was used as a filler item in Study 4, and
the terror management view that MS encourages aggression only
we did not expect any group differences, we included the nonevaluative
against others who threaten one's worldview, just as it has been impression formation form after participants had read the essay arid the
found to encourage derogation of only such targets (e.g., evaluative impression formation form after participants had allocated
Greenberg et a l , 1990; Rosenblatt et al., 1989). the hot sauce and completed manipulation checks.
Finally, participants were debriefed. Because some participants in this
study drank the unpleasant tasting juice, we made sure that participants
Method understood why that manipulation was used. While participants reported
Participants. Sixteen male and 44 female psychology students par- that the juice tasted very unpleasant, no one reported being upset about
ticipated in partial fulfillment of a course requirement. Five participants our use of this manipulation or about any other aspect of our procedure.
were excluded because they reported suspicion about the experimental
procedure, two were excluded because of experimental error, and two
s
were excluded because they were nonnative speakers and had trouble The bad juice was a 2 oz (57 g) mixture of unsweetened grape
following the experimenter's instructions. Kool-Aid with 1 tsp of white vinegar added. The vinegar had the effect
Materials and procedure. Participants were randomly assigned to of making it extremely unpleasant to drink. The neutral juice consisted
conditions in a 2 (MS vs. control) X 2 (noxious vs. neutral juice) of 2 oz (57 g) water, with one drop of red food coloring and one drop
between-subjects design. The design of the study was identical to Study of blue food coloring added. The food coloring was used to give the
1, with the following exceptions. neutral sample an appearance identical to that of the bad juice.
602 MCGREGOR E T AL.

Results Table 6
Study 4: Cell Means for Weight (in Grams), and Self-Report
The primary analyses to be reported are 2 (MS vs. dental Estimates (in Spoonfuls) of Hot Sauce Allocated
pain) x 2 (neutral vs. bad juice) ANOVAs. As in Studies 1 and
2, participants were aware that the target disliked hot sauce (M Mortality salient Dental pain control
= 3.13) and found this information useful in allocating hot sauce Juice sample M SD M SD
amounts (M = 16.18); neither measure varied by condition
(Fs < 2.10). For the question assessing the extent to which Weight
participants used the inventory, the mean response was 15.97. 22.82 14.26 26.53 25.78
Bad
There was an unexpected main effect of juice sample, F ( l , 56) Neutral 17.07 9.39 13.64 15.74
= 4.59, p < .04. Participants in the bad juice condition reported
using the taste preference inventory less than participants in the Self-report estimates
neutral juice condition (Ms = 14.40 and 17.53, respectively). Bad 2.00 1.20 2.42 2.15
This may indicate that participants in the bad juice condition Neutral 1.25 0.80 1.24 1.34
were irritated by receiving their juice sample and as a result
were significantly less interested in taking the target's taste pref-
erences into account.
Juice sample evaluations. Participants also evaluated the mates of how many spoonfuls of hot sauce they allocated to the
flavor and taste of the juice samples. Not surprisingly, partici- target, an ANOVA again yielded a significant effect for sample
pants in the bad juice condition reported it to be significantly type, F ( l , 56) = 6.56, p < .02. Bad juice participants reported
more tart (M = 7.8) than did participants in the neutral juice more spoonfuls allocated than did neutral juice participants (Ms
condition (M = 2.0), F ( l , 56) = 130.32, p < .01. On the = 2.21 and 1.25, respectively). Cell means are displayed in
second question, which asked them to rate their level of liking Table 6. As with the weight measure, there were no hints of a
for the juice, the difference between participants in the neutral main effect of MS or an interaction of MS and sample type, F( 1,
juice (M = 2.0) and bad juice conditions (M = 1.7) was not 56) = .29, p > .59, and F ( l , 56) = .33,p > .57, respectively.
