University of Southern California Dissertations and Theses

The impact of naturally-occurring rumination on aggression

(USC Thesis Other)

The impact of naturally-occurring rumination on aggression

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

THE IMPACT OF NATURALLY-OCCURRING
RUMINATION ON AGGRESSION
by
Cathryn Ashley Borders

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)

May 2007

Copyright 2007 Cathryn Ashley Borders

ii

Dedication

For Jen

and in memory of Russ
iii

Acknowledgements
I would like to thank the following students for their help with data collection, coding,
and data entry: Andrew Gonzalez, Lindsay Dye, Desiree Barnett, and Kirsten Lander. I
would also like to thank Mitch Earleywine for his tireless feedback and encouragement,
and Norman Miller for suggesting much of the methodology used in these studies.
Thank you to my committee: Mitch Earleywine, Norman Miller, Stan Huey, Gayla
Margolin, and John Brekke. Finally, love and thanks to my friends and family without
whom I wouldn’t have wanted to go through this.
iv

Table of Contents

Dedication ii

Acknowledgements iii

List of Tables v

List of Figures vi

Abstract vii

Introduction 1

Study 1 12
Method 14
Results 23
Discussion 45

Study 2 55
Method 58
Results 63
Discussion 87

General Discussion 98

References 107

v
List of Tables

Table 1. Correlations between Study 1 variables for entire sample (N=95). 24
Table 2. Correlations between Study 1 variables by mood condition. 25
Table 3. Correlations between Study 1 variables by gender. 44
Table 4. Correlations between Study 2 variables, for entire sample (N=60). 64
Table 5. Correlations between rumination and Study 2 variables by 65
mood condition.

Table 6. Correlations between rumination and Study 2 variables by gender. 85

vi
List of Figures

Figure 1. Procedure for Study 1. 21

Figure 2. The impact of mood manipulations on state rumination (Study 1). 28

Figure 3. Mood indices mediate the effect of mood manipulations on 34
state rumination (Study 1).

Figure 4. Accessibility of arousal words mediates the link between angry mood 40
and hostile attributions (Study 1).

Figure 5. State rumination mediates the link between angry mood and 42
accessibility of arousal words, for participants in the sad condition (Study 1).

Figure 6. Procedure for Study 2. 60

Figure 7. The impact of mood manipulations on state rumination (Study 2). 69

Figure 8. Mood indices mediate the effect of mood manipulations on 71
state rumination (Study 2).

Figure 9. Angry mood at Time 2 mediates the link between state rumination 81
and aggression measures (Study 2).

Figure 10. Evaluation of partner mediates the link between state rumination 83
and behavioral aggression (Study 2).
vii

Abstract
Experimental studies show that manipulated rumination increases aggressive responding
to perceived provocations. This study expands on previous work by instead manipulating
negative mood and measuring naturally-occurring rumination and subsequent aggression.
A pair of studies found that both angry and sad moods increased self-reported rumination,
compared to neutral mood. Moreover, angry mood generally led to increased aggression.
Study 1 failed to show an association between rumination, hostile attributions, and
aggressive intentions. However, Study 2 found robust correlational links between
increased rumination and aggressive responding to a mild trigger. Furthermore, although
rumination led to increased angry and sad moods following a mild trigger, only angry
mood mediated the association between rumination and increased aggression.
Interpretations and implications of these results are discussed.

1
Rumination is associated with a wide range of mental disorders. Formally defined
as repetitive, uncontrollable thoughts about negative internal or external experiences
(Ingram, 1990; Martin & Tesser, 1996), rumination involves harping on something
negative, seemingly without end or control. These repetitive thoughts often focus on
current feelings, related causes, consequences, and potential solutions (Lyubomirsky &
Nolen-Hoeksema, 1995). For instance, individuals might become fixated on thoughts
about why they can’t cope with things, why their partner treated them badly, or how they
will perform in an upcoming speech.
Several important reviews and meta-analyses highlight the connection between
rumination, depression and anxiety (Ingram, 1990; Mor & Winquist, 2002; Papageorgiou
& Wells, 2004; Pyszczynski & Greenberg, 1987; Segerstrom et al., 2003). In addition,
recent research documents important connections between rumination, anger, and
aggression. However, no reviews of this literature exist and only a few researchers have
proposed relevant theories (Miller et al., 2003; Pedersen et al., 2005). Given the current
prevalence of aggressive crimes and bullying and the emotional consequences of
violence, a better understanding of how rumination exacerbates aggression seems
especially relevant to the mental health field. In the following sections, I first review the
literature linking rumination to anger and aggression. I then turn to a discussion of some
important unresolved questions in the literature.
Rumination, anger, and aggression
Recent studies support the association between rumination and anger. Rusting and
Nolen-Hoeksema (1998) conducted a set of experiments investigating the effect of self-

