Article Critique

Journal of Experimental

Journal of Experimental Social Psychology 41 (2005) 61–67

www.elsevier.com/locate/jesp

Social Psychology

Journal of Experimental Social Psychology 41 (2005) 61–67

The moderating effect of trigger intensity on triggered displaced aggression

q

Eduardo A. Vasquez, a,* Thomas F. Denson,

a William C. Pedersen,

b

Douglas M. Stenstrom, a and Norman Miller

a

a University of Southern California, Seeley G. Mudd Building, Room 501, Los Angeles, CA 90089-106, USA

b Department of Psychology, California State University, Long Beach, USA

Received 4 September 2003; revised 7 April 2004

Available online 2 July 2004

Abstract

Many instances of aggression result in excessive retaliation in response to a seemingly trivial triggering event. The triggered

displaced aggression paradigm (TDA; Miller, Pedersen, Earleywine, & Pollock, 2003) provides an experimental vehicle for exploring

such occurrences. Participants were either provoked or not and were subsequently exposed to a neutral, mild, or moderately strong

triggering event from a second bogus participant. Consistent with TDA theory (Miller et al., 2003), disjunctively escalated aggressive

behavior occurred only among previously provoked participants when responding to the mild triggering event, but not the mod-

erately strong or neutral trigger. Independent of provocation, the neutral triggering event elicited very low levels of aggression,

whereas the moderately strong trigger elicited moderate levels of aggression. Implications for instances of real world aggression are

discussed.

� 2004 Elsevier Inc. All rights reserved.

Keywords: Aggression; Displaced aggression; Triggered displaced aggression; Provocation

Introduction

Displaced aggression occurs when a person is pro-

voked, is prevented from retaliating against the original

provocateur, and subsequently aggresses against a

seemingly innocent target (Dollard, Doob, Miller,

Mowrer, & Sears, 1939; Hovland & Sears, 1940; Marcus-

Newhall, Pedersen, Carlson, & Miller, 2000). For in-

stance, a man insults his wife for no apparent reason after having been berated previously by his boss. In this

case, the target has provided no justification or instiga-

tion to warrant a retaliatory response from the aggressor.

q The current research was supported in part by Grant #R21-

AA013343 from NIAAA and John Randolph Haynes and Dora

Haynes Foundation fellowships to Eduardo A. Vasquez and Thomas

F. Denson. * Corresponding author. Fax: 1-213-746-9082.

E-mail address: evasquez@usc.edu (E.A. Vasquez).

0022-1031/$ - see front matter � 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jesp.2004.05.007

Of greater theoretical and ecological interest is trig- gered displaced aggression (TDA; Miller et al., 2003;

Pedersen, Gonzales, & Miller, 2000). In the TDA par-

adigm, participants are exposed to an initial Time 1

provocation under conditions that preclude retaliation

against the provocateur. At Time 2, a second, usually

trivial and ambiguous triggering event is presented as an

instigation to aggress. Aggression directed toward the

source of this Time 2 triggering event can disjunctively exceed the independent additive effects of the Time 1

provocation and Time 2 trigger (Pedersen et al., 2000).

For instance, the same man who is berated by his boss

and later severely physically abuses his wife in response

to her query about why he did not mow the lawn an-

ecdotally illustrates the disjunctive escalation of ag-

gression that can be seen in triggered displaced

aggression. Recently, Miller et al. (2003) have suggested that the

intensity of the Time 2 trigger is of theoretical impor-

tance. Specifically, compared to participants exposed to

62 E.A. Vasquez et al. / Journal of Experimental Social Psychology 41 (2005) 61–67

a moderately strong trigger or no trigger at all, only previously provoked participants exposed to a relatively

mild triggering event should display disjunctively esca-

lated displaced aggression. There are two reasons for

this. First, consistent with the cognitive neoassocia-

tionistic model of aggression (CNA; Berkowitz, 1993)

and the more recent general aggression model (GAM;

Anderson & Bushman, 2002), provocation primes ag-

gression related cognition, affect, and arousal, such that negative features of subsequent events are likely to be

made highly salient to provoked individuals relative to

unprovoked individuals. Indeed, participants simply

primed with aggressive constructs interpret ambiguous

situations in a more aggressive manner than control

participants (for a review see Todorov & Bargh, 2002).

Second, mild triggering events are susceptible to attri-

butional distortion whereas strong triggers are always highly salient and unambiguously perceived as provo-

cations in and of themselves. Thus, strong triggering

events lend themselves to the tit-for-tat ‘‘matching rule’’

(Axelrod, 1984) and norms of reciprocity (Gouldner,

1960) whereas mild triggers are ambiguous and leave

room for biased interpretation as a result of prior

provocation.

To date, only five published aggression studies have orthogonally manipulated both a Time 1 provocation

and Time 2 triggering event (Baron, 1972; Baron & Bell,

1975; Pedersen et al., 2000, Studies 1 & 2; Worchel,

1966). Although the three earlier studies failed to find

disjunctive escalation (Baron, 1972; Baron & Bell, 1975;

Worchel, 1966), two more recent studies did find the

expected interaction (Pedersen et al., 2000; studies 1 and

2). In the first three studies the intensity of the Time 2 trigger matched or exceeded the intensity of the Time 1

provocation. For instance, in Worchel (1966) students in

a psychology course were told by a graduate teaching

assistant that the entire class would be subjected to a

pop quiz (Time 1 provocation). Subsequently, they were

interrupted and insulted by the course instructor as they

completed a bogus intelligence test (Time 2 triggering

event). Clearly, the Time 2 triggering event was not trivial. More likely is that it was of similar if not greater

intensity than the Time 1 provocation. Although a slight

increase in aggression did occur relative to the Time 1

provocation only condition, not even additive effects

were obtained.

In two studies by Baron (1972) and Baron and Bell

(1975) the Time 1 provocation entailed spending the

experimental session in a very hot room (91.1–95.5 �F). In both studies, the Time 2 triggering events were nei-

ther mild nor ambiguous. In the Baron (1972) study,

triggered participants received nine electric shocks and a

negative task evaluation. In Baron and Bell (1975),

participants were triggered by an experimental confed-

erate who, in a written statement, insulted the partici-

pant. Interestingly, in both studies participants who

were in both the hot room and trigger conditions, re- acted with decreased aggression compared to partici-

pants who received the trigger in a cooler (74–75 �) room.

Pedersen et al. (2000), however, employed a mild

Time 2 triggering event. In these two studies, the pre-

dicted interaction between the Time 1 provocation and

Time 2 trigger was observed such that previously pro-

voked participants exposed to the mild trigger displayed greater disjunctively escalated displaced aggression than

participants who were not provoked, exceeding the in-

dependent additive effects of the Time 1 provocation

and Time 2 trigger. In Study 1, the experimenter insulted

participants on their (poor) performance on a difficult

anagram task (Time 1 provocation) or not (no provo-

cation control condition). Participants then performed a

second trivia game task that was presented to them by either an annoying and incompetent research assistant

(Time 2 trigger) or a competent research assistant (no

trigger control condition). Participants were then given

the opportunity to evaluate the research assistant. Par-

ticipants evaluated the research assistant negatively only

when they had previously been provoked. In Study 2, an

experimenter provoked participants by telling them to

speak louder in a contemptuous and irritated tone of voice during the same difficult anagram task used in

Study 1. A second (bogus) participant then informed the

participants in writing that their performance on the

anagram task could have been ‘‘somewhat stronger’’

(Time 2 trigger) or were given a neutral evaluation (no

trigger control). Participants were then asked to evaluate

the bogus participant for a coveted research position

job. The same pattern of results as observed in Study 1 were obtained. Thus, despite the distinctly different ex-

perimental manipulations of triggering events in these

two studies, participants exhibited disjunctively esca-

lated aggression in the presence of a mild triggering

event only when previously provoked. The mild trigger

by itself did not affect displaced aggression.

The current research

The between-study differences in outcomes that we

have described suggest that within the TDA paradigm,

mild triggers are more likely to elicit disjunctive escala-

tion of aggression than are strong triggering events.

Thus, in the research we report herein, our primary

purpose was to investigate the effects of differences in

intensity of triggering events on displaced aggression within a single study. Half of the participants were ex-

posed to a Time 1 provocation and half were not pro-

voked. Participants were then exposed to either a

moderately strong trigger, mild trigger, or neutral trig-

ger. We expected differences in trigger intensity to

moderate the interaction between the Time 1 provoca-

tion and Time 2 triggering event, yielding a disjunctive

E.A. Vasquez et al. / Journal of Experimental Social Psychology 41 (2005) 61–67 63

escalation of aggression only among previously pro- voked participants who were exposed to the mild trig-

ger, by contrast with those exposed to either a

moderately strong or neutral trigger. Due to inconsis-

tencies in previous studies with both strong Time 1 and

Time 2 events (Baron, 1972; Baron & Bell, 1975; Wor-

chel, 1966), no predictions were made regarding the

strong trigger conditions, other than an expectation of

no disjunctive escalation of aggression. Our secondary goal was to provide convergent validity for the TDA

paradigm by using a measure of aggression that reflected

an act of physical harm, rather than the verbal aggres-

sion measured in Pedersen et al. (2000). Previous reviews

have concluded that verbal and physical laboratory

aggression measures assess the same latent construct of

aggression (Anderson & Bushman, 1997; Giancola &

Chermack, 1998). Therefore, we anticipated our out- come with a physical measure of aggression to parallel

that previously obtained with verbal measures (viz.

Pedersen et al., 2000).

Method

Participants and design

Participants were 63 University of Southern Califor-

nia (USC) undergraduates, 44 women and 19 men. All

participants were part of the psychology department

subject pool and volunteered for extra course credit. The

design was a 2 (provocation: yes or no) � 3 (trigger: moderately strong, mild, neutral control) between sub-

jects factorial. Participants were randomly assigned to one of the six experimental conditions within blocks

such that one full replication of the design was com-

pleted before starting a new block.

Procedures

Participants were run individually. Upon arrival to

the laboratory, participants were told that the study was investigating the effects of cognitive ability and physical

distraction on social impression formation. Participants

were told that they would complete a measure of general

cognitive ability while being distracted and then interact

with another (bogus) participant in another room on the

floor. In order to reduce suspicion, the experimenter

ostensibly checked to make sure the other participant

was ready to begin the experiment while the actual participant completed a demographic information sheet.

Approximately 3 min later, the experimenter returned

to the laboratory room and told the participant that the

first part of the study involved a test of cognitive ability

under conditions of auditory distraction. Specifically,

participants listened to cacophonous music (Stravin-

sky’s ‘‘Rites of Spring’’) at a moderately loud volume

(approximately 80 dB) while completing a sheet with 15 difficult anagrams (e.g., tophhapogr ¼ photograph). The experimenter informed the participant that they would

have 4 min to complete all 15 anagrams, started the

music, and left the room. When the 4 min had elapsed,

the experimenter re-entered, took the anagram answer

sheet, and left the room to ostensibly score the partici-

pant’s performance.

Provocation manipulation

Approximately 3 min later, the experimenter re-en-

tered. In the provocation condition, participants were

told that their performance was far below average

compared to a sample of engineering students. Fur-

thermore, the experimenter insulted participants in an

irritated and exasperated tone of voice:

You really got a lot of these wrong. This data is useless to me.

We should probably just start all over, but to be perfectly hon-

est with you, I don’t want to waste my time.

In the no provocation condition, participants were

told that their performance was average compared to a

sample of engineering students and were not insulted.

Trigger manipulation

Participants then completed a trait-listing task. Spe-

cifically, they were asked to list six traits necessary to be

a good astronaut (Bettencourt, Brewer, Croak, & Miller,

1992). The experimenter then appeared, ostensibly to

take the participant’s astronaut task performance to the

other participant. Two minutes later, the experimenter

returned with the bogus astronaut task and an evalua- tion form for the participant to fill out. Allegedly, it

would be exchanged with the other participant. Similar

to procedures used in prior research (Pedersen et al.,

2000; Study 2), this exchange of evaluation forms served

as the Time 2 trigger manipulation.