significant, F( 1,56) = 1.07, p > .43. The lack of significance Attitude toward and agreement with the target. As in the
on this measure is most likely a reflection of the specific wording previous studies, composite measures of evaluation of the essay-
of the question. Participants evaluated the juice on a scale that ist and agreement with the essay were computed and subjected
ranged from no liking at all (1) to extreme liking (9) with a to 2 X 2 ANOVAs. These measures were included to ensure
midpoint of moderate liking (5). The endpoint of no liking does procedural equivalence with Study 3. Because the participants
not permit participants to express extreme disliking they may were presented with a neutral essay, we had no strong expecta-
have felt for the sample. In addition, the midpoint of 5 reflects tions regarding this measure; however, we felt it possible that a
moderate likiag rather than a neutral evaluation that one would negative reaction to the target who delivered the noxious stimu-
expect from water. Consequently, it may have been impossible lus would be found.
for participants to demonstrate a distinction between reactions A two-way ANOVA performed on the attitude measure
of neutral feelings toward, and strong dislike for, the sample. yielded no significant effects, Fs < 1.06. Means are displayed
In other words, a floor effect may have kept a difference from in Table 7. However, there was a main effect for juice type on
emerging on this measure. This is particularly likely given the the agreement measure, F(\, 56) = 4.42,p < .05. Participants
significant differences found on participants' evaluation of the in the bad juice condition reported less agreement with the target
flavor of the juice and on the hostility subscale of the PANAS- (M = 5.9) than participants in the neutral juice condition (M
X measure (reported later in this section). In addition, in a = 6.8). This may indicate that anger at the target on the part
later study (Lieberman & Greenberg, 1997) the endpoint of no of participants in the bad juice condition carried over to the
liking at all was replaced with extreme disliking, and significant evaluative rating of agreement with the targets. No main effects
differences in the expected direction were obtained. or interactions involving MS were found (Fs < .14).
MS and aggression. As expected, an ANOVA on the weight Affect measure. Two-way ANOVAs performed on each of
(in grams) of hot sauce allocated to the target yielded a signifi- the 13 PANAS-X subscales revealed several significant effects.
cant main effect for sample type, F ( l , 56) = 4.33, p < .05. Significant main effects of sample type were found on hostility,
Bad juice participants allocated more hot sauce than neutral F ( l , 49) = 9.71, p < .01, overall negative mood, F ( l , 49) =
juice participants (Ms = 24.67 and 15.36, respectively). How- 5.62, p < .03, and surprise F( 1,49) = 7.64, p< .01. Bad juice
ever, there were no hints of a main effect of MS or an interaction participants reported feeling more hostile (Ms = 1.8 vs. 1.2),
of MS and sample type, F( 1,56) - .01, p > .97, a n d F ( l , 56) having a more negative mood (Ms = 1.48 vs. 1.18), and feeling
= .64, p > .42, respectively. Cell means are displayed in Table more surprised (Ms = 2.04 vs. 1.45) than neutral juice partici-
6. Thus, it appears that although our manipulation of sample pants. Thus, it appears that we were successful in our attempts
type was successful in increasing aggression, MS did not moder- to irritate our participants in a manner that did not violate their
ate this effect; in fact, in the bad juice condition, the mean hot worldview.
sauce allocation was higher in the dental pain condition than in An unexpected main effect of MS occurred on the surprise
the MS condition. Consequently, the effects of MS on aggression subscale as well, F ( l , 49) = 6.38, p < .02. MS participants
do not appear to extend to non-worldview-threatening targets. reported feeling less surprised in general (M = 1.5) than dental
On the secondary measure of participants' self-report esti- pain participants (M = 2.0). Although, as in Study 3, this effect
TERROR MANAGEMENT AND AGGRESSION 603

Table 7 have different views of the world. However, the findings of


Study 4: Cell Means on Composite Measures of Target Studies 2 and 3 indicate that, as Solomon et al. (1991) have
Rating, and Agreement with Target's Essay proposed, aggression is only one of a number of responses
to a worldview threat motivated by mortality concerns. When
Dental pain participants were first given the opportunity to aggress against
Mortality salient control
the target, they did so and did not express a particularly negative
Juice sample M SD M SD attitude toward the target. Similarly, when first given the oppor-
tunity to express dislike of the target, they did that and did not
Target rating engage in significant aggression. This pattern of results suggests
Bad 5.77 1.62 6.11 1.06 that derogation and aggression are two alternative ways to de-
Neutral 6.31 1.59 6.31 1.16 fend the worldview in response to the increased need for protec-
tion produced by MS.