2
focused rumination on angry mood. They viewed anger-related rumination as self-focus
relating to the symptoms, causes, and consequences of anger. Participants underwent an
angry mood induction and then participated in either rumination or distraction
manipulations. The rumination manipulation involved focusing attention on thoughts
that were emotion- and self-focused but not explicitly about anger (e.g., “why people
treat you the way they do”). In the distraction condition, participants focused their
attention on non-emotional thoughts (e.g., “the layout of the local post office”).
Following these manipulations, participants completed stories that were later rated for
anger content. In all experiments, rumination increased anger, whereas distraction
decreased or had no effect on pre-existing anger.
Similarly, Mischel and colleagues wrote about “hot” versus “cool” attentional
focus (Ayduk, Mischel, & Downey, 2002; Metcalfe & Mischel, 1999). The former
involves directing attention to the arousing and emotional aspects of situations, whereas
“cooling” strategies include distraction and cognitive reflections. In one experiment,
participants were asked to recall a significant and hurtful rejection from their past. They
then focused their attention on either the physiological and emotional reactions associated
with the experience (hot focus), or on the characteristics of the physical setting of the
experience (cool focus). All participants then reported on their mood and thoughts during
the experiment and completed a lexical decision task measuring reaction times to hostility
words. Both the explicit and implicit dependent measures indicated that hostile thoughts
and angry feelings were more accessible in the hot-focus condition than in the cool-focus
condition.

3
Several studies have also found that rumination increases hostile or vengeful
thoughts. Measures of both general rumination (McCullough et al., 1998; 2001) and
anger rumination (Barber, Maltby, & Macaskill, 2005) correlate with increased
vengefulness, or the disposition to seek revenge after suffering an interpersonal offense.
Moreover, people who ruminated more about a particular offense showed less progress
towards forgiveness eight weeks later (McCullough et al., 2001). Similarly, Caprara and
colleagues (1985, 1986) found that high dispositional ruminators exhibited more hostility
and deliberated more over thoughts of retaliation following a hostile experience than did
low ruminators (Caprara, 1985; see also Pedersen et al., 2005). Moreover, high
ruminators made less favorable evaluations following an insult than did low ruminators
who were also insulted (Caprara, 1985).
Researchers have also examined the impact of rumination on various forms of
aggression. In a quasi-experiment, Collins and Bell (1997) randomly assigned low
ruminators and high ruminators to provocation or control conditions. Provocation
consisted of negative feedback about performance on a memory task. Participants then
competed with an “opponent” to score the most number of points. Participants could
choose to increase their own number of points (by pressing one button), deduct points
from their opponent (by pressing another button), or deliver a loud white noise to the
opponent (by pressing still another button). The total number of noise blasts delivered to
the opponent constituted the aggression measure. Results indicated that high ruminators
in the provocation condition exhibited more aggression than low ruminators in the same
condition.

4
Bushman (2002) also investigated whether rumination while venting exacerbates
angry mood and aggression. Participants received negative feedback from a partner and
then underwent rumination, distraction, or control manipulations. Participants in the
rumination condition were instructed to hit a punching bag as many times as they wanted
(venting) and at the same time to think about the partner who criticized their work
(rumination). By contrast, participants in the distraction condition focused their thoughts
on becoming physically fit while they hit the punching bag. Finally, participants in the
control condition sat quietly for several minutes. After completing a mood adjective
checklist, participants engaged in a competitive reaction time task with a supposed
partner. In this task, opponents compete to press a button the fastest, and whoever is
slower receives a blast of loud noise. Participants choose how loud of a blast their
opponent would receive, over 25 trials. Results indicated that rumination while venting
increased both anger and aggressive behavior. Unfortunately, the design did not include
a pure rumination condition, so effect of cognitive rumination could not be separated
from the effect of hitting a bag.
Several studies investigated the impact of rumination on triggered displaced
aggression (Bushman, 2002; Bushman et al., 2005; Pedersen et al., 2005). Triggered
displaced aggression refers to instances in which a person who is provoked but cannot
retaliate directly against the source of provocation subsequently becomes aggressive with
a person who merely provides a trivial annoyance or provocation (Pedersen, Gonzalez, &
Miller, 2000). In two studies (Bushman et al, 2005), participants received negative
feedback and then ruminated or distracted themselves for 25 minutes, