University of Southern California Dissertations and Theses

Frustration within the triggered displaced aggression paradigm

(USC Thesis Other)

Frustration within the triggered displaced aggression paradigm

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content FRUSTRATION WITHIN THE TRIGGERED DISPLACED AGGRESSION
PARADIGM

by

Jeffrey J Lamoreaux

A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(PSYCHOLOGY)

December 2006

Copyright 2006 Jeffrey J Lamoreaux
ii
Table of Contents

List of Tables iii

List of Figures iv

Abstract v

Chapter 1: Introduction 1
Triggered Displaced Aggression 2
Frustration 10

Chapter 2: The Current Research 13
Hypotheses 15

Chapter 3: Method 22
Participants and Design 22
Procedure 25

Chapter 4: Results 33

Chapter 5: Discussion 52

References 62

Appendix 66

iii
List of Tables

Table 1- Age of participants in total and by experimenter. 23

Table 2- Means and Standard Deviations of the responses to the six significantly
different emotion words from the frustration manipulation check, by
condition. 37

Table 3 - Means and Standard Deviations of the responses for all words from
the frustration manipulation check, by condition. 72
iv
List of Figures

Figure 1- The General Affective Aggression Model from Anderson, Anderson, &
Deuser (1996). 7

Figure 2- Mean percentage of words completed aggressively out of total words
completed as a function of provocation and frustration. 41

Figure 3- Mean percentage of aggressive action-related words out of total words
completed that had possible answers in that category as a function of
provocation and frustration. 44

Figure 4- Mean percentage of arousal words out of total words completed with
answers in that category as a function of provocation and frustration. 44

Figure 5- Mean percentage of anger-related affect words out of total words
completed that had possible answers in that category as a function of
provocation and frustration. 45

Figure 6- Mean rating of the experimenter for a coveted research position as a
function of provocation and frustration. 47

Figure 7- Path model representing cognitive accessibility as a mediator of the
effects of provocation and trigger on aggression. 50

Figure 8- Path model representing aggression as a mediator of the effects of
provocation and trigger on cognitive accessibility. 50

Figure 9 - Diagram of path analysis involving aggression and the cognitive
accessibility of action words, first “forwards”, then “backwards”. 73

Figure 10 - Diagram of path analysis involving aggression and the cognitive
accessibility of affect words, first “forwards”, then “backwards”. 74

Figure 11 - Diagram of path analysis involving aggression and the cognitive
accessibility of arousal words, first “forwards”, then “backwards”. 75

v
Abstract
Tested the hypotheses, consistent with previous research on Triggered
Displaced Aggression (Miller et al. 2003) and the predictions of both the Cognitive-
neoassociationistic model (Berkowitz, 1983) and the General Affective Aggression
Model (Anderson et al. 1996), that a major provocation followed by a minor
frustrating triggering event would result in disjunctive escalation of aggression
greater than the additive effects of the two events experienced separately as well as
similar magnification of the cognitive accessibility of aggressive constructs. Both
hypotheses were confirmed, supporting the theoretical ideas that frustrations and
provocations are functionally equivalent and that Triggered Displaced Aggression
occurs when a Time 2 trigger follows a Time 1 provocation. The path of mediation
between accessibility of aggression-related cognitions and aggressive inclinations is
also discussed.

1
Chapter 1: Introduction
Imagine that Jim, an ordinary man, makes a mistake at work resulting in a
stern chastisement from his boss. He arrives home later that day, only to be greeted
by a mildly annoying comment from his wife that he should “get a haircut”. In
response, Jim strikes his wife.
This anecdote is an example of a well-documented phenomenon known as
Triggered Displaced Aggression (TDA) (Bushman, Bonnacci, Pedersen, Vasquez, &
Miller, 2005; Miller, Pedersen, Earlywine, & Pollock, 2000; Pedersen, Gonzalez, and
Miller, 2000). TDA can lead to anything from verbal fights to serious acts of
domestic abuse or even public violence (Denson, Pedersen & Miller, 2006). Put
simply, TDA consists of three events occurring at subsequent points in time: first the
Time 1 Provocation, then the Time 2 Trigger, and concluding with aggression on the
part of the victim of the provocation/trigger combination (Pedersen et al., 2000).
TDA is of special interest because the resulting aggression runs counter to
typical expectations by not always being additive in nature. When the trigger is
minor, aggression disjunctively escalates (Miller et al., 2003, Pedersen et al., 2000;
Vasquez et al, 2005). This means that if the aggression levels in response to the
provocation and to the trigger were each measured separately without the TDA
relationship, when added together their sum would be significantly lower than the
aggression resulting from the interaction of the Time 2 trigger with the Time 1
provocation. Thus, TDA can exceed the level of response to the trigger that one
2
would expect on the basis of the tit-for-tat matching rule (Axelrod, 1982), a rule that
is typically followed in most social interactions. Given this unusual severity in the
potential consequences of TDA, it is important that we come to understand its nature,
and how to influence its outcomes. Through TDA, seemingly minor transgressions
can lead to unexpectedly severe retaliation.
The purpose of this study was to extend the TDA paradigm by examining
frustrating instigations, by contrast with the provocations that have been used
heretofore within the TDA paradigm. The introduction is organized as follows: First,
TDA is examined in more detail, discussing it in light of two major frameworks of
aggression theory: the Cognitive-neoassociationistic model of aggression (CNA;
Berkowitz, 1993) and the General Affective Aggression Model (GAAM; Anderson,
1997). Second, the role of frustration is examined within both of the aforementioned
aggression models. Finally, these concepts are linked together to develop the
argument that Triggered Displaced Aggression is not limited to reactions to
interpersonal provocations, but also occurs in response to combinations of any of the
other sources of provocation outlined in the CNA model or GAAM, in this case,
frustrations.
Triggered Displaced Aggression
As previously stated, Triggered Displaced Aggression occurs when a person
first receives a major Time 1 provocation in circumstances that preclude retaliation
(Miller, et al. 2003) which is then followed by a second Time 2 triggering
3
provocation. The individual who is the source of the trigger is typically the target of
the aggression (Pedersen, 2000).
In order for the trigger to elicit a disjunctively escalated aggressive
retaliation, it is important that it be relatively minor; a major provocation at Time 2
will simply invoke the matching rule rather than leading to disjunctive escalation, as
implied by two studies which used strongly provoking Time 2 triggers (Worchel,
1966; Baron & Bell, 1975) and as experimentally shown by Vasquez et al. (2005).
For disjunctive escalation to occur, the trigger must instead be ambiguous in its
meaning and intentionality, which leaves it open to attributional distortion (Miller, et
al, 2003). In effect, a previously provoked person, in comparison to someone who
has not been provoked, is more likely to interpret a minor trigger as intentionally
harmful. It should be noted that displaced aggression still occurs in response to a
trigger that is as intense as a prior provocation, but the synergistic interaction that is
greater than the sum of the provocation and the trigger only occurs when the trigger
is relatively trivial (Vasquez et al., 2005).
Cognitive-neoassociationistic Model
The interactive effect in the TDA paradigm can be explained within the
framework of either of two major theories of aggression. The Cognitive-
neoassociationistic (CNA) model of aggression (Berkowitz, 1983) is considered first.
The CNA model proposes a two-step process in reaction to negative events.
The first step consists of the initial expressive-motor reactions, feelings, thoughts,
4
and memories associated with primitive fight (related to anger) and flight (related to
fear) responses: the immediate, visceral reaction. The second step, which may or
may not be relevant to the aggressive response that actually manifests itself, follows
almost immediately as higher-order processing begins and cognitions other than the
initial fight-and/or-flight reaction begin to emerge, influencing experience of the
negative affect that initialized with the automatic response. Moreover, one of the
strengths of the CNA model is that it makes no claim that this second stage will
necessarily occur, allowing the model to explain instances of aggression in which
attributions are crucial while simultaneously accounting for those in which
attributions do not seem to play a part.
Further, the CNA model posits that aversive events (any type of aversive
event, including but not limited to provocations, frustrations, and even experiences
of physical pain and discomfort) lead to aggression by generating negative affect,
which then activates an associated network consisting of the thoughts, feelings,
memories, and physiological and motor responses related to that negative affect.
According to the theory, these related feelings, ideas, and memories are linked
together in a cognitive network, and activation of one component in the network
tends to activate and/or make more readily available the cognitions with which it is
associated. Therefore, any prior priming of a part of the cognitive network (such as
aggressive cognitions related to a provocation) will make other related cognitions
(such as further aggressive cognitions) more readily accessible.
5
Miller et al. (2003) argue that Triggered Displaced Aggression is consistent
with the predictions of the CNA model. Under this model, the Time 1 provocation,
being relatively severe and obviously aggressive, would almost certainly be
interpreted as an attack, leading to the activation of aggressive thoughts and their
related cognitive-neoassociationistic networks. With this network activated and
aggression-related cognitions more readily accessible, the Time 2 trigger is likely to
be subject to attributional distortion, being interpreted as provoking even though it
would have been instead interpreted as innocuous in the absence of the prior
aggressive priming. Interestingly, in the CNA framework, a person’s reaction to the
Time 2 trigger will be determined more by the previous negative affect (part of step
1 of their experience of the event) than by objective attributions made about the
event itself. Therefore, the attributional distortion resulting in greater aggressive
response seen in TDA can be explained within the CNA model.
Rumination
Normally, aggressive arousal, and the activation of the associative network
elicited by the initial provocation, only persists over a short window (10-20 minutes).
However, this limit is routinely circumvented by rumination, a process in which the
person continues to think about the provocation and/or its negative affect (Bushman
et al., 2005). As interpreted within the context of the CNA model, rumination keeps
the network of aggressive thoughts, feelings, and memories active and accessible,
allowing the aggression from the provocation to be displaced to a later time as long
6
as the rumination continues. In our opening anecdote, Jim might have spent the day
ruminating about what his boss said to him, thus preserving his state of being primed
to aggress.
General Affective Aggression Model
The GAAM (Anderson et al. 1996, Anderson, 1997, Lindsay & Anderson,
2000) also integrates aggression-related cognitive processes into its
conceptualization of aggression, but considers them one of three variables that
produce aggressive behavior, the other two being an anger-related affective state and
arousal. The GAAM posits that any or all of these three factors may be involved in
any particular instance of affectively based aggressive behavior. Whatever
combination of the three that occurs would then influence the person’s appraisal
processes, which include both the interpretation of the situation (which includes
assigning blame) and the interpretation and experience of affect.