Agreement
In addition to providing support for terror management the-
Bad 5.90 1.98 5.87 1.92 ory, these findings also tell us something new about terror man-
Neutral 6.67 1.67 6.97 1.19 agement that goes beyond the level of specification of the basic
Note. Lower means indicate more negative evaluations of target.
theory. The results indicate that MS can lead to at least two
different defensive reactions to a woridview threat and that once
one mode of addressing the threat is used, the other mode is
not needed. As previously noted, terror management may be
may have been influenced by the juice tasting that followed the similar in this regard to other psychological defenses such as
MS induction, it may indicate some denial of the affective im- dissonance reduction and self-esteem defense. It might be
pact of MS. worthwhile in future research to assess the possibility that if
As in the first 3 studies, within-cell correlations were con- the threat were particularly great or persistent, multiple modes
ducted between all PANAS-X subscales, including the hostility of defense might be used in concert.
subscale, and both the weight and self-report of hot sauce alloca-
Another issue worth exploring is whether people might use
tion measures. Again, the correlations involving the hostility
other modes besides derogation and aggression to defend their
subscale did not attain significance, although the relationships
worldviews in response to MS. Indeed, following Berger and
were in the expected direction in the MS conditions (e.g., be-
Luckmann (1967), Solomon et al. (1991) noted that, histori-
tween hostility and weight r(15) = .23, in the MS-bad juice
cally, in addition to derogation and aggression, people have also
condition). No other correlations between the remaining sub-
often tried to defuse worldview threat through assimilation and
scales and the weight or self-report measures approached sig-
accommodation. Assimilation involves attempting to convert
nificance. One might have expected some evidence that self-
others to one's own point of view. Missionary activity is one
reported hostility mediated aggression against the target; how-
example of this approach. Accommodation involves incorporat-
ever, none was found (Geen, 1990, noted that self-reported anger
ing certain compelling or appealing components of a threatening
and hostility and aggression often do not covary as well as
alternative worldview into one's own while discarding the
might be expected).
threatening components. A historical example of this may be the
incorporation and meaning-altering transformation of various
Discussion aspects of the 1960s American hippie counterculture into main-
stream American society. Hence, blue jeans became high fash-
The results of Study 4 indicate that MS does not increase
ion, granola bars became chocolate covered, and revolution-
aggression against an appropriate noxious target who did not
espousing rock music became soothing background stimulation
threaten participants' worldview—someone who made them
for shopping mall excursions. Of course, research is needed to
consume a very unpleasant tasting drink. This lack of effect
determine whether MS does indeed motivate efforts at assimila-
cannot be attributed to a general lack of effects of MS, because
tion and accommodation in response to worldview threat as well
randomized conditions run simultaneously (reported as Study
as derogation and aggression.
3) revealed that MS did increase aggression against a worldview
threatener. Thus, the evidence from these studies as well as We believe that it is also important to investigate factors that
prior research seems to support the specificity of MS effects determine which mode of defense is most likely to be used by
to worldview-relevant targets. These findings also add to prior a given person in a given situation. The present research suggests
research indicating that MS effects do not result from arousal that people will choose the first mode of worldview defense
or the generation of negative affect, for if they did, one would provided for them; however, there are undoubtedly also other
expect MS to amplify aggression against any appropriate target. situational and dispositional factors that affect which mode of
worldview defense is chosen. Indeed, the aggression literature
has revealed a variety of factors that facilitate or inhibit aggres-
General Discussion sion (see, e.g., Baron & Richardson, 1994; Geen, 1990; Ham-
Taken together, these four studies indicate that MS motivates mock & Richardson, 1992a); perhaps these would steer MS
aggression against those who threaten one's worldview. Thus, individuals toward either aggression or alternative responses to
these studies provide the first empirical evidence consistent with worldview threat. Of course, in the present research, participants
the idea that mortality concerns contribute to the historically all were not aware of alternative ways to respond to the threat when
too common phenomenon of aggression between those who the first dependent measure was assessed. By simultaneously
604 MCGREGOR E T A L .

providing people with a number of possible ways to address Becker, E. (1971). The hirth and death of meaning (2nd ed.). New
worldview threats following MS, future research may be able \brk: Free Press.