To implement the trigger conditions, participants

received from the other participant an evaluation of the

degree to which their astronaut task exhibited original- ity, quality, effort, a variety among traits listed, and

made sense. In addition, an overall evaluation was

provided. In the mild trigger condition the individual

ratings and overall evaluation were 3, 4, 3, 3, 4, and 4,

respectively on 7-point Likert-type scales (1 ¼ no good at all, 7 ¼ extremely good). In addition, space was available for participants to indicate additional comments. In this

space, the following statement was written: ‘‘The per- formance was not great and I think a USC student could

do better.’’ In the moderately strong trigger condition,

the evaluation consisted of ratings of 3, 2, 1, 1, 3, and 2

and the statement: ‘‘Personally, I think the performance

was bad. People who come to USC should be more

aware of our space program. This needed more work.’’

In the neutral trigger condition, the participant received

a neutral evaluation (6, 5, 6, 5, 5, and 5) and the

1 All robust statistical analyses were completed using functions

written by Rand Wilcox (available at http://www-rcf.usc.edu/~rwilcox)

for the computer program R (available at http://www.r-project.org).

All analyses were also conducted using traditional ANOVA methods

on untrimmed means. Both sets of analyses revealed an identical

pattern of results. 2 Student’s t test on untrimmed means also revealed a significant

mean difference: tð56Þ ¼ 2:88, p < :01.

64 E.A. Vasquez et al. / Journal of Experimental Social Psychology 41 (2005) 61–67

following statement: ‘‘My partner did a decent job. I think the task was well done.’’

Physical aggression

Participants were given approximately 3 min to look

over their evaluation. The experimenter returned with a

cup containing four pieces of paper and explained that

the next portion of the study would investigate the ef-

fects of physical distraction on impression formation. The participant then drew one of the four pieces of

paper to determine which distraction condition they

would be in (visual, auditory, tactile, or control). In

reality, all four pieces of paper contained the no dis-

traction control condition. The experimenter then left

the room ostensibly to see which distraction condition

the other (bogus) participant received.

Approximately 2 min later the experimenter returned with a bucket of cold water (10 �C) and the dependent measures. He informed the participant that the other

participant had received the tactile distraction condition.

The participant was told that s/he would determine how

long the other (bogus) participant would submerge one

hand in the bucket of cold water while performing an

impression formation task. Participants were instructed

to place one hand in the bucket of water for 5 s, osten- sibly because such experience was necessary to make an

informed decision about the length of time that their

partner would be distracted. Pretesting had determined

that 5 s of exposure to the cold water was perceived as

painful. Next, participants were instructed to circle the

amount of time that the other participant should be

distracted on a 9-point Likert-type scale starting at

‘‘1 ¼ no distraction at all’’ which increased by 10 s inter- vals to ‘‘9 ¼ 80 s/very strong distraction.’’ Participants were asked to slide the aggression sheet under the door so

that a second research assistant could administer the task

to the bogus participant. Participants then completed the

remaining dependent measures at their own pace.

Manipulation checks

In order to assess affect from the provocation, par- ticipants completed a modified version of mood adjec-

tive checklist (MACL; Nowlis, 1965). After having been

told how they did on the anagram task (i.e., the prov-

ocation), participants rated the degree to which they

experienced each of 25 emotions. Each emotional de-

scriptor was rated on a 7-point Likert-type scale (1 ¼ not at all, 7 ¼ extremely so).

An additional six items assessed the reaction to the trigger. Specifically, participants were asked to rate how

happy, complimented, pleased, annoyed, irritated, and

angry they felt upon receiving the evaluation from the

bogus participant. Each item was rated on a 7-point

Likert-type scale (1 ¼ not at all, 7 ¼ extremely so). Six additional questions assessed attributional dis-

tortion. Participants were asked to rate the degree to

which their partner provided a good, unfair, useless, and unbiased evaluation. In addition participants also rated

the degree to which their partner meant to provide a

negative evaluation and a valuable evaluation. All items

were rated on 9-point Likert-type scales (1 ¼ strongly disagree, 9 ¼ strongly agree).

All participants were thanked, probed for suspicion

with a funnel debriefing, and dismissed.

Results

Statistical analyses

Data were analyzed with modern robust statistical

methods. 1 Even arbitrarily small violations of the as-

sumptions of normality or homogeneity of variance may greatly decrease the sensitivity of traditional analysis of

variance methods and produce biased results (Wilcox,

1998, 2003). In experimental research, truly normal

distributions are rarely, if ever, observed in practice.

Recently, Wilcox and Keselman (2003) reviewed a sub-

stantial body of evidence suggesting that traditional

methods of inferential statistics based on means perform

poorly under most circumstances encountered in psy- chological research (e.g., heavy tails, slight skewness,

heteroscedasticity, etc.). These authors demonstrated

that bootstrap methods and analyses with trimmed

means provide superior performance relative to tradi-

tional procedures. Specifically, modern methods accu-

rately control Type I error rate, provide increased

power, and tolerate violations of the homogeneity and

normality assumptions. Four participants were removed due to suspicion of

experimental procedures or guessing the study hypoth-

esis. No significant gender effects were observed. Thus,

males and females were combined into the final analyses.

Manipulation checks

Mood adjectives from the mood adjective checklist were formed into reliable composites for positive

(a ¼ :81) and negative (a ¼ :92) affect in order to assess emotional reactions to the provocation. Yuen (1974)

robust t test on 20% trimmed means demonstrated that participants in the provocation conditions reported

experiencing more negative affect than participants in

the no provocation conditions, Ty ¼ 2:74, p < :001; Mts ¼ 3:10 and 2.32, respectively.2 No differences

Fig. 1. Physical aggression means and standard errors as a function of

Time 1 provocation and Time 2 trigger intensity.

E.A. Vasquez et al. / Journal of Experimental Social Psychology 41 (2005) 61–67 65

emerged for positive affect. Thus, it appears that the provocation manipulation was successful.

To assess affective reaction to the trigger, composites

were formed from the items assessing their emotional

reaction to the evaluation they received. Both the posi-

tive items (happy, complimented, pleased) and negative

items (annoyed, irritated, offended) formed reliable

composites, as ¼ :96 and .94, respectively. A one-way ANOVA on 20% trimmed means across trigger condi- tions revealed a main effect for negative affect,

Ftð2; 20:72Þ ¼ 86:87, p < :001. Percentile bootstrap lin- ear contrasts on 20% trimmed means (Wilcox, 2003)

revealed that participants in the mild trigger condition

reported more negative affect as a result of the evalua-

tion than participants in the neutral trigger condition,

w ¼ 6:95, p < :0001. In addition, participants in the moderately strong trigger condition reported more negative affect than participants in the neutral trigger

and mild trigger conditions, w ¼ 3:80 and w ¼ 12:60, respectively, ps < :0001.3 There was also a main effect of trigger condition on positive affect, Ftð2; 21:51Þ ¼ 71:94, p < :001. Percentile bootstrap linear contrasts on 20% trimmed means revealed that participants in the neutral

trigger condition reported more positive affect that

participants in the mild and moderately strong trigger conditions, w ¼ 9:54 and w ¼ 12:15, respectively, ps < :0001. These two latter conditions did not differ from each other in amount of reported positive affect,

w ¼ 1:81, ns.4

Physical aggression

A 2 (provocation: yes, no) � 3 (trigger: mild, moder- ately strong, neutral) between subjects ANOVA with

20% trimmed means (Wilcox, 2003) revealed a main

effect of Trigger, Ft ¼ 6:17, p ¼ :02. However, this main effect was qualified by a Trigger � Provocation interac- tion, Ft ¼ 20:52, p < :05. Planned percentile bootstrap linear contrasts with 20% trimmed means were con-

ducted. As expected, the only differences between the

provocation and no provocation conditions were in the mild trigger conditions such that previously provoked

3 A traditional one-way between subjects ANOVA and post hoc

tests on untrimmed means of negative affect also revealed an identical

pattern of results. There was a main effect for trigger, F ð2; 55Þ ¼ 39:71, p < :001. Participants in the mild trigger condition reported more negative affect than participants in the neutral trigger condition,

p < :001, while participants in the moderately strong trigger condition reported more negative affect than the mild (p < :005) and neutral trigger (p < :001) conditions.

4 A traditional one-way between subjects ANOVA and post hoc

tests on untrimmed means of positive affect also revealed an identical

pattern of results. There was a main effect for trigger, F ð2; 55Þ ¼ 60:66, p < :001. Participants in the neutral trigger condition reported more positive affect than participants in the mild and moderately strong

trigger conditions, ps < :001. Means in the mild and moderately strong trigger conditions did not differ from each other, p ¼ :40.

participants (Mt ¼ 6:57) displayed disjunctively higher levels of aggression than unprovoked participants

(Mt ¼ 3:83; w ¼ 3:79, p ¼ :005).5 Aggression did not reliably differ in the neutral trigger and moderately

strong trigger conditions as a function of provocation.

Fig. 1 displays the raw means and standard errors.

Attributional distortion

The six items designed to assess attributional distor-

tion formed a reliable composite (a ¼ :86). A 2 (provo- cation: yes, no) � 3 (trigger: mild, moderately strong, neutral) between subjects ANOVA with 20% trimmed

means revealed a main effect for trigger condition,

Ft ¼ 68:28, p < :0001. Percentile bootstrap linear con- trasts on 20% trimmed means revealed that participants

in the mild trigger condition perceived the evaluation

more negatively than participants in the neutral trigger

condition, w ¼ 6:51, p < :0001. In turn, participants in the moderately strong trigger condition perceived the

evaluation more negatively than participants in both the

mild, w ¼ 3:01, p < :002, and neutral trigger conditions, w ¼ 12:03, p < :0001. Although the expected interaction between the provocation and mild trigger did not occur,

the attributional distortion composite did reliably

5 A traditional 2 (provocation: yes, no) � 3 (trigger: mild, moder-

ately strong, neutral) between subjects ANOVA on untrimmed means

also revealed a main effect of trigger condition, F ð2; 53Þ ¼ 11:65, p < :001, and a trigger � provocation interaction, F ð2; 53Þ ¼ 4:24, p < :05. Planned contrasts on untrimmed means revealed an identical pattern of results. Within each trigger condition, the only significant

difference was between provoked and unprovoked participants in the

mild trigger condition, tð53Þ ¼ 2:76, p < :01.

66 E.A. Vasquez et al. / Journal of Experimental Social Psychology 41 (2005) 61–67

predict aggression, Theil-Sen estimate (Wilcox, 2003) bts ¼ :37, p < :0001.6

Mediation analyses

To explore the hypothesis that negative affect result-

ing from the Time 2 triggering event (but not the Time 1

provocation) elicits aggression toward the target, medi-

ation analyses (Baron & Kenny, 1986) were conducted. Trigger condition predicted physical aggression, Theil-

Sen bts ¼ 1:50, p < :0001. Trigger condition also reliably predicted negative affect, such that increasing trigger

intensity resulted in higher levels of negative affect,

Theil-Sen bts ¼ 1:84, p < :0001. Negative affect from the trigger predicted aggression, Theil-Sen bts ¼ :75, p < :0001. Finally, when controlling for negative affect, trigger condition no longer predicted aggression, Theil- Sen bts ¼ :40, p ¼ :46, but negative affect remained a significant predictor of aggression, Theil-Sen bts ¼ :60, p < :0001. Thus, the four criteria suggested by Baron and Kenny (1986) for establishing mediation were ful-

filled. Finally, although negative affect resulting from

the trigger predicted physical aggression, negative

affect resulting from the provocation did not predict

aggression. 7

Discussion

The current study provided support for Miller et al.

(2003) prediction that relative to a no provocation

condition, disjunctively escalated displaced aggression

would only occur when the Time 2 triggering event is of mild intensity. Indeed, the only observed difference in

aggression was in the mild trigger conditions, such that

previously provoked participants displayed displaced

aggression that exceeded the independent additive effects

of the Time 1 provocation and Time 2 triggering event.

6 The pattern of results for the attributional distortion composite

with traditional statistics was identical to the results obtained with

robust methods. The 2 (provocation: yes, no) � 3 (trigger: mild, moderately strong, neutral) between subjects ANOVA on untrimmed

means also revealed a main effect of trigger condition,

F ð2; 52Þ ¼ 30:96, p < :001. Post hoc testing revealed that participants in the mild trigger condition perceived the evaluation more negatively

than participants in the neutral trigger condition, p < :001. Partici- pants in the moderately strong trigger condition perceived the

evaluation more negatively than participants in both the mild

(p < :01) and neutral trigger conditions (p < :001). Ordinary least squares (OLS) regression analyses demonstrated that the attributional

distortion composite predicted aggression, b ¼ :43, p < :01. 7 Mediation analyses with OLS revealed an identical pattern of

results. Trigger condition predicted both physical aggression, b ¼ :48, p < :001, and negative affect, b ¼ :76, p < :001. In turn, negative affect from the trigger predicted aggression, b ¼ :55, p < :001. Finally, when controlling for negative affect, trigger condition no longer predicted

aggression, b ¼ :14, p ¼ :43, but negative affect remained a significant predictor of aggression, b ¼ :44, p < :05.

Participants in the neutral trigger and moderately strong trigger conditions did not significantly differ in levels of

displaced aggression. Thus, the current study provides

strong support for a central feature of the TDA para-

digm. Specifically, among those previously provoked,

aggressive responding should be greatest in response to

a mild triggering event. When not previously provoked

participants should not overreact to the mild triggering

event. This was the pattern of results obtained in this and two previous studies (Pedersen et al., 2000). In ad-

dition, evidence of convergent construct validity was

obtained by extending the TDA paradigm to measures

of physical aggression. By contrast, only measures of

verbal aggression have been used in our in two prior

tests of disjunctive escalation of aggression within the

TDA paradigm (Pedersen et al., 2000; studies 1 and 2).

Although irrefutably convincing evidence was not obtained for the attributional distortion hypothesis,

limited aspects of the current results are consistent with

TDA theory predictions (Miller et al., 2003). For ex-

ample, negative affect resulting from the Time 2 trigger

mediated aggression, but negative affect resulting from

the provocation did not. A more convincing test of the

attributional distortion hypothesis would have entailed

a significantly greater mean score on the attributional distortion composite in the provocation condition than

the no provocation condition only among participants

in the mild trigger conditions. Future studies with more

sensitive measures or different situational moderators

may yet demonstrate this relationship.

One interesting, albeit minor, aspect of the current

findings concerns the moderately strong trigger condi-

tions. Specifically, aggression was nonsignificantly lower among participants who received both the Time 1

provocation and Time 2 strong trigger. These results are

consistent with the two out of the three previous studies

that manipulated strong Time 1 and Time 2 events

(Baron, 1972; Baron & Bell, 1975). Researchers may

wish to explore the intriguing possibility that high levels

of provocation may actually decrease aggressive be-

havior over time. It is perfectly conceivable, and con- sistent with contemporary models of aggression

(Anderson & Bushman, 2002; Berkowitz, 1993), that

successive high levels of provocation over time may in-

crease avoidance tendencies while lower levels may in-

crease approach tendencies. Indeed, Baron and

colleagues suggested that this may be the reason for

their significant decrease in aggression observed in un-

comfortably hot, by comparison with cool rooms (Bar- on, 1972; Baron & Bell, 1975).

Perhaps the most novel aspect of the TDA theory is

its potential to explain seemingly puzzling real-world

aggressive phenomena. One can easily think of instances

in which the aggression observed in response to a minor

triggering event seemed unwarranted. In such instances,

the effects of two provocations over time disjunctively

E.A. Vasquez et al. / Journal of Experimental Social Psychology 41 (2005) 61–67 67

exceed the independent additive combination of the ef- fects of each provocation. For example, episodes of road

rage or spousal abuse are frequently elicited by minor

events. Thus, in many of these episodes, a prior provo-

cation or frustration may have contributed to the ag-

gression observed in response to a minor impoliteness

on the highway or a wife’s reminder about the uncut

lawn. Moreover, aggressive individuals may not even be

explicitly aware that their response to the person who provided the minor triggering event may be inordinately

disproportionate. Although the process details pre-

sented here are incomplete, future research conducted

within the TDA paradigm may eventually lead to a

complete understanding of situational and individual

moderators of aggressive responding among previously

provoked participants in response to a mild triggering

provocation. It is hoped that through understanding these moderators and process variables, efforts to limit

aggressive responding may be developed.

References

Anderson, C. A., & Bushman, B. J. (1997). External validity of

‘‘trivial’’ experiments: The case of laboratory aggression. Review of

General Psychology, 1, 19–41.

Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual

Review of Psychology, 53, 27–51.

Axelrod, R. (1984). The evolution of cooperation. New York: Basic

Books.

Baron, R. A. (1972). Aggression as a function of ambient temperature

and prior anger arousal. Journal of Personality and Social

Psychology, 21, 183–189.

Baron, R. A., & Bell, P. A. (1975). Aggression and heat: Mediating

effects of prior provocation and exposure to an aggressive model.

Journal of Personality and Social Psychology, 31, 825–832.

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator

variable distinction in social psychological research: Conceptual,

strategic, and statistical considerations. Journal of Personality and

Social Psychology, 51, 1173–1182.

Berkowitz, L. (1993). Aggression: Its causes, consequences, and control.

New York: McGraw-Hill.

Bettencourt, B. A., Brewer, M. B., Croak, M. A., & Miller, N. (1992).

Cooperation and the reduction of intergroup bias: The role of

reward structure and social orientation. Journal of Experimental

Social Psychology, 28, 301–319.

Dollard, J., Doob, L. W., Miller, N. E., Mowrer, O. H., & Sears, R. R.

(1939). Frustration and aggression. New Haven, CT: Yale Univer-

sity Press.

Giancola, P. R., & Chermack, S. T. (1998). Construct validity of

laboratory aggression paradigms: A response to Tedeschi

and Quigley (1996). Aggression and Violent Behavior, 3, 237–

253.

Gouldner, A. W. (1960). The norm of reciprocity. American Sociolog-

ical Review, 25, 161–178.

Hovland, C., & Sears, R. (1940). Minor studies in aggression: VI.

Correlation of lynchings with economic indices. Journal of

Psychology, 9, 301–310.

Marcus-Newhall, A., Pedersen, W. C., Carlson, M., & Miller, N.

(2000). Displaced aggression: A meta-analytic review. Journal of

Personality and Social Psychology, 78, 670–689.

Miller, N., Pedersen, W. C., Earleywine, M., & Pollock, V. E. (2003).

A theoretical model of triggered displaced aggression. Personality

and Social Psychology Review, 7, 75–97.

Nowlis, V. (1965). Research with the mood adjective checklist. In S. S.

Tompkins & C. S. Izard (Eds.), Affect, cognition, and personality:

Empirical studies. Oxford, UK: Springer Verlag.

Pedersen, W. C., Gonzales, C., & Miller, N. (2000). The moderating

effect of trivial triggering provocation on displaced aggression.

Journal of Personality and Social Psychology, 78, 913–927.

Todorov, A., & Bargh, J. A. (2002). Automatic sources of aggression.

Aggression and Violent Behavior, 7, 53–68.

Wilcox, R. W. (1998). How many discoveries have been lost by

ignoring modern statistical methods? American Psychologist, 53,

300–314.

Wilcox, R. W. (2003). Applying contemporary statistical techniques.

San Diego, CA: Academic Press.

Wilcox, R. W., & Keselman, H. J. (2003). Modern robust data analysis

methods: Measures of central tendency. Psychological Methods, 8,

254–274.

Worchel, P. (1966). Displacement and the summation of frustration.

Journal of Experimental Research in Personality, 1, 256–261.

Yuen, K. K. (1974). The two-sample trimmed t for unequal population

variances. Biometrika, 61, 165–170.