7
Figure 1: The General Affective Aggression Model from Anderson, Anderson, & Deuser
(1996).
The TDA effect is also predicted within the GAAM framework, albeit in a
different way. According to the GAAM, the Time 1 provocation would result in a
response consisting of some combination of aggressive cognitions, negative affect,
and arousal. These elements, still being intact (or perhaps preserved through
rumination) at Time 2, would influence the person’s appraisal processes, as they
would had they actually been caused by the trigger. For instance, Jim might notice
8
that he is very aroused and angry at the time his wife annoys him and misinterpret
those factors as being a response to her comments when they are actually due to the
provocation he underwent at Time 1, thus influencing his appraisals and causing him
to react more aggressively towards his wife at Time 2.
The disjunctive escalation of aggression seen in TDA can be explained in the
context of the GAAM because the model indicates that the responses elicited by the
Time 1 provocation will influence the appraisal of the Time 2 trigger. However, this
appraisal of the trigger as a truly provoking event occurs before a decision is made
about how aggressively to react, and therein lies the opportunity for the disjunctive
escalation of aggressive response expected within the TDA paradigm. Once the
evaluation of the Time 2 trigger as a more severely provoking event is made, it
follows logically that greater generation of anger-related cognitions, affect, and
arousal will occur in response to that Time 2 trigger in comparison to the results
produced when a person evaluates the trigger as truly trivial. Therefore, not only
would the original cognitions, affect, and/or arousal from Time 1 remain, but they
would then be compounded with those that were generated by the Time 2 trigger
(which was only interpreted as more severely provoking as a result of the Time 1
event). Thus, instead of an aggressive response commensurate with the additive
effects of a major provocation and a minor trigger, we see aggression that is
essentially in response to what has been interpreted as two major provocations.
Naturally, this combined effect is greater than that expected when the trigger has
9
been interpreted as minor, therefore resulting in the disjunctive escalation of
aggressive response displayed within the TDA paradigm. The Time 1 provocation
not only makes it likely that the Time 2 trigger will be appraised as a serious offense,
but it also adds its own cognitions, affect, and/or arousal into the equation (which
now also includes the greater aggression-related response to the trigger) that
determines the actor’s decision about the appropriate level of aggressive action.
Therefore, the CNA model and the GAAM can both provide a context in
which the disjunctive escalation seen in TDA can be explained. Though their
predictions of the outcome of a TDA situation are similar, there are subtle
differences between the expectations of the two models. According to the CNA
model, greater accessibility of aggressive constructs and negative affect lead to
attributional distortion (which may be automatic, bypassing conscious attribution
altogether), thereby producing disjunctive escalation of aggression. Conversely, the
GAAM predicts that remaining aggressive thought, anger-related emotions, and
arousal from the initial provocation will likely be misattributed to the Time 2 trigger,
leading the person to interpret the minor trigger as actually having been major, which
compounds aggression to cause a display of disjunctively escalated aggression. The
key difference between the two TDA paths is that CNA theorizes that the
accessibility leads to an unconscious misinterpretation of the trigger itself as a more
major attack, while the GAAM claims that the thoughts, feelings, and arousal
remaining from Time 1 are misattributed to the Time 2 event and then compounded
10
before action is taken. Thus, though the two models arrive at their conclusion by
different paths, both conclude that disjunctive escalation of aggression is likely to
occur as a result of a major Time 1 provocation followed by a minor Time 2 trigger.
Frustration
Unlike a provocation, which is an interpersonal event that produces anger,
frustration refers to interference with task completion (Dollard et al., 1939).
However, frustration is not simply the absence or prevention of a reward; a number
of conditions must be met in order for an experience to qualify as frustrating. First,
the frustration must block a desired outcome. If the person does not care whether or
not they reach the goal, they will not be frustrated by interference, and in fact would
not be striving to achieve the goal in the first place, making the blocking essentially
inconsequential. Second, the person must not only view the goal as attainable, but
the person must expect to attain it. For example, because one does not expect to
actually be able to fly by flapping one’s arms, being unable to do so does not
generally result in frustration. The severity of a frustration can be adjusted by
modifying these variables; frustration will be increased as desire for the goal
increases and when attainability of the goal increases (in terms of how close the
person is to reaching the goal) (Worchel, 1974).
Both the CNA model and the GAAM have grown, at least in part, from the
decades-long debate over Dollard et al.’s 1939 monograph presenting the
Frustration-Aggression Hypothesis, which posited that “Aggression is always a
11
consequence of frustration.” (p.1) The hypothesis has since been narrowed and
refined by other aggression theorists, but it was originally thought that only
illegitimate thwartings (ones that violate social rules, also called “arbitrary”
frustrations) would lead to an instigation to aggression, an idea supported by
Pastore’s 1952 study that asked students to rate how frustrated they would be by an
arbitrary frustration (bus driver drove by and for no apparent reason failed to pick
you up as you waited at the stop) compared to a non-arbitrary frustration (the bus
drove by, but it was marked special because it was “on its way to the garage”).
Though the students evaluated the arbitrary frustration as more provoking, the study
had a number of problems that made its results inconclusive (see Berkowitz, 1982.
Baron & Ransberger 1978, and Carlsmith & Anderson 1979).
More recent work on the subject suggests that even non-arbitrary thwartings
can elicit aggressive inclinations. For instance, Averill reported a study in 1982 in
which 11 percent of participants reported being angered by someone’s deliberate and
socially inappropriate behavior, but also 7 percent admitted being angered by an
unavoidable accident or event. According to Berkowitz (1993), there is every
indication that non-arbitrary thwartings do result in inclinations to aggression, but
these inclinations are “suppressed because the frustrated persons think they will be
punished in some way (at least by disapproval) if they are openly aggressive.” (pg.
43-44). The perceived likelihood of punishment would naturally increase when the
12
frustration occurs within the normal social rules (that is, when the frustration is non-
arbitrary).
Both the CNA model and the GAAM posit that frustrations lead to
aggression in the same way that interpersonal provocations lead to aggression; in
other words, all types of aversive events (assuming equal severity) are functionally
interchangeable. In the case of the CNA model, both frustrations and provocations
(as well as other aversive events, such as pain and discomfort) are theorized to elicit
an inclination to aggress in the same way (though not necessarily always to the same
degree) and activate aggression-related cognitions and their associated networks
(Berkowitz, 1989). Similarly, the GAAM maintains that both frustrations and
provocations bring forth a combination of aggression-related cognitions, anger-
related affect, and arousal (Lindsay & Anderson, 2000). If these models are
accurate, it follows that frustrations and provocations will elicit similar inclinations
to aggress within the TDA paradigm as well, though Miller et al. (2003) “would not
expect [frustrations] to yield escalations of retaliatory aggression that are as potent as
those elicited by provocations.” (pg. 79)
One criticism of the equivalence of provocations and frustrations as
instigators of aggression is that provocations tend to elicit much stronger aggressive
reactions than frustrations (Buss, 1963 and 1966, Baron 1977). However, Berkowitz
(1993) argues that the studies used to support this criticism only compare major
provocations and relatively minor frustrations. In this case, there is no reason to
13
dismiss the notion that provocations and frustrations are not distinct in the way that
they cause aggression, or that a frustration of equal magnitude to a particular
provocation will not elicit an identical response. The aggression measured in
separate studies by Pastore and by Worchel in response to frustrations, as well as a
recent study by Vasquez, Bartsch, Pedersen, & Miller (under review) which found
that direct displaced aggression that is facilitated by a frustration trigger results in
disjunctive escalation similar to that seen in TDA, buttress this argument.

Chapter 2: The Current Research
Now, imagine Jim’s scenario again, only this time, Jim is not married, and
instead has a dog. Jim arrives home from work after being berated by his boss and
takes a seat on his couch; the dog is sitting nearby in a non-irritating manner. Jim
picks up the television remote, expecting to watch his favorite TV show to relax after
a rough day. However, he finds that cable TV service for his area is down
temporarily, and the screen informs him that it is due to an unavoidable accident
resulting from a storm. Having been denied his expected pleasure, and since no one
is there to see him and thus punish him or disapprove of him, Jim kicks the dog.
This is another example of TDA at work, but in this case the typical
interpersonal trigger was replaced by a frustrating trigger. This type of trigger is
especially different from an interpersonal provocation in that it is “unsourced”: there
is no single entity that can rationally be personally blamed for the frustration (and
were such blame possible, it would liken it to an interpersonal trigger). Typical
14
examples of such frustrations include jammed vending machines, computer crashes,
and unusually tight jar lids. In our example, Jim experienced a non-arbitrary
frustration (it happened for a “good reason”) that could not be attributed to a specific
target. Jim experienced a non-arbitrary frustration, and lashed out against another
living thing as a result. Without the constraint of potential punishment, Jim feels
free to act on his anger, which he would have been less likely to do had there been a
risk of retaliation or punishment.
Previous TDA research being limited to interpersonal provocations at both
Time 1 and Time 2, the goal in the current research was the examine the effects of
replacing an interpersonal trigger with a non-arbitrary frustration. A non-arbitrary
frustration was chosen because arbitrary frustrations are very similar to provocations,
with both emerging from apparently malicious intent on the part of a specific
aggressor. This initial inquiry was limited to placing the frustration in the Time 2
triggering position and leaving the strong Time 1 interpersonal provocation intact, in
the belief that adhering as closely as possible to established research would give us
greater confidence in interpreting our results. The current research provides an
opportunity to test the predictions of the CNA and GAAM as they relate to
frustration within the context of TDA.
One important variable measured in this study was the cognitive accessibility
of aggressive constructs. This term specifically refers to the aggression-related
cognitions that make up the elements of a cognitive-neoassociationistic network.
According to CNA theory, higher accessibility of the cognitions within this network
15
corresponds to higher levels of aggressive inclination. A high level of accessibility
of these cognitions indicates that aggression is on the subject’s mind, regardless of
whether they are consciously aware of the intensity of their own thoughts about
aggression (Anderson et al., 2004, Pedersen et al., under review). This variable was
of interest as a way to corroborate previous findings related to the CNA model and to
support the idea that the activation of these cognitive networks relates to TDA. As
will be indicated in the method section, the GAAM was integrated into the
operationalization of the cognitive accessibility measure in order to test the
relationship between frustration-triggered TDA and the three elements that are
theorized to lead to aggression in the GAAM.
Hypotheses
Three hypotheses were formulated in light of the predictions of the CNA and
GAAM models with respect to TDA. They are based on the expected outcomes in a
2(provocation/no provocation) x 2(frustrating trigger, no trigger) experimental
design.
(H1) Participants who are provoked at Time 1 and experience a (non-
arbitrary) triggering frustration at Time 2 will display disjunctively
escalated aggression in comparison to those who receive either the
provocation or frustration exclusively or neither, responding
significantly more aggressively when given the opportunity to aggress
against another person.
16
(H2) Consistent with the CNA model’s expectation that greater cognitive
accessibility of aggressive constructs accompany greater aggression, the
cognitive accessibility of aggressive constructs (which encompasses all
three GAAM categories of aggression-related cognitions, anger-related
affect, and arousal) will follow a similar pattern to aggression itself:
those who are provoked and then receive a minor frustration will
experience greater overall accessibility of aggression-related constructs
than those who only undergo the provocation, the frustration, or neither.
This hypothesis is also in line with the GAAM, as that aggression
model predicts that a greater combined presence of the three categories
of aggressive constructs will accompany a disjunctive escalation of
aggression.
(H3) The provocation and the frustrating trigger will each elicit different
reactions in reference to the three GAAM categories. Specifically, the
provocation, being closer in nature to a direct attack, will produce more
action-related thoughts than the frustrating trigger. In contrast, the
frustrating trigger, being more recent and fresh in the participant’s
mind, will result in a greater display of arousal when compared to the
provocation. Finally, the provocation, being somewhat more severe,
will induce more negative affect than the frustrating trigger. To be
clear, both a provocation and a frustrating trigger will each create
17
arousal, but as time passes, arousal will dissipate and/or be dominated
by aggressive action or affect. In the experimental design, the
provocation comes 2-6 minutes before the measure of cognitive
accessibility, whereas the trigger always comes immediately before the
measure.
It is clear how this third hypothesis is consistent with the GAAM, in that the
model theorizes that aggressive responses, even those of similar magnitude, can be
caused by very different mixes of the three categories of aggressive instigation. It is
less immediately clear how this hypothesis is consistent with the CNA model.
However, the CNA model does not preclude the possibility that instigations to
aggress, though all similarly causing negative affect, may activate different areas of
the aggression-related cognitive network based on their attributes. Thus, while still
fitting within the CNA model’s framework, a direct attack will activate more
cognitions about aggressive action than a non-attack instigation to aggression, such
as the frustration trigger. Similarly, the triggering frustration, being more recent, will
evince more remaining activation of the arousal areas of the aggression-related
cognitive network. Therefore, the third hypothesis can be explained by both models
and, if corroborated in the data analysis, will support both models.
These three hypotheses can each be supported by specific statistical
outcomes, assuming the manipulations are effective. The aggressive response in the
no provocation/no trigger cell can be considered to be the baseline level of
18
aggression, because participants in that condition experience no instigation to
aggress. Regarding the first hypothesis, in the absence of a Time 1 provocation, no
differential aggression is expected between the frustrating and the non frustrating
trigger cells because the frustration is quite minor (and because previous TDA
research that has used a minor provocation as the triggering event has generally not
found a reliable difference between the means in trigger-only cells and those in
control cells). By contrast, in the absence of a triggering frustration, the Time 1
provocation, being more major, will reliably yield greater aggression than the no
provocation condition. Finally, the provocation/frustrating trigger cell is also
expected to reliably differ from the no provocation/no frustrating trigger cell.
Additionally, however, in accord with the definition of disjunctive escalation of
aggression as being a level of aggression that exceeds the independent additive
effects of the contributing instigations, the difference between the mean aggressive
responses in the provocation/frustrating trigger cell and the no provocation/no
frustrating trigger cells will exceed the sum of the differences between the no
provocation/no frustrating trigger cell and the remaining two cells -- specifically, the
provocation/no frustrating trigger and the no provocation/frustrating trigger cells.
Put more simply, the frustrating trigger will produce very little aggression
because it is minor, the provocation will produce a noticeable level of aggression
because it was designed to be more potent than the trigger, and the combination of
the two will produce substantial aggression, exceeding the combined independent
19
effects of the provocation and the frustrating trigger. The assumption of disjunctive
escalation would be supported if a 2x2 ANOVA revealed a significant interaction of
provocation x frustration trigger. That is, if subtracting the mean in the control cell
from both the provocation-only and trigger-only cells, summing these differences,
and then subtracting that sum from the difference between the
provocation/frustrating trigger cell and the control cell yields a positive number, the
hypothesis of disjunctive escalation will be supported.
For the second hypothesis, which concerns change in the general cognitive
accessibility of aggressive constructs, between-cell effects identical to those
predicted for aggression are expected. More specifically, accessibility in the no
provocation/no frustrating trigger constitutes the baseline. No increase in
accessibility is expected from the minor frustrating trigger alone. A reliable increase
over that in the no provocation/no frustrating trigger cell is expected as a result of the
provocation alone. And likewise, a reliable difference is also expected between the
no provocation/no frustrating trigger cell and the provocation/frustrating trigger cell.
Again, the difference between the mean accessibility in the control cell and the
provocation/frustrating trigger cell is expected to exceed the sum of the mean
increase in accessibility elicited in the remaining two cells, thereby indicating
disjunctive escalation in the accessibility of aggressive cognitions. Thus, if the
outcomes from the statistical tests outlined above for aggression also hold for the
cognitive accessibility scores as well, the second hypothesis would be supported.
20
The preceding considerations and expectations suggest a specific a priori
analytic strategy, namely, the application of a set of contrast weights to each cell to
test the specific predictions. As explained above, the no provocation/no frustrating
trigger was expected to display the least aggression and cognitive accessibility of
aggressive constructs. The no provocation/frustrating trigger cell was expected to be
essentially equal to the control cell. Therefore, both of these cells should be assigned
the same weight, and this weight should be the lowest. The provocation/frustrating
trigger cell was expected to yield the highest scores for both aggression and
cognitive accessibility, meaning that this weight should be the highest.
The provocation/no frustrating trigger cell warrants further discussion. As
indicated, theory predicts that it will elicit aggression and cognitive accessibility
scores that exceed those in the control or trigger-only cells, but be substantially
lower than the provocation/frustrating trigger cell. Contrary to this prediction,
however, previous TDA research such as Pedersen et al. (2000 has found a contrast
effect in which aggression toward the triggering co-participant was reduced in the
provocation-only cell compared to the no provocation/ no trigger control cell. In
these studies, participants received mildly negative feedback on their work on a
problem-solving task from a partner as the trigger manipulation, and neutral
feedback in the no trigger condition. Thus, this contrast effect occurred in the
context of an interpersonal trigger manipulation. After having endured a major
provocation from one person and then having received neutral feedback from another
21
person (the displaced aggression target), participants could compare the behavior of
the displaced aggression target against the behavior of the provocateur from Time 1.
In prior TDA research (e.g. Pederseen. 2000) such interpersonal comparison resulted
in the non-triggering partner being seen as nicer when one had previously been
strongly insulted by the Time 1 provocateur than when one had not, resulting in the
obtained contrast effect. In the context of the current research, however, there is no
reason to expect such a contrast effect because participants will not view the
displaced aggression target as having passed on an opportunity to aggress (which is
what happened in the control conditions of previous TDA research with interpersonal
triggers). . A key feature of the current design is that the presence or absence of the
frustrating trigger is not due to an interpersonal decision of the part of the target of
displaced aggression. Since the displaced aggression target’s behavior is not directly
responsible for the trigger in the current research (they are not displaying a choice
not to aggress because the study does not give them a specific opportunity to aggress
the way past studies have), participants will have no basis on which to contrast the
target with the provocateur. Therefore, contrary to previous TDA research, the
provocation-only cell is expected to elicit greater aggression and cognitive
accessibility of aggressive constructs than the control cell. The preceding
considerations suggested the following cell weights for the four cells of the 2 x 2
design: 2, 0, -1, -1, with 2 representing the provocation/frustrating trigger cell, 0
representing the provocation/no frustrating trigger cell, and the -1s representing the
22
no provocation/frustrating trigger cell and the no provocation/no frustrating trigger
cell.
Chapter 3: Method
Participants and Design
Ninety-three volunteers were recruited by two female and two male
experimenters, all four of whom were students ranging in age from 19 to 24 years.
Each experimenter collected data from volunteers in various locations on the
University of Southern California campus. The first female experimenter recruited
22 volunteers, the second female experimenter recruited 21 volunteers, the first male
experimenter recruited 16 volunteers, and the second male experimenter recruited 34
volunteers. The recruited volunteers were predominantly students, with only about 6
percent of the participants falling outside of the 18-24 age range and over 80 percent
between the ages of 18 and 22 (see Table 1 for a detailed summary of participant
ages in total and by experimenter).
23
Table 1: Age of participants in total and by experimenter.
Age