to clarify the roles that dispositional and situational variables Becker, E. (1973). The denial of death. New \brk; Free Press.
play in determining which of the many possible responses to Becker, E. (1975). Escape from evil New "fork: Free Press.
Berger, P., & Luckmann, T. (1967). The social construction of reality:
MS a given individual tends to choose.
A treatise in the sociology of knowledge. Garden City, NY: Anchor
Press.
Conclusion Berkowitz, L. (1964). Aggressive cues in aggressive behavior and hostil-
ity catharsis. Psychological Review, 71, 104-122.
The present studies are the first to provide behavioral evi-
Berkowitz, L. (1993). Aggression: Its causes, consequences, and con-
dence of aggression in response to MS. In so doing, they offer
trol. Philadelphia: Temple University Press.
what is probably the most compelling evidence to date that
Bernstein, S., Richardson, D., & Hammock, G. (1987). Convergent
mortality concerns play a significant role in social behavior; and discriminant validity of Buss and Taylor measures of physical
Clearly this work demonstrates that death-related thoughts can aggression. Aggressive Behavior, 13, 15-24.
motivate a highly significant and problematic type of social Brooks, S. (1992, May 14). State wants to close day care accused of
behavior. Indeed, the hot sauce allocations by MS participants mistreating children. The Columbus Dispatch, p. IB.
could have caused considerable pain not only to the mouths but Buss, A. H. (1961). The psychology of aggression. New York: Wiley.
also to the gastrointestinal systems of the intended targets. Columbus (Ohio) Dispatch. (1992, July 22). p. 2C.
These findings are thus consistent with Solomon et al.'s DeCharms, R., & Wilkins, E. J. (1963). Some effects of verbal expres-
(1991) contention that terror management processes contribute sion of hostility. Journal of Abnormal and Social Psychology, 66,
to aggression, as well as with the writings of Becker (1973, 462-470.
Diener, E. (1976). Effects of prior destructive behavior, anonymity,
1975) and others who have argued that the fear of death plays
and group presence on deindividuation and aggression. Journal of
a major role in intergroup aggression. Prom a practical perspec-
Personality and Social Psychology, 33, 497-507.
tive, further work is of course needed to assess the scope of
Diener, E., Dineen, J., Endresen, K., Beaman, A. L., & Fraser, S. C.
terror management processes in both interpersonal and in-
(1975). Effects of altered responsibility, cognitive set, and modeling
tergroup conflict, and to identify ways of attenuating the aggres- on physical aggression and deindividuation. Journal of Personality
sion that MS can produce. Perhaps methods to address the fear and Social Psychology, 31, 328-337.
of death could be developed individually or culturally that would Donahue, T. (1989, October 24). Woman denies killing son. United
reduce the propensity for people to resort to violence against Press International.
those who view the world differently than they do. As Becker Epstein, S. (1994). Integration of the cognitive and psychodynamic
(1975) noted: unconscious. American Psychologist, 49, 709-724.
Florian, Y, & Mikulincer, M. (1997). Fear of personal death and the
If men kill out of animal fears, then conceivably fears can always judgment of social transgressions: A multidimensional test of terror
be examined and calmed. . . . Elie Wiesel, who survived a Nazi management theory. Journal of Personality and Social Psychology,
concentration camp, summed it all up . . . "Man is not human." 73, 369-380.
But it is one thing to say that man is not human because he is a Geen, R. G. (1990). Human aggression—Mapping social psychology.
vicious animal and another to say that it is because he is a frightened Great Britain: Open University Press.
creature who tries to secure a victory over his limitations. . . . Gotz-Marchand, B., Gotz, J., & Irle, M. (1974). Preference of disso-
Cultural developments . . . might influence the fear of death and nance reduction modes as a function of their order, familiarity, and
forms of heroism; and so blunt the terrible destructiveness they reversibility. European Journal of Social Psychology, 4, 201-228.
have caused, (p. 169) Greenberg, J., & Pyszczynski, T. (1985). Compensatory self-inflation:
A response to the threat to self-regard of public failure. Journal of
Even before the first drop of blood is shed, many conflicts Personality and Social Psychology, 49, 273-280.
probably involve elevated mortality salience; this research sug- Greenberg, J., Pyszczynski, X, Solomon, S., Rosenblatt, A., Veeder, M.,
gests that such thoughts of mortality may be not only conse- Kirkland, S., & Lyon, D. (1990). Evidence for terror management
quences of these conflicts but significant contributors to their theory: II. The effects of mortality salience reactions to those who
violent escalation as well. threaten or bolster the cultural worldviews. Journal of Personality
and Social Psychology, 58, 308-318.
Greenberg, J., Pyszczynski, T., Solomon, S., Simon, L., & Breus, M.
References
(1994). The role of consciousness and accessibility of death-related
Arizona Daily Wildcat (University of Arizona). (1995, April 6). p. 1. thoughts in mortality salience effects. Journal of Personality and
Amdt, J., & Greenberg, J. (1996). Mortality salience, worldview de- Social Psychology, 67, 627-637.
fense, and physiological arousal. Manuscript in preparation, Univer- Greenberg, J., Simon, L., Harmon-Jones, E., Solomon, S., Pyszczynski,
sity of Arizona, Tucson. T, & Lyon, D. (1995). Testing alternative explanations for mortality
Amdt, J., Greenberg, J., Solomon, S., Pyszczynski, T., & Simon, L. salience effects: Terror management, value accessibility, or worrisome
(1997). Suppression, accessibility of death-related thoughts, and cul- thoughts. European Journal of Social Psychology, 25, 417-433.
tural worldview defense: Exploring the psychodynamics of terror man- Greenberg, J., Solomon, S., & Pyszczynski, T. (1997). Terror manage-
agement. Journal of Personality and Social Psychology, 73, 5 - 1 8 . ment theory of self-esteem and cultural worldviews: Empirical assess-
Baldwin, M. W., & Wesley, R. (1996). Effects of existential anxiety ments and conceptual refinements. In M. P. Zanna (Ed.), Advances
and self-esteem on the perception of others. Basic and Applied Social in experimental social psychology (Vol. 29, pp. 61-139). San Diego,
Psychology, 18, 75-95. CA: Academic Press.
Baron, R. A., & Richardson, D. R. (1994). Human aggression (2nd Greenberg, J., Solomon, S-, Pyszczynski, T., Rosenblatt, A., Burling, J.,
ed.). New "fork: Plenum Press. Lyon, D., & Simon, L. (1992). Assessing the terror management
TERROR MANAGEMENT AND AGGRESSION 605
analysis of self-esteem: Converging evidence of an anxiety-buffering (1989). Evidence for terror management theory: I. The effects of
function. Journal of Personality and Social Psychology, 63, 913— mortality salience on reactions to those who violate or uphold cultural
922. values. Journal of Personality and Social Psychology, 57, 681-690.
Hammock, G., & Richardson, D. (1992a). Aggression as one response Scheier, M. F., & Carver, C. S. (1980). Private and public self-attention,
to conflict. Journal of Applied Social Psychology, 22, 298-311. resistance to change, and dissonance reduction. Journal of Personality
Hammock, G., & Richardson, D. (1992b). Predictors of aggressive and Social Psychology, 39, 390-405.
behavior. Aggressive Behavior, 18, 219-229. Serrano, R. A. (1989, January 24). Two care providers plead not guilty
Harmon-Jones, E., Brehm, J. W., Greenberg, J., Simon, L., & Nelson, in child-abuse case. Los Angeles Times, Pt. 2, p. 1.
D. E. (1996). Evidence that the production of aversive consequences Shemberg, K. M., Leventhal, D. B., & Allman, L. (1968). Aggression
is not necessary to create cognitive dissonance. Journal of Personality machine performance and rated aggression. Journal of Experimental
and Social Psychology, 70, 5-16. Research in Personality, 3, 117-119.
Harmon-Jones, E., Greenberg, J., Solomon, S., & Simon, L. (1995). Simon, L., Greenberg, J., & Brehm, J. (1995). Trivialization: The forgot-
The effects of mortality-salience on intergroup bias between minimal ten mode of dissonance reduction. Journal of Personality and Social
groups. European Journal of Social Psychology, 25, 1-5. Psychology, 68, 247-260.
Harrington, A. (1969). The immortalist. New York: Random House. Simon, L., Greenberg, J., Harmon-Jones, E., Solomon, S., Pyszczynski,
Kirk, R. E. (1995). Experimental design: Procedures for the behavioral T., Arndt, J., & Abend, T. (1997). Terror management and cognitive-
sciences (3rd ed.). Pacific Grove, CA: Brooks/Cole. experiential self-theory: Evidence that terror management occurs in
Lieberman, J. D., & Greenberg, J. (1997). Cognitive-experiential self- the experiential system. Journal of Personality and Social Psychol-
theory and displaced aggression. Manuscript in preparation, Univer- ogy, 72, 1132-1146.
sity of Arizona, Tucson. Solomon, S., Greenberg, J., & Pyszczynski, T. (1991). A terror manage-
Lifton, R. J. (1983). The broken connection; On death and the continu- ment theory of social behavior: The psychological functions of self-
ity of life. New "York: Basic Books. esteem and cultural worldviews. In M. P. Zanna (Ed.), Advances in
Martin, N. S. (1994, December 26). Child killing a losing battle in S. experimental social psychology (Vol. 24, pp. 91-159). San Diego,
Florida. Sun-Sentinel, p. 1A. CA: Academic Press.
Milne, J. (1995, April 20). Some don't like it hot; Fry cook facing assault Taylor, S. P. (1967). Aggressive behavior and physiological arousal as
charges denies he spiked trooper's eggs. The Houston Chronicle, p. a function of provocation and the tendency to inhibit aggression.
A15. Journal of Personality, 35, 197-310.
Nelson, L. J., Moore, D. L., Olivetti, J., & Scott, T. (1997). General and Thirty-nine year old charged with child abuse (1992, January 24). St.
personal mortality salience and nationalistic bias. Personality and Petersburg Times, p. 1.
Social Psychology Bulletin, 23, 884-892. Watson, D., & Clark, L. A. (1991). The PANAS-X: Preliminary manual
N.Y. foster mother accused of torturing 3-year-old boy. (1993, August for the positive and negative affect schedule-expanded form. Unpub-
26). The Record (Bergen, New Jersey), p. A09. lished manuscript, Southern Methodist University.
Ochsmann, R., & Mathy, M. (1994). Deprecating and distancing from Wheeler, L., & Caggiula, A. R. (1966). The contagion of aggression.
foreigners: Effects of mortality salience. Unpublished manuscript, Journal of Experimental Social Psychology, 2, 1-10.
Universitat Mainz, Mainz, Germany. Zillmann, D. (1979). Hostility and aggression. Hillsdale, NJ: Erlbaum.
Ochsmann, R., &Reichelt, K. (1994). Evaluation of moral and immoral
behavior; Evidence for terror management theory. Unpublished manu- Received November 16, 1995
script, Universitat Mainz, Mainz, Germany. Revision received May 6, 1997
Rosenblatt, A., Greenberg, J., Solomon, S., Pyszczynski, T, & Lyon, D. Accepted May 9, 1997 •

You might also like