Experimenter 1

Experimenter 2

Experimenter 3

Experimenter 4 Total
N= 22 21 15 35 93
average 20.59 22.43 19.87 20.34 20.8075

18 5 3 1 9 18
19 4 1 8 8 21
20 4 3 1 4 12
21 2 3 3 7 15
22 2 2 1 3 8
23 2 4 1 1 8
24 1 3 0 1 5

25 1 0 0 1 2
26 0 0 0 0 0
27 1 0 0 0 1
28 0 1 0 0 1

33 0 0 0 1 1
40 0 1 0 0 1
Each participant was randomly assigned to one of four conditions. The study
was a 2 (provocation: Yes/No) X 2 (frustrating trigger: Yes/No) design. All four
cells contained similar age distributions: provocation/frustration, M=21, SD=3.3;
provocation/no frustration, M=20.8, SD=2.5; no provocation/frustration, M=21.6,
SD=5.4; no provocation/no frustration, M=19.94, SD=1.8. The differences in the
standard deviations (as well as the slight differences in the means) of the ages
between these cells can be accounted for by the fact that the no
provocation/frustration condition (which had the highest M and SD) contained the
only 40-year- old and the only 28-year-old, while the no provocation/no frustration
cell (with the lowest M and SD) contained no participants over the age of 24 except
24
for the study’s only 25-year-old. Of the participants over age 24, there was also a
27-year-old in the provocation/no frustration cell and a 33-year-old in the
provocation/frustration cell.
The general procedure of the study was as follows: First, the participants read
an article that served as either a provoking or neutral stimulus and briefly gave their
opinion of the article. Then, the participants in the frustration condition attempted to
complete a problem-solving task (opening a trick box) within a limited span of time.
Participants were led to believe that they should be able to complete the task during
that time, though this was nearly impossible. This problem-solving task served as
the manipulation of frustration, as participants were given a goal that they expected
to be able to accomplish, only to have the attainment of that goal blocked.
Participants in the no frustration condition did not attempt the problem solving task,
but instead skipped to the next step, a measure of the cognitive accessibility of
aggressive constructs (which those in the frustration condition also completed, albeit
after the problem-solving task). Finally, participants rated the experimenter (who
had neither aggressed nor directly caused the frustration) as a behavioral measure of
aggression. The participants believed that their ratings of the experimenter’s
performance would influence the experimenter’s chances of receiving a highly
coveted research position; therefore, the participants thought that by giving the
experimenter a poor rating they could harm the experimenter by reducing the
experimenter’s chances of being hired for the position. Participants were led to
believe the experimenter would never see the evaluations. As will be explained in
25
more depth below, these ratings constitute a measure of aggression because giving a
poor rating would be acting with the intent to cause harm to the experimenter.
Procedure
All participants were recruited using the same script. An experimenter
approached them and said, “Hello, I’m running studies for the Psychology
Department, would you be willing to participate? The studies usually take about 15
minutes.” Upon agreement, participants received an informed consent form. On
average, roughly two out of every three people who were asked to participate
refused. After reading and signing the consent form, participants were told that in
order to make good use of their time, they would participate in two brief studies that
were unrelated to each other. They were informed that the first study focused on
opinions, and the second on problem-solving ability. Participants were recruited
from a number of locations on campus, with an eye toward avoiding people who
were eating or with friends. The areas from which participants were recruited
included a pedestrian road flanked by an administrative building on one side and a
football practice field on the other (with a large, low-lying stairway and ledges on
which students tended to sit), a courtyard near the engineering building (also with
tables and ledges near a large fountain surrounded by trees), at University tram (bus)
stops throughout campus, in front of the psychology building and other classroom
buildings, and at various sidewalk areas (experimenters would walk along the
26
sidewalks of campus looking for recruits). The majority of the participants were
recruited from the first three locations mentioned above.
Time 1 provocation
Participants then received a newspaper clipping, ostensibly from a past
edition of the USC student newspaper, and a response sheet that contained questions
about basic demographic information and two free-response items, one asking “What
is your opinion about the subject of the article you read?” and “What do you think of
the author of the article?” These two items served both to bolster the cover story and
as manipulation checks of the provocation. Participants in the no provocation
condition were given a neutral article about a museum exhibit (see Appendix). This
article was selected because it did not use particularly positive or negative language
and the subject matter was not likely to cause an emotional reaction for participants.
In contrast, participants in the provocation condition received an op-ed piece of
roughly the same length (the neutral article was 415 words long while the provoking
article 401 words long) and appearance as the neutral article. However, it should be
noted that the neutral article appeared to be longer because it tended to contain
longer words, as can be seen in the Appendix. The provoking article used strongly
negative language to belittle college students and told them to “grow up” and “stop
whining” about yearly increases to tuition costs (see Appendix). This article was
specifically fabricated for this experiment with the purpose of insulting the
27
participant while remaining believable. The names of the authors for both articles
were hidden.
Time 2 frustrating trigger
After reading the article and filling out the opinion response form,
participants were thanked and told that they would now be moving on to the next
study. Participants in the frustration condition were told that the second study
consisted of two different kinds of problem-solving tasks, while those in the no
frustration condition were told that it consisted of one problem-solving task. The
first of these tasks, administered only to those in the frustration condition, actually
served as the Time 2 Frustration Trigger manipulation, while the second,
administered to all participants, actually served as the measure of the cognitive
accessibility of aggressive constructs. As such, only participants in the frustration
condition took part in the first problem solving task (the frustration manipulation)
while participants in the no frustration condition skipped the first problem solving
task completely, being unaware of its existence. All participants completed the
second problem solving task (the cognitive accessibility of aggressive constructs)
regardless of condition. Therefore, roughly half of the participants (those in the
frustrating trigger condition) completed two problem solving tasks (the frustration
manipulation and the measure of cognitive accessibility of aggressive constructs)
while the other half (those in the no frustrating trigger condition) only completed one
28
of those problem solving tasks (the measure of cognitive accessibility of aggressive
constructs). All else was held constant between these two conditions.
The frustration manipulation was administered as follows: participants in the
frustration condition were given a small (about 8 inches long, 3 inches wide, and 3
inches thick) wooden “Magic Box.” The box was designed as a puzzle, in that there
was a very specific “trick” to opening it, and no other method would work. In this
case, the box had two moving parts to its lid, one of which was obviously easy to
rotate, and the second of which seemed locked in place, but needed to be rotated to
access the inside of the box. Linking these two parts, but invisible to the participant,
was an internal shaft containing a small pin. When the box was turned upside-down,
the pin would fall “up” the shaft, and could be locked into that position by rotating
the first moving part 180 degrees before turning the box back right-side-up. Locking
the pin like this kept it from holding the second movable part in place, thus allowing
the second (previously locked) part to be rotated, opening the box. Without the use
of this method, it would appear to the participant that one part of the box could be
moved, but the other was locked into place, and the box could not be opened.
Participants were instructed to attempt to open the box but not to try to force
it open, as it would open easily when done correctly. Participants who were given
this box were told that the average person usually opens the box within about two
minutes. In truth, practically no one can open the box in under two minutes without
prior knowledge of the correct method. However, it was necessary for participants to
29
believe otherwise because the expectation of success is one of the requirements for
the occurrence of frustration. No participants were able to open the box in two
minutes. At the end of the two minutes, the experimenter told participants that “time
is up” and took back the box. Failure to achieve the expected outcome of opening
the box served as the Time 2 Frustrating Trigger.
Since the goal was only minimally attractive, the resultant frustration was
expected to be minor and thus well suited to produce disjunctive escalation of
aggression in accord with prior work in the TDA paradigm. This manipulation also
qualified as a non-arbitrary frustration, as the actions of the experimenter were
within the acceptable social rules set by the experiment itself. The experimenter was
not purposely breaking the rules to thwart the participant, but instead simply
following the experimental design. Therefore, participants were not expected to
blame the experimenter directly or attribute to the experimenter a deliberate intent to
thwart or frustrate the participant. This attribute of the manipulation also contributed
to keeping the severity of the frustration low.
Cognitive Accessibility of Aggressive Constructs
As previously stated, all participants completed the second ostensible
problem-solving task, which was a word completion task that was actually meant to
measure the cognitive accessibility of aggressive constructs (see Appendix).
Participants in the frustration condition completed this word completion task after
completing the “Magic Box” frustration manipulation problem-solving task, while
30
participants in the no frustration condition, having skipped the frustration
manipulation task completely, completed this word completion task as the only
component of what they were told was “the second study” in which they were being
asked to participate.
This pen-and-paper word completion task consisted of a series of 34
incomplete words, with some letters provided and some missing. Each was
specifically selected so that participants could fill in letters to form either an
aggressive or a non-aggressive word. For instance, “ang _ _” could be completed as
“anger” or “angle,” “fi _ _ t” could be completed as “fight” or “first,” and “off _ _ _”
could be completed as “offend” or “office.” Participants were asked to complete the
task quickly, but without rushing. This instrument is Pedersen et al.’s (under review)
modification of a measure created by Anderson et al. (2004). Whereas the original
version only included words in the action category, the revision added affect and
arousal words. The measure was used successfully by Pedersen et al. to assess
differences in the cognitive accessibility of aggressive constructs.
Measure of Aggression
Following the word completion task, participants were thanked and led to
believe that the study was over. This was done to reduce the possibility of suspicion
about the true purpose of the final measure by making it appear to not actually be
part of the experiment. At this point, the experimenter explained that he or she was
running these studies as part of the application process for a highly coveted Research
31
Position that paid well and “looks good on a resume” (this was a fictional cover
story). As such, the experimenter asked the participant to complete an evaluation of
the experimenter’s work to be sent to the Psychology department to aid in the
department’s hiring decision. This procedure is identical to the one used by
Pedersen, et al. (2000) to measure aggression and intent to harm.
The experimenter then handed the participant an experimenter evaluation
sheet and a stamped envelope addressed to Dr. Norman Miller of the Psychology
Department (who was ostensibly in charge of the hiring process). Participants were
instructed complete 5 items on the sheet on a 7 point linear scale; examples of items
were “I think the experimenter is competent” and “The experimenter did a good
job.” Participants were told that their responses would be completely anonymous
(they would only be identified by a participant number that was kept separate from
their consent form), and that they should seal their response in the envelope and mail
it so that the experimenter would never see it. After the participant had sealed the
envelope, the experimenter offered to drop the envelope in the mailbox as a
convenience for the participants. All but one of the participants gave the envelope to
the experimenter to mail, and that participant’s evaluation was successfully received
through the mail.
It was important that the evaluation appeared to be anonymous because, as
established in our discussion of frustration, fear of reprisal and/or disapproval is the
major reason that aggressive inclinations resulting from non-arbitrary frustrations are
32
often not behaviorally expressed. Assuring anonymity removed the potential for
punishment and/or disapproval, freeing the participants to express their aggression.
Finally, participants were thanked, given a debriefing document, given the
opportunity to ask questions and/or withdraw their participation with no penalty, and
reminded to mail the evaluation if necessary.
Frustration Manipulation Check Study
In order to limit the length of the study itself and reduce potential for
suspicion, a check of the frustration manipulation was conducted separately. The
check was a two condition design: participants were either in the frustration
condition, or the no frustration condition. Twenty people participated in this study,
with nine in the frustration condition and eleven in the no frustration condition.
The procedure of this manipulation check was as follows: In both the
frustration and no frustration conditions, participants were approached in the field at
public locations on the same campus as the one from which participants for the main
study were recruited. They were asked to take part in a brief survey that involved
some reading, a problem-solving task, and a questionnaire, overall taking about five
minutes. After reading an informed consent form, participants read the neutral
article from the main study. Those in the frustration condition were then given the
“magic box” frustration manipulation exactly as it had been administered in the main
study. This was followed by a list of twenty-eight words describing feelings or
moods, such as angry, confused, pleased, frustrated, and sad. Participants were
33
asked to rate their own current mood by indicating the degree to which they felt each
of the items on the list on a 7-point linear scale ranging from 1 = not at all to 7 =
extremely so. This instrument is a modification of Watson, Clark, & Tellegen’s
(1998) Positive Affect Negative Affect Scale (PANAS), which has been validated by
Sandin, et al. (1999) and Crook, et al. (1998). Finally, participants were thanked and
given a verbal debriefing.
The no frustration condition was identical to the frustration condition except
that participants did not undergo the frustration manipulation. Instead, they skipped
directly from the neutral article to the mood questionnaire. This type of control
condition was more practical than a condition in which the participant succeeded in
opening the box, as this would elicit positive feelings derived from accomplishment
that would obscure the influence of the frustration.

Chapter 4: Results
Of the ninety-three original participants, sixteen were removed from the
analysis because of distraction. Though an effort was made to administer the study
in an environment devoid of distractions, data were gathered in the field on public
portions of the University of Southern California campus; occurrences of distractions
such as a passing friend of the participant or a ringing cellular phone were not
entirely avoidable. As such, participants in this study displayed a higher rate of
distraction than that seen in laboratory studies. Distracted participants were removed
34
if their distraction met any of the following three criteria. First, participants were
disqualified if they held a conversation with a passerby that was longer than a wave
or simple “hello”. Explanations of why the conversation could not continue were
allowed, as long as they did not exceed two sentences. For instance, saying “I am
busy now, I will call you later” was acceptable. Second, participants were
disqualified if they held a cellular phone conversation that lasted longer than thirty
seconds. Third, participants were disqualified if the distracting event was deemed
disruptive to the emotions manipulated by the experiment. The decision that the
event was disruptive was made only if both the experimenter who was present at the
time and the head experimenter agreed that it was disruptive. The head experimenter
made this decision based on an oral report of the event by the experimenter who had
been present. If the head experimenter was also the one who administered the study
on the occasion in question, the head experimenter would make the decision alone.
This decision was always made after the participant had been run but before the
envelope containing the aggression results had been opened. For instance, one
participant was removed because a squirrel ran in the midst of the experiment and
startled the participant, and another was removed as a result of a phone call that was
shorter than thirty seconds, but the caller informed the participant that a friend had
been in a car accident.
A boxplot also revealed two subjects to be very significant outliers; they were
removed from the analyses presented here. All analyses were run with and without
35
the two outliers, but the results were no different (nothing that had reached
significance in one analysis failed to reach significance in the other, or vice versa).
Therefore, the analysis included a total of 75 undistracted subjects. There was no
age trend with those removed: directly paralleling the overall participant pool, over
half (10 of 17) of those removed were between 18 and 20, and the rest a smooth
spread of ages between 21 and 25. All ages and age groups were removed in
numbers roughly commensurate with the age or group’s representation in the overall
data set, thus providing evidence that removal was not influenced by age. Similarly,
both the overall data set and the set of those removed displayed a nearly exact half-
and-half gender split, indicating that removal also was not biased by gender.
Provocation Manipulation Check
After reading the provoking or neutral article, participants wrote free
responses to the two “opinion” items: “What is your opinion about the subject of the
article you read?” and “What do you think of the author of the article?” Three
independent raters who were blind to the hypothesis evaluated these free responses
by rating them on three seven-point linear scales: “This person felt offended by the
article,” “This person disliked the article’s author,” and “The article made this person
angry.” The points on the scale were described by the following: “strongly
disagree”(1), “moderately disagree”(2), “slightly disagree”(3), neither agree nor
disagree”(4), “slightly agree”(5), “moderately agree”(6), and “strongly agree”(7).
36
Inter-rater reliability (alpha=.8) and inter-item reliability (alpha=.92) were
acceptably high. Therefore, the data of the three raters were collapsed across the
three items into one mean score for each participant. As expected, raters evaluated
the responses of the participants as being significantly more negative (t(50)=5.58,
p=.001) after reading the provoking article (M=4.67, SD=1.24) as opposed to the
neutral article (M=3.19, SD=1.71), indicating greater anger, offense, and dislike of
the author when provoked. Though this difference was significant, the mean rating
of the provoking article only exceeded the scale midpoint by .67, suggesting that
though the article was provoking, it may not have been as strongly so as was desired.
Frustration Manipulation Check Study
As previously indicated, the frustration induced by failure to open the box
was assessed in a separate manipulation check study. Participants responded to
twenty-eight words describing moods or feelings. They rated the degree to which
they currently felt each on a 7-point linear scale with described endpoints: 1
indicated “not at all” and 7 indicated “extremely so.” A boxplot revealed three
values that were significant outliers, which were then removed from the analysis.
Each of these outliers was provided by a different participant and occurred in
response to a different one of the following three items: playful, pleased (outliers for
both of these were found in the frustration condition), and curious (outlier found in
the no frustration condition). Please refer to the table in the Appendix for the means
and standard deviations for all words after outliers were removed.
37
Independent-samples t-tests (N=20) revealed six adjectives on which those in
the frustration and no frustration conditions differed reliably. These were:
frustrated, t(8)=4.86, p=.001, down, t(8)=3.46, p=.009, distressed, t(12.25)=2.52,
p=.026, uncertain, t(14.3)=2.65, p=.019, annoyed, t(8)=3.59, p=.007, and
concentrating t(16)=4.08, p=.002. Examples of other potentially relevant words that
did not display a difference are angry, sad, anxious, happy and offended. See Table 2
for a summary of means and standard deviations for the six significant items, and see
the Appendix for means and standard deviation of all twenty-eight items.

Table 2: Means and Standard Deviations of the responses to the six significantly different
emotion words from the frustration manipulation check, by condition.

Emotion Frustration M SD
Frustrated yes 3.78 1.72
no 1.00 0.00

Down yes 2.00 0.87
no 1.00 0.00

Distressed yes 2.33 1.12
no 1.27 0.65

Uncertain yes 4.44 1.88
no 2.45 1.37

Annoyed yes 3.22 1.86
no 1.00 0.00

Concentrating yes 4.33 1.32
no 2.00 1.15
38
A Bonferroni correction on the data was applied to protect against
experiment-wise error, which becomes a concern when conducting a large number of
t-tests. The Bonferroni correction divides the upper limit of the significance level of
the individual test by the number of tests being conducted in the overall experiment.
Because one would normally conduct all of the tests at the .05 level, and .05/28 is
about .002, experiment-wise error is controlled by conducting the tests at the .002
significance level. This left only frustrated and concentrating as significant items.
These results support our contention that participants felt frustrated as a result
of the frustration manipulation, but did not find it offensive or saddening (clearly
differentiating it from interpersonal provocations used in past research). Though
concentrating showed a difference, this effect was not accompanied by higher scores
on the items assessing anxiety, distress, or vigorousness, suggesting that it is unlikely
that it influenced the aggression of concern. It is also likely that the difference in
concentration would disappear had this questionnaire been administered as part of
the main study because participants in the no frustration condition also would have
just finished the word completion task, which also requires concentration.
Consideration of the differences that were found before the experiment-wise
correction was applied supports the conclusion of a successful manipulation of
frustration : the fact that down, distressed, uncertain, and annoyed showed reliable
differences between the conditions before application of the Bonferroni correction is
39
consistent because although all of them indicate negative affect, none were
accompanied by differences on interpersonal provocation-related items such as
offended, angry, and hostile. Thus, participants did not feel provoked even when the
data are interpreted more liberally.
The idea that this manipulation distinctly influenced frustration is further
bolstered because frustration displayed the largest and most reliable difference
among all of the items that reached significance, though it was not reliably more
significant than the next best item or a composite of the other significant items.
Nevertheless, the results of the manipulation check study outcomes supported the
argument that the trigger indeed manipulated frustration (and not provocation).
Cognitive Accessibility of Aggressive Constructs.
The thirty-four word completion items meant to test the cognitive
accessibility of aggressive constructs was partitioned into three categories based on
the aggression-related words that could potentially be constructed using the provided
letters. These three categories were: 1) Action Words (indicating behavioral
intention), made up of fourteen items including such words as “shoot” and “injure,”
2) Arousal Words, made up of nine items such as “excited” and “tense,” and 3)
Affect Words, made up of eleven items such as “annoyed” and “outrage”.
To examine effects on this instrument, percentages that represented how
many words had been filled in with a potential aggressive word out of the total
number of completed words in that category were created, as not all participants
40
completed every item. For instance, if a participant completed 30 words but only 3
of those were filled in with the aggressive word choice, that participant received a
total cognitive accessibility of aggressive constructs score of .1. Four values were
computed for each subject: percentage of action words, percentage of arousal words,
percentage of affect words, and total percentage of aggressive words, which
combines the three categories.
First, a 2X2 ANOVA was applied to the total scores, ignoring the
differentiation among the three types of aggressive cognitions. Though this did not
yield the expected significant interaction (F(1,71)=1.28, p=.26), it did reveal
significant main effects for both provocation (F(1,71)=9.71, p=.003) and frustration
(F(1,71)=3.92, p=.052). Both effects, however, were largely driven by the higher
value in the provocation/frustration cell compared to the other cells. Direct t-test
comparisons of the provocation/frustrating trigger cell (M=.32, SD=.16) with each of
the other three cells indicated that the cell in which disjunctive escalation was
expected was significantly different from every other cell: provocation/no frustrating
trigger, M=.23, SD=.08, t(31)=2.21, p=.034; no provocation/frustrating trigger,
M=.21, SD=.08, t(33)=2.77, p=.009; and no provocation/no frustrating trigger,
M=.18, SD=.09, t(34)=3.24, p=.003. The difference between the provocation/no
frustrating trigger cell and the control cell approached, but did not reach,
significance, t(37)=1.5, p=.07, while the difference between the no
provocation/frustrating trigger cell and the control cell did not even approach
41
significance t(28)=.752, p=.46. Supporting the expectation of disjunctive escalation
of the accessibility of aggressive constructs under the combined presence of both a
provocation and a subsequent frustrating trigger, the provocation/frustrating trigger
cell exceeded the baseline control cell of 18% by 14%, which is greater than the sum
of the deviation from the baseline found in the provocation-only and frustrating
trigger-only cells (5% and 3% respectively =8%). The overall results for the word
completion task combining all three categories are presented in Figure 2.
Figure 2: Mean percentage of words completed aggressively out of total words completed as
a function of provocation and frustration.
0
10
20
30
40
50
60
70
80
90
100
provocation no provocation
Percentage of Words Completed
Aggressively
frustration
no frustration

Though the initial 2X2 ANOVA did not reveal an interaction, we applied an
a priori test of our hypothesis using the contrasts weights 2 (provocation/frustrating
trigger), 0 (provocation/no frustrating trigger), -1 (no provocation/frustrating trigger),
42
and -1 (no provocation/no frustrating trigger), as explained in the Introduction, to test
the specific model predicted by theory and supported by the t-tests and the related
aggression results. As expected, our results showed the percentage of words filled in
aggressively compared to the control cell was significantly higher (roughly three
times higher) in the provocation/frustration cell compared to the provocation/no
frustration cell, while the latter had greater cognitive accessibility scores than the
remaining two cells, F(1,71)=5.16, p =.003. The residual variance was not
significant at the p=.05 level, F(1, 74)=3.54, signaling that no major segment of the
results failed to be explained by our model.
A series of 2X2 ANOVAs was then applied to each of the three GAAM
categories. The analysis of aggression-related Action words revealed only a main
effect for provocation, F(1,71)=8.62, p=.004, while, conversely, Arousal words
displayed only a main effect for frustration, F(1,71)=4, p=.049. Anger-related Affect
words also showed a main effect for frustration, F(1,71)=8.55, p=.005. However,
this Affect main effect was due almost entirely to the provocation/frustration cell,
though neither the interaction (F(1,71)=2.69, p=.11) nor the main effect for
provocation (F(1,71)=3.21, p=.078) reached significance. Again, consistent with the
expectation of disjunctive escalation, the provocation/frustrating trigger cell
exceeded the baseline control cell of 19% by 17%, which is greater than the sum of
the deviation from the baseline found in the provocation-only and frustrating trigger-
only cells (5% and 1% respectively = 6%). Direct contrasts comparing affect in the
43
provocation/frustration cell to each of the other three cells using t-tests indicated that
participants in the provocation/frustration cell (M=.36, SD=.19) reliably filled in
more affect words aggressively than did those in the provocation/no frustration cell
(M=.24, SD=.13, t(43)=2.44, p=.019), the no provocation/frustration cell (M=.19,
SD=.14, t(36)=2.99, p=.005), and the no provocation/no frustration cell (M=.19,
SD=.12, t(34)=2.93, p=.006).
An a priori contrast for this interaction, as predicted by theory and supported
by the individual t-test results, was applied to the affect data by using the contrasts
weights 2, 0, -1, and -1 to compare the cells in the same manner used with the total
cognitive scores. The significant result supported the idea that the percentage of
affect words filled in aggressively compared to the control cell was significantly
higher (roughly three times higher) in the provocation/frustration cell compared to
the provocation/no frustration cell, while the latter had greater affect cognitive
accessibility scores than the remaining two cells, F(3,71)=5.35, p=.002. The residual
variance was not significant at the p=.05 level, F(1,73)=3.78. Results for each of the
three individual categories of words are presented in Figures 3, 4, and 5.

44
Figure 3: Mean percentage of aggressive action-related words out of total words completed
that had possible answers in that category as a function of provocation and frustration.
0
10
20
30
40
50
60
70
80
90
100
provocation no provocation
Percentage of Action Words
Completed Aggressively
frustration
no frustration

Figure 4: Mean percentage of arousal words out of total words completed that had possible
answers in that category as a function of provocation and frustration.
0
10
20
30
40
50
60
70
80
90
100
provocation no provocation
Percentage of Arousal Words
Completed Aggressively
frustration
no frustration

45
Figure 5: Mean percentage of anger-related affect words out of total words completed that
had possible answers in that category as a function of provocation and frustration.
0
10
20
30
40
50
60
70
80
90
100
provocation no provocation
Percentage of Affect Words
Completed Aggressively
frustration
no frustration

Thus, responses on the word completion task supported the possibility that
the provocation, being an unambiguous attack, was the main source of cognitions
about aggressive action while the frustration, being temporally more proximal to the
measurement, was responsible for most of the arousal present. T-tests comparing
first the provocation/no frustrating trigger cell, then the no provocation/frustrating
trigger cell, to the control cell were both non-significant. Consistent with theory,
these results also suggest that while the provocation directionally generated anger-
related affect more than the minor frustration does on its own (this difference was
unreliable, t(37)=1.1, p=.278 with a mean difference of .05) when the frustration that
occurred at Time 2 was preceded by a Time 1 provocation, it generated greater
anger-related affect than it would in the absence of a previous provocation.
46
Aggression.
A 2x2 ANOVA on the single aggressive behavioral intention item (“I
recommend the experimenter to be a Student Research Assistant”), as expected,
showed main effects for provocation, F(1,71)=14.21, p=.00, and frustration,
F(1,71)=20.67, p=.00, that were driven by the interaction between provocation and
trigger, F(1,71)=4.88, p=.03.
Given the high alpha for the composite of the five items used by the
participants to evaluate the experimenter’s performance for a coveted research
position (alpha= .98) they were collapsed into one score indicating aggression and
intent to harm. A 2x2 ANOVA revealed both a main effect for provocation, F(1,
71)=20.55, p=.00, and for frustration, F(1, 71)=19.10, p=.00. Again, both of these
main effects were driven by the expected interaction of provocation and frustration,
F(1, 71)=15.22, p=.004. The expectation of disjunctive escalation is confirmed by
inspecting the means: compared to the baseline of 1.7, the provocation/frustrating
trigger cell deviated by 2.3, which is greater than the sum of the deviation from the
baseline found in the provocation-only and frustrating trigger-only cells (.1 and .2
respectively =.3). The aggression results for the 5-item composite are presented in
Figure 6.
47
Figure 6: Mean rating of the experimenter for a coveted research position as a function of
provocation and frustration. Higher scores indicate a more unfavorable rating and therefore
greater aggression.
1
2
3
4
5
6
7
provocation no provocation
Ratings of Experimenter
frustration
no frustration
An analyses of simple effects indicated that participants in the
provocation/frustration cell (M=3.95, SD=2.25) reliably rated the experimenter more
negatively than those in the provocation/no frustration cell (M=1.67, SD=.62,
t(24)=4.64, p=.00), the no provocation/frustration cell (M=1.62, SD=.69, t(26)=4.48,
p=.00), and the no provocation/no frustration cell (M=1.19, SD=.35, t(23)=5.66,
p=.00). The provocation/no frustrating trigger cell reliably differed from the no
provocation/no frustration cell t(35)=2.87, p=.007 whereas a similar comparison of
the no provocation/frustration cell with the no provocation/no frustration cell yielded
a smaller, but reliable effect t(23)=2.43, p=.023. The difference in the strength of
these effects suggests that the provocation resulted in somewhat more aggression.
48
An a priori contrast for the predicted interaction was applied to the
aggression data by using the contrasts weights 2, 0, -1, and -1 as outlined in the
Introduction. The significant result supported the idea that the ratings of the
experimenter when compared to the control cell were significantly more negative
(roughly three times so according to our model) in the provocation/frustration cell
compared to the provocation/no frustration cell, while the latter itself had more
negative experimenter ratings than the remaining two cells, F(3,71)=17.863, p=.00.
However, the residual variance was significant at the p=.05 level, F(1,73)=9.76,
indicating that the model did not explain all of the variance between the groups.
Inspection of the means suggests that the significant residual variance reflects (a) the
unanticipated reliable effect of the frustration alone and (b) an even stronger
disjunctive escalation of aggression in the provocation/frustration cell than that
predicted by the +2, 0, -1, -1 model.
Because four different experimenters administered the study, the effect of
experimenter on the aggression dependent variable measure was examined. First, a
4x2x2 ANOVA including the four experimenters, the two provocation conditions,
and the two frustration conditions only showed main effects for provocation and
frustration and the interaction between provocation and frustration. The
experimenter variable did not present a main effect, nor did it interact with any of the
other variables. For experimenter x frustration F(1,59)=.428 and p=.734 while for
49
experimenter x provocation F(1,59)=1.56 and p=.209 and for experimenter x
provocation x frustration, F(1,59)=.13 and p=.942.
A 2 x 2 x 2 ANOVA evaluating interactions amongst the provocation,
frustration, and the gender of the experimenter also failed to produce any significant
main effects or interactions involving experimenter gender. Therefore, there is no
indication that participants responded differently to any one experimenter in
comparison to the others or to female experimenters in comparison to male
experimenters.
Having found evidence that supported the idea that the combination of
provocation and frustrating trigger resulted in disjunctively escalated aggression and
significantly greater accessibility of aggression-related cognitions compared to the
other three conditions, a path analysis using Structural Equation Modeling (SEM)
was applied using the standardized values to determine the nature of the relationship
amongst the variables of concern. Heightened accessibility of aggressive cognitions
was expected as a consequence of the aversive events (the provocation and the
frustrating trigger) and, in turn, the accessibility of cognitions was expected to
mediate the influence of aversive events on displaced aggression. If so, analysis
would support a model that simulates such a path (see Figure 9), evidencing a high
“goodness of fit”, as indicated by a non-significant chi-squared value, a high (near 1)
Comparative Fit Index (CFI), and a low (near zero) standard mean-square residual
(SRMR). A SEM analysis applied to a model that reversed the accessibility of
50
aggression-related cognitions and aggression model is shown in Figure 10. Support
for the theoretically predicted model shown in Fig.9 would be further bolstered if
this reversed model (Fig. 10) had a goodness of fit that was inferior to that of our
postulated model.

Figure 7: Path model representing cognitive accessibility as a mediator of the effects of
provocation and trigger on aggression.
.02*
.1*
-.012*

.014*

Figure 8: Path model representing aggression as a mediator of the effects of provocation and
trigger on cognitive accessibility.
.34*

.1*
-.012*
. 37*
Cognitive
Accessibility
Aggression
Provocation
Frustrating
Trigger
Cognitive
Accessibility
Frustrating
Trigger
Provocation
Aggression
51
Contrary to expectations, the first model was a poor fit to the data,
2
(2) =
22.03, p = .00, CFI = .6, SRMR = .126. Meanwhile, the “backwards” model was an
extremely good fit to the data,
2
(2) = 2.484, p = .289, CFI = .99, SRMR = .043.
Thus, instead of confirming our expectations, the path analysis supported the
argument that level of aggression actually mediated the effect of provocation and
frustration trigger on the accessibility of aggression-related constructs.
To further investigate the mediation, we performed additional path analyses
of the individual elements of the total cognitive accessibility scores (action, affect,
and arousals), examining the relationships, as before, both “forwards”, i.e. with
cognitive accessibility influencing aggression, and “backwards”, i.e. with aggression
influencing cognitive accessibility.
The results of these path analyses largely confirmed the outcome of the initial
overall path analysis. For action word scores, the forwards model yielded a poor
goodness of fit,
2
(2) = 29.26, p = .00, CFI = .27, SRMR = .141, while the
backwards model was a good fit to the data,
2
(2) = 4.79, p = .09, CFI = .93, SRMR
= .074. Similarly, models of the arousal word scores achieved a very poor fit
forwards,
2
(2) = 28.15, p = .00, CFI = .4, SRMR = .42, but an extremely strong fit
backwards,
2
(2) = .77, p = .68, CFI = 1, SRMR = .03. Finally, path analyses
considering only affect word scores revealed a poor fit using the forwards model,

2
(2) = 24.63, p = .00, CFI = .47, SRMR = .141, and an good fit using the backwards
model,
2
(2) = 3.67, p = .16, CFI = .96, SRMR = .053. Diagrams of all of the path
52
analyses for the individual cognitive accessibility elements, with notation for beta
weights and statistical significance, are provided in the appendix.

Chapter 5: Discussion
In accordance with our hypotheses, the combination of a major Time 1
provocation and a minor Time 2 frustration magnified both the participants’ negative
evaluation of the experimenter and the number of words completed aggressively on
the word completion task. Further, as predicted, the provocation was responsible for
more activation of cognitions about aggressive action while the frustrating trigger
was responsible for more arousal, but both contributed to greater negative affect.
Though it was hypothesized that the provocation would produce more aggression,
general cognitive accessibility of aggressive constructs, and greater negative affect
compared to the control cell than would the trigger, we did not find a significant
result on a t-test when comparing either of these cells to the control for either
cognitive score. At the same time, both comparisons revealed differences in the
aggression scores. In all cases, the provocation came closer to significance, or was
significant at a higher level of confidence, than the trigger, though the differences
between these differences were never reliable.
These aggression results are mostly consistent with past research on the TDA
paradigm, the CNA model, and the GAAM, and parallel Vasquez et al.’s (under
review) similar findings on frustrating trigger-facilitated direct aggression.
53
Mirroring previous TDA research, the reliable interaction between the Time 1
provocation and the Time 2 trigger reflected disjunctively escalated aggression.
Contrary to expectations based on previous research and theory, we saw an
aggression effect for frustrating trigger and, also unexpectedly, an aggression effect
for provocation that did not differ reliably from the frustrating trigger.
Further, the results of the present study also build on the aforementioned lines
of research. The present study supports the idea that TDA can occur in response to
events other than the interpersonal provocations on which previous research has been
focused, and specifically in reaction to a non-arbitrary frustration as the trigger.
The present results indicate the possibility that not only is intent to frustrate
someone not necessarily required to incline them to aggress in a TDA situation, but
even inadvertently causing someone to be frustrated when simply following socially
acceptable rules might lead to such aggression. However, more research is needed to
verify that participants did not actually attribute intent to harm to the experimenter,
as the present results also leave open that alternative possibility.
Our results also support the theoretical idea presented within the CNA model
and the GAAM that frustration and provocation are functionally interchangeable; we
obtained similar results when using a minor frustration trigger as prior TDA
researchers obtained when using a minor interpersonal provocation as a Time 2
trigger, though the aggression effect we found for frustrating trigger was not found
for interpersonal triggers in previous research.
54
Excitation Transfer Theory and Triggered Displaced Aggression
On the surface, it may appear that excitation transfer theory could provide an
alternative explanation for our results. Excitation transfer theory claims that
previously-present physiological arousal will amplify aggression in response to a
provocation. One well-known example of excitation transfer theory is the “cycling
study” (Zillmann et al., 1974), in which participants exercised on a stationary bicycle
(causing physiological arousal), and then responded more aggressively to
provocation a few minutes later when compared to those who had not exercised
beforehand. This physiological arousal tends to persist for only about ten to twenty
minutes after the initial event, well within the bounds of the current study (Zillmann,
1988).
Though it may appear relevant at first glance, excitation transfer theory does
not apply to our current research, for two reasons. First, in order for excitation
transfer to occur, the person must be unaware of the arousal they are experiencing
(Zillmann, 1988). However, our experiment purposely brought these feelings of
arousal to the attention of the participants when asking their opinion on the
provoking article. Participants in the present research responded to requests for their
opinions on the article and the author by expressing their negative reactions, thus
making themselves aware of their own arousal resulting from the provocation. In
addition, participant’s scores on the word completion task suggested that they were
55
not only aroused, but were also experiencing thoughts about aggression-related
action as well as feeling anger-related affect. In light of these criticisms, it seems
unlikely that excitation transfer theory can adequately explain the results of the
present study.
Limitations
One limitation to our study is its inability to test the reasons why Hypothesis
3 was supported. Though an argument for the hypothesis was proffered, (that the
directly offensive nature of the provocation would activate more action-related
cognitions while the temporal proximity of the frustrating trigger to the measure of
cognitive accessibility would reveal more arousal attributable to the frustrating
trigger) there were no measures that would allow verification of that explanation.
The design can only support the idea that greater cognitive accessibility of arousal-
related constructs occurs in response to the trigger than to the provocation given the
proximity of each to the cognitive accessibility measure in our design. It cannot
provide any useful information about how or why this happens, meaning that taken
alone, the data from the present study does not support one theoretical explanation
over another in accounting for the effect that was found. Further research integrating
measures designed to explain this phenomenon in experimental designs specifically
designed for such a test would be helpful.
Another limitation of this study is that the selected a priori contrasts
involving the aggression scores revealed significant residual variance, indicating that
56
not all of the variance in the data was explained by our model. Since the model
tested by our contrast perhaps expected both greater response to the provocation but
certainly expected a lesser response to the frustrating trigger than the data show, the
residual variance is most likely due to those two cells. The lower-than-desired
scores seen on the provocation manipulation check might seem to account for the
lower-than-expected result in the provocation-only cell. However, since our contrast
model did fit the cognitive accessibility scores, this explanation seems unlikely.
These cognitive results support the view that our manipulations did have the desired
effect of activating cognitive-neoassociationistic networks of aggressive constructs,
but for some reason (and in accord with the outcomes of the path analyses) these
effects did not translate into aggression but instead, were its consequence.
The reason for the obtained deviation from the expected model seems most
likely to be due to the unexpected aggression produced by frustration alone. In
retrospect, it appears that in interpersonal trigger-based TDA paradigms, participants
recognize that their anger is directed at the triggering aggression target and
consciously inhibit retaliation against what they recognize as a minor attack. They
still feel bad, but do not allow these feelings to manifest in aggressive behavior. In
comparison, consistent with the assumption that provocations have a source who can
be personally blamed for negative feelings whereas non-arbitrary frustrations do not,
participants in the frustration only condition had no reason to inhibit the aggressive
inclinations resulting from the negative feelings they experienced as a result of the
57
trigger. Consequently, when later given opportunity to aggress against the
experimenter, no inhibitory mechanism was in an activated state and instead,
aggressive inclinations elicited by the frustration were then directed toward the
experimenter. This argument, however, is clearly post hoc and needs to be tested
more carefully – an enterprise beyond the scope of the present study.
The fact that the model in Figure 8 was such a good fit highlighted another
limitation of our study. Contrary to expectations, the path analyses suggested that it
is aggression that mediates the relationship between the aversive events (the
provocation and the frustrating trigger) and the cognitive accessibility of aggression-
related constructs, not vice-versa. It seems likely, however, instead of revealing
evidence counter to aggression theory, the path analysis reflected a flaw in the
methodology of the current study: cognitive accessibility of aggressive constructs
was only measured after the frustrating trigger rather than between the provocation
and the frustrating trigger. The attempt to demonstrate that the provocation and
frustrating trigger would interact to produce greater cognitive accessibility of
aggressive constructs (which requires a measurement at the time it was made),
inadvertently limited examination of any mediating relationship between cognitive
accessibility and aggression (which, conversely, benefits from a measurement made
between the provocation and the frustrating trigger). This obscures the initially
expected path of mediation, because theory assumes that it is the differences in
cognitive accessibility of aggressive constructs that are induced after the provocation
58
but prior to the trigger, that are key in producing TDA. Without knowing the
cognitive accessibility scores from this point in time, it is difficult to evaluate the
mediation of interest, since it is those scores, not the post-trigger accessibility scores,
which are expected to mediate aggression. At the same time, there is no way to be
sure that the accessibility as it was in fact measured did not precede the frustration.
Though the path analyses suggest the “backwards” path of mediation, measurements
at different times would be needed to come to a confident conclusion about the true
path of mediation.
Therefore, the results neither support nor oppose the possibility that the
accessibility of aggression-related constructs mediates the effect of aversive events
on aggression. Instead, they imply that cognitive accessibility, when measured after
the aversive events and around the same time as aggression is measured, will appear
to be influenced by one’s inclination to aggress, rather than solely and directly
affected by the events themselves. This lends credence to the CNA model, which
proposes that aversive events lead to immediate fight and flight responses, which in
turn activate the related cognitive neo-associationistic networks. Since appraisals are
not required in the CNA model; the accessibility of aggression-related cognitions
need not have much influence on the actual aggressive behavior that is displayed.
Though the obtained results do not provide the strong support for the GAAM that
would have been provided by evidence of the opposite path of mediation, they are
not necessarily inconsistent with the GAAM. The GAAM allows for one’s early
59
appraisals to further influence levels of anger-related affect, aggression-related
cognitions, and arousal during the “re-appraisal” process between initial appraisals
and aggressive action. Unfortunately, the implications of the results of our path
analysis are inconclusive due to the current study’s limited scope and measurements.
More specific and detailed research into the exact processes and their ordering,
conducted within a variety of paradigms and using a variety of measures, would be
required for any serious conclusions about the impact of our path analysis results on
aggression theorizing.
Further testing of the hypotheses using different methods is especially
important considering two weaknesses of the provocation manipulation. First, as
noted in the Results section, the provoking article created negative responses that
were not strong as intended (though it did elicit responses that reliably differed from
those to the neutral article). Second, despite efforts made to match the two articles in
appearance and word count, the neutral article still appeared to be longer than the
provoking article because the former contained more lengthy words.
Another limitation of the study was that though, as explained earlier, the
experimenter was operating within the bounds of the study protocol and gave no
reason for the participant to blame the experimenter for the frustration, it could be
argued that the reason participants aggressed against the experimenter was because
they viewed the experimenter as somehow responsible for the frustration. Further
research is needed to ascertain whether TDA-related aggressive actions will be
60
limited to being targeted towards someone who is in some way related to the cause
of the Time 2 frustrating trigger, or if the aggression would be readily directed at any
available target (perhaps even at the source of the provocation in a case of direct, as
opposed to displaced, aggression), regardless of their relation to the frustrating
trigger. Such a question was beyond the scope of the current study.
It would also aid us in our understanding of TDA and aggression processes in
general if future research delineated which attributes of the aggressor, target, and
situation facilitate the expression of TDA in response to a frustration and which
attributes repress attacks in this context. It seems likely that if there are any
characteristics of the target or the situation that are necessary for frustration-triggered
displaced aggression, they will be somewhat different from those necessary for
provocation-triggered displaced aggression, as frustrations have no explicit human
instigator and, in this case, conform to social rules. For instance, frustration-triggered
displaced aggression might be uniquely influenced by the group membership of the
target in a situation where aggressing against the target would not otherwise be
considered socially acceptable. This may be the case because an aggressor might
view aggressing against an outgroup member as (to some degree) socially acceptable
even in the absence of the objective justification that would be required otherwise.
Salience of group membership would likely moderate such an effect. Conversely, if
participants really do directly blame the experimenter for the frustration, group
salience might instead lead to a “black sheep effect” (Marques, Yzerbyt, & Leyens,
61
1988), in which the participant would rebuke an ingroup member more strongly than
an outgroup member in response to a transgression. Either way, since the retaliation
in a provocation-triggered displaced aggression situation would probably seem
socially acceptable to the aggressor in and of itself, these variables would likely not
influence provocation-triggered displaced aggression. The current research is
limited to explaining situations in which there exists an available opportunity to
aggress, in complete safety from retaliation, against a target who was in at least some
way involved in causing a non-arbitrary frustration.

62
References
Anderson, C. A., Anderson, K. B., & Deuser, W. E. (1996). Examining an affective
aggression framework: Weapon and temperature effects on aggressive
thoughts, affect, and attitudes. Personality and Social Psychology Bulletin,
22, 366-376.

Anderson, K. B., Anderson, C. A., Dill, K. E., & Deuser, W. E. (1998). The
interactive relations between trait hostility, pain and aggressive thoughts.
Aggressive Behavior, 24, 161-171.

Anderson, C. A., & Bushman, B. J. (2002). Human Aggression. Annual Review of
Psychology, 53, 27-51.

Anderson, C. A., Carnagey, N. L., Flanagan, M., Benjamin, A. J., Eubanks, J., &
Valentine, J. C. (2004). Violent video games: Specific effects of violent
content on aggressive thoughts and behavior. Advances in Experimental
Social Psychology, 36, 199-249.

Anderson, C. A., Deuser, W. E., & DeNeve, K. (1995). Hot temperatures, hostile
affect, hostile cognition, and arousal: Tests of a general model of affective
aggression. Personality and Social Psychology Bulletin, 21, 434-448.

Averill, J. R. (1982). Anger and aggression: An essay on emotion. New
York/Heidelber: Springer-Verlag.

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(5), 825-832.

Baron, R. A., Bell, P. A. (1977). Sexual arousal and aggression by males: Effects of
type of erotic stimuli and prior provocation. Journal of Personality and
Social Psychology, 35(2), 79-87.

Baron, R. A., Ransberger, V. M. (1978). Ambient temperature and the occurrence of
collective violence: The "long, hot summer" revisited. Journal of
Personality and Social Psychology, 36(4), 351-360.

Berkowitz, L., Donnerstein, E. (1982). External validity is more than skin deep:
Some answers to criticisms of laboratory experiments. American
Psychologist, 37(3), 245-257.

63
Berkowitz, L. (1983). The experience of anger as a parallel process in the display of
impulsive, “angry” aggression. In R.G. Geen & E.I. Donnerstein (Eds.),
Aggression: Theoretical and empirical reviews, 1. New York: Academic
Press.

Berkowitz, L. (1989). Frustration-aggression hypothesis: Examination and
reformulation. Psychological Bulletin, 106(1), 59-73.

Berkowitz, L., Heimer, K. (1989). On the construction of the anger experience:
Aversive events and negative priming in the formation of feelings.
Berkowitz, Leonard (Ed). Advances in Experimental Social Psychology, 22,
1-37, San Diego, CA, US: Academic Press, Inc.

Berkowitz, L. (1990). On the formation and regulation of anger and aggression: A
cognitive-neoassociationistic analysis. American Psychology, 45(4), 494-
503.

Bushman, B.J., Bonacci, A.M., Pedersen, W.C., Vasquez, E.A., & Miller, N. (2005).
Chewing on it can chew you up: Effects of rumination on triggered displaced
aggression. Journal of Personality and Social Psychology, 88(6), 969–983.

Buss, A. H. (1963). Physical aggression in relation to different frustrations. Journal
of Abnormal & Social Psychology, 67(1), 1-7.

Buss, A. H. (1966). Instrumentality of aggression, feedback, and frustration as
determinants of physical aggression. Journal of Personality and Social
Psychology, 3(2), 153-162.

Carlsmith, J. M., Anderson, C. A. (1979). Ambient temperature and the occurrence
of collective violence: A new analysis. Journal of Personality and Social
Psychology, 37(3), 337-344.

Crook, K., Beaver, B. R., Bell, M. (1998). Anxiety and depression in children: A
preliminary examination of the utility of the PANAS-C. Journal of
Psychopathology and Behavioral Assessment, 20(4), 333-350.

Denson,T., Pedersen,W.C., & Miller, N. (2006). The Displaced Aggression
Questionnaire. Journal of Personality and Social Psychology, 90, 1032-
1051.

Dill, J.C. & Anderson, C.A. (1995). Effects of frustration justification on hostile
aggression. Aggressive Behavior, 21(5), 359-369.
64
Dollard, J., Doob, L.W., Miller, N.E., Mowrer, O.H., & Sears, R.R. (1939).
Frustration and Aggression. Oxford: Yale University Press.

Gustafson, R. (1989). Frustration and successful vs. unsuccessful aggression: A test
of Berkowitz’ completion hypothesis. Aggressive Behavior, 15(1), 5-12.

Loew, C.A. (1967). Acquisition of a hostile attitude and its relationship to
aggressive behavior. Journal of Personality & Social Psychology, 5(3), 335-
341.

Lyubomirsky, S. & Nolen-Hoeksema, S. (1995). Effects of self-focused rumination
on negative thinking and interpersonal problem solving. Journal of
Personality & Social Psychology, 69(1), 176-190.

Marques, J. M., Yzerbyt, V. Y., & Leyens, J.-P. (1988). The black sheep
effect: Judgmental extremity towards ingroup members as a function of
group identification. European Journal of Social Psychology, 18, 1–16.

Miller, N., & Marcus-Newhall, A. (1997). A conceptual analysis of displaced
aggression. In R. Ben-Ari & Y. Rich (Eds.), Enhancing education in
heterogeneous schools: Theory and application (pp. 69-108). Ramat-Gan,
Israel: Bar-Ilan University Press.

Miller, N., Pedersen, W.C., Earleywine, M., & Pollock, V.E. (2002). A Theoretical
Model of Triggered Displaced Aggression. Personality & Social Psychology
Review, 7(1), 75-97.

Pastore, N. (1952). The role of arbitrariness in the frustration-aggression hypothesis.
Journal of Abnormal Psychology, 47, 728-731.

Pedersen, W.C., Gonzales, C., &Miller, N. (2000). The moderating effect of trivial
triggering provocation on displaced aggression. Journal of Personality &
Social Psychology, 78(5), 913-927.

Pedersen, W.C., Bushman, B.J., Vasquez, E.A., & Miller, N. (2001). The
Moderating Effect of Target Attributes on Triggered Displaced Aggression.
Manuscript submitted for publication, . Journal of Personality & Social
Psychology.
65
Rusting, C.L. & Nolen-Hoeksema, S. (1998). Regulating responses to anger: Effects
of rumination and distraction on angry mood. Journal of Personality &
Social Psychology, 74(3), 790-803.

Sandin, B., Chorot, P., Lostao, L., Joiner, T. E., Santed, M.A., Valiente, R.M.
(1999). The PANAS Scales of Positive and Negative Affect: Factor analytic
validation and cross-cultural convergence. Psicothema, 11(1), 37-51.

Vasquez, E. A., Bartsch, V. O., Pedersen, W. C., Miller, N. (under review). The
Impact of Aggressive Priming, Rumination, and Frustration on Prison
Sentencing. European Journal of Social Psychology.
Vasquez, E .A., Denson,T .F., Pedersen.W .C., Stenstrom, D .M., & Miller, N.
(2005). The moderating effect of trigger intensity on triggered displaced
aggression. Journal of Experimental Social Psychology, 41, 61-67.

Watson, David, Clark, Lee A., Tellegen, Auke (1998). Development and validation
of brief measures of positive and negative affect: The PANAS scales.
Journal of Personality & Social Psychology 54(6), 1063-1070.

Worchel, S. (1974). The effect of three types of arbitrary thwarting on the
instigation to aggression. Journal of Personality, Vol 42(2), 300-318.

Zillmann, D., Bryant, J. (1974). Effect of residual excitation on the emotional
response to provocation and delayed aggressive behavior. Journal of
Personality and Social Psychology, 30(6), 782-791.

Zillmann, D., Bryant, J., Comisky, P. W., Medoff, N. J. (1981). Excitation and
hedonic valence in the effect of erotica on motivated intermale aggression.
European Journal of Social Psychology, 11(3), 233-252.
66
Appendix
[Neutral article: Edited, from an article that originally appeared in the Daily
Trojan, for neutrality and length and to match the formatting of the provoking
article. Please note that the author’s name and title, as well as the writer credit at
the end, were blacked out with a marker on the hard copies shown to participants.]

Clever Flavor to New Exhibit
You don't need the glint of Egyptian treasure or the
shock value of an ancient mummy to draw a massive
crowd through museum doors. In fact, you may be
able to entice more than 200,000 visitors to an exhibit
featuring a 66-million-year-old Triceratops skull, a red
1970s Winnebago and a makeshift Mayan temple -- all
in close range of each other -- with a little ingenuity,
and a challenge to viewers to read between the lines.
"Collapse?" a new exhibit at the Los Angeles
Natural History Museum, provides visitors with more
than a walk through of an object-littered marble hall.
While it does host an intriguing buffet of mixed-media
eye candy, the exhibition dares viewers to think about
the fate that modern society faces by taking a journey
through the past to see how man's choices brought
about the successes and decline of civilizations in
different eras across the globe.
Inspired by Pulitzer Prize winning author Jared
Diamond's best selling book "Collapse: How societies
Choose to Fail or Succeed", the exhibit takes museum
visitors on a journey that starts in modern-day
Montana, and then travels backward in time through
Mayan civilization and Tokugawa, Japan before
returning to the contemporary scene in Australia and
Southern California.
“Collapse?" is intended to test new approaches to
learning, said Vitali. She also said the exhibit is meant
to be visited more than once, as it will remain at the
Natural History Museum until next January.

“The exhibit is designed to develop a new language of
natural history that is more appropriate to our times,
particularly for a city like Los Angeles, that bathes in
new media and contemporary art," said Vitali, who
compared the creative process to an orchestra, with
herself as the conductor. "It involves a lot of creative
minds from the city.”
The behind-the-scenes process was a particular
challenge for Vitali and her crew because Diamond had
not yet finished his book when plans for the exhibit
began.
In the book-to-exhibit shift in medium, the exhibit
staff went beyond adapting some of Diamond's ideas,
expanding and extending the context to suit a 3-D form.
Working with the Pulitzer Prize-winner was a
particularly special treat for the staff.
"After the concept was defined, he left us museum
people to determine how the exhibit would look and
feel. As a great professional, he respects other
professionals," Vitali said.
"Collapse" will continue until Jan. 31, 2006.
Admission to the Los Angeles National History
Museum is free for USC students.

W Staff Writer Mitch Vernon is a sophomore majoring in
business management. He can be reached at
dtrojan@usc.edu or at (213) 740 5665.

Mitch Vernon
Staff Writer
67
[Provoking Article. Please note that the author’s name and title, as well as the
writer credit at the end, were blacked out with a marker on the hard copies shown to
participants

Stop complaining about tuition
It’s that time of year again: time for the annual
announcement that – surprise! – tuition is going up! It’s
time to begin the annual chorus . . . of whining.
It seems like each year, everybody is worried and up
in arms about the annual hike in the tuition. Stop
complaining about it!
Yeah, sure tuition has gone up. Yeah, every year the
amount that tuition goes up gets higher and higher (the
raise is pushing the $2000 dollar mark this year). So
what?
Sure, $2000 extra seems like a lot at first. But think
about how much you are paying for school already. Not
even counting room and board, the tuition increase
comes to about 7 percent more money than you were
paying. That’s barely even a tuition hike when you take
into account the fact that the inflation rate has been
around 3 percent this year.
Besides, most of the students here at USC come from
rich families anyway. Don’t worry, mommy and daddy
will still pay for gas for your BMW even though tuition
went up.
Yet every year, there is the same report on the tuition
increase in the paper, without fail only barely managing
to avoid veering out of the realm of journalism and into
the realm of opinion. This is unavoidably followed by a
torrent of letters complaining about the tuition increase,
as though we are all shocked

and appalled at an event we all knew was coming
anyway. It happens every year. You can handle it.
Get over it!
You pretty much sign on for tuition increases as part
of the deal when you enroll, whether you realize it or
not. All whining does is annoy the rest of us who
understand that this is a fact of life and it’s time to just
move on.
Tuition goes up every year, and tuition will continue
to go up every year. It is clear that school
administrations have no plans to halt or lessen the
increases in tuition, no matter how much students
whine. It’s not like you can just decide to go to a
school that doesn’t raise tuition; they all do it. You are
stuck paying more tuition every year, so you should all
just learn to grin and bear it.
So step away from the keyboards, shut your mouths,
and stop whining so we can all get on with our lives in
peace.

W Editorialist Mitch Vernon is a sophomore majoring
in business management. He can be reached at
dtrojan@usc.edu or at (213) 740 5665.
Mitch Vernon
Editorial Writer
68
[Demographics and provocation manipulation check]

Opinion Questionnaire

Directions: Please fill in the following:

Age = _____________

Gender = _________________

Major = _________________

What is your native language? ____________________

What is your ethnic/racial background? ___________________

What is your political orientation? ___________________

Opinion: Please write a brief response in the space under each question

1. What is your opinion about the subject of the article you read?

2. What do you think of the author of the article?
69
[Measure of Cognitive Accessibility of Aggressive Constructs]

Word Completion Task
Directions: Below is a list of words with missing letters. Please fill in
the missing letters to form a word. All of the items can be made into
several possible words so please just make the first word that comes to
mind.
Please complete this task as fast as you can.
(1) s h o _ t (16) m _ d (31) d i s g _ _ _ e d

(2) u _ _ e _ (17) a t t _ c _ (32) s _ _ b

(3) _ r _ _ s e d (18) j _ _ _ y (33) _ e l l

(4) h a _ e (19) s l _ p (34) _ _ _ a g e

(5) p r o v _ _ e (20) _ _ a t e

(6) _ _ t t e r y (21) t e _ s _

(7) a l _ _ _ (22) a _ _ _ _ e d

(8) o u t _ _ _ e (23) l _ _ e l y

(9) i n s _ _ _ (24) f i _ _ t

(10) k i _ _ (25) i n _ e n _ e

(11) e x _ _ _ e d (26) o f f _ _ _

(12) _ _ m i n g (27) _ e e _ e d

(13) a n g _ _ (28) h _ t

(14) h _ r _ (29) _ _ t t e r

(15) ac_ _ _ _ (30) i n _ _ r e
70
[Measure of Aggression]

Important Instructions (PLEASE READ THE FOLLOWING)

The USC Psychology Department has initiated a new procedure for the hiring
of Student Research Assistant (SRA) positions. As you can imagine, these positions
are very attractive because they look good on a resume and they offer a reasonable
salary. Because the Psychology Department has received more applicants than the
number of STA positions available, we have decided to gather input from the
participants in research studies to help us make our hiring decisions. The
Experimenter conducting your study today is a Student Research Assistant applicant
and we would like you feedback (on the 5 questions below) regarding this
individual’s suitability for the SRA position. We would ask you to please take this
task seriously.

Experiment: Experiment #74
SRA Applicant: #5
Directions: Please use the 7-point scale below to individually answer the following
statements. Please do not write your name, student ID number, or any other
identifying mark on this sheet since we want the answers to be completely
anonymous.
1---------------2---------------3--------------4----------------5--------------6---------------7
strongly moderately slightly neither agree slightly moderately strongly
disagree disagree disagree nor disagree agree agree agree

1. I recommend this experimenter to be a Student Research Assistant. _____________
2. I think that the experimenter is a competent individual. _____________
3. I like the experimenter. _____________
4. I think that the experimenter is friendly. _____________
5. The experimenter is intelligent. _____________
71
[Adjective Checklist for Frustration Manipulation Check]

Instructions: Please write in the number next to each of the following words that
best describes how much you are experiencing that feeling or mood on the following
scale:

1---------------2---------------3--------------4----------------5--------------6---------------7
Not at Very little slightly moderately strongly very strongly extremely
all so so so so so so

______ Angry ______ Down ______ Cheerful

______ Anxious ______ Distressed ______ Uncertain

______ Fearful ______ Playful ______Hostile

______Sad ______Scornful ______Grouchy

______Frustrated ______Regretful ______Avoidant

______Confused ______Curious ______Annoyed

______Pleased ______Disgusted ______Vigorous

______Upset ______Offended ______Happy

______Competent ______Irritable ______Guilty

______Concentrating

72
[Table of means and standard deviations for all frustration manipulation check]
Table 3: Means and Standard Deviations of the responses for all words from the frustration
manipulation check, by condition.
Emotion Frustration Mean SD Emotion Frustration Mean SD
Frustrated yes 3.78 1.72 Scornful yes 1.11 0.33
no 1.00 0.00 no 1.00 0.00

Down yes 2.00 0.87 Grouchy yes 2.00 1.41
no 1.00 0.00 no

Distressed yes 2.33 1.12 Regretful yes 1.78 0.97
no 1.27 0.65 no 1.55 1.21

Uncertain yes 4.44 1.88 Avoidant yes 1.56 0.88
no 2.45 1.37 no 1.55 1.21

Annoyed yes 3.22 1.86 Confused yes 2.56 1.42
no 1.00 0.00 no 2.18 1.25

Concentrating yes 4.33 1.32 Curious yes 4.56 1.51
no 2.00 1.15 no 4.6 2.01

Angry yes 1.89 1.36 Pleased yes 1.5 0.53
no 1.36 0.81 no 1.55 0.82

Cheerful yes 2.33 1.22 Disgusted yes 1.11 0.33
no 3.09 1.70 no 1.36 0.81

Anxious yes 2.78 1.72 Vigorous yes 1.44 5.27
no 1.91 2.02 no 1.18 0.40

Fearful yes 1.33 0.50 Upset yes 1.78 1.64
no 1.00 0.00 no 1.36 0.81

Playful yes 2.25 1.16 Offended yes 1.56 0.88
no 3.27 2.28 no 1.00 0.00

Hostile yes 1.33 0.71 Happy yes 2.33 1.41
no 1.00 0.00 no 2.91 1.51

Sad yes 1.67 1.32 Competent yes 2.89 1.69
no 1.55 1.21 no 2.55 1.81

Irritable yes 2.78 1.48 Guilty yes 1.00 0.00
no 1.55 1.21 no 1.18 0.40
73
[Diagrams of path analyses separating the three cognitive accessibility elements
(action, affect, and arousal).]

Figure 9: Diagrams, including beta weights and statistical significance, of path analyses
involving the cognitive accessibility of action words and aggression, first “forwards”, then
“backwards”.

.33*

.27*
.05*

.05*

.45*

.27*
.05*

.42*

Aggression

Frustrating
Trigger
Provocation
Action
Words
Action
Words
Frustrating
Trigger
Provocation
Aggression
74
Figure 10: Diagrams, including beta weights and statistical significance, of path analyses
involving the cognitive accessibility of affect words and aggression.

.32*

.38*
.05*

.23*

.45*

.38*

.05*

.42*

Aggression
Frustrating
Trigger
Provocation
Affect
Words
Affect
Words
Frustrating
Trigger
Provocation
Aggression
75
Figure 11: Diagrams, including beta weights and statistical significance, of path analyses
involving the cognitive accessibility of arousal words and aggression.

.11*

.43*
.05*

.25*

.45*

.43*
.05*

.42*

Aggression
Frustrating
Trigger
Provocation
Arousal
Words
Arousal
Words
Frustrating
Trigger
Provocation
Aggression

Abstract (if available)

Abstract Tested the hypotheses, consistent with previous research on Triggered Displaced Aggression (Miller et al. 2003) and the predictions of both the Cognitiveneoassociationistic model (Berkowitz, 1983) and the General Affective Aggression Model (Anderson et al. 1996), that a major provocation followed by a minor frustrating triggering event would result in disjunctive escalation of aggression greater than the additive effects of the two events experienced separately as well as similar magnification of the cognitive accessibility of aggressive constructs. Both hypotheses were confirmed, supporting the theoretical ideas that frustrations and provocations are functionally equivalent and that Triggered Displaced Aggression occurs when a Time 2 trigger follows a Time 1 provocation. The path of mediation between accessibility of aggression-related cognitions and aggressive inclinations is also discussed.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The moderating effect of triggering provocation on direct aggression

PDF

Children's triggered displaced aggression

PDF

The angry brain: neural correlates of interpersonal provocation, directed rumination, trait direct aggression, and trait displaced aggression

PDF

The impact of naturally-occurring rumination on aggression

PDF

Implicit theory ideology about human attributes and aggression

PDF

Projection on outgroups: complementary projection in aversive intergroup contexts

PDF

Personality and cognitive aging: are positive traits protective?

PDF

The home environment and Latino preschoolers' emergent literacy skills

PDF

Couples’ neuroendocrine activity in response to family conflict discussions: the role of self-reported anger and previous marital aggression

PDF

Within session therapist behavior in multisystemic therapy

PDF

Evaluating a cultural process model of depression and suicidality in Latino adolescents

PDF

Assessing therapeutic change mechanisms in motivational interviewing using the articulated thoughts in simulated situations paradigm

PDF

Underlying neural mechanisms of depression and dementia

PDF

Age-related androgen depletion and the development of Alzheimer's disease

PDF

Age-related changes in cognitive functioning and their implications for the reacquisition of conditioned taste avoidances

PDF

Dynamic analyses of the interrelationship between mothers and daughters on a measure of depressive symptoms

PDF

Effectiveness of a heart plus brain health-focused nutrition intervention among Latinas: a randomized controlled trial of Buenos Hábitos Alimenticios Para Una Buena Salud (Good Eating Habits for ...

PDF

Health-related quality of life correlates of the treatment decision making process of newly diagnosed prostate cancer patients

PDF

Examining the effect of fertility on lifespan using data from the Swedish Twin Registry

PDF

Social support, constraints, and protective buffering in prostate cancer patients and their partners

Asset Metadata

Creator Lamoreaux, Jeffrey J. (author)

Core Title Frustration within the triggered displaced aggression paradigm

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychology

Publication Date 12/02/2006

Defense Date 09/07/2006

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aggression,frustration,OAI-PMH Harvest,triggered displaced aggression

Language English

Advisor Miller, Norman (committee chair), Gatz, Margaret (committee member), Lickel, Brian (committee member)

Creator Email jefflamx@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m210

Unique identifier UC1460382

Identifier etd-Lamoreaux-20061202 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-157307 (legacy record id),usctheses-m210 (legacy record id)

Legacy Identifier etd-Lamoreaux-20061202.pdf

Dmrecord 157307

Document Type Thesis

Rights Lamoreaux, Jeffrey J.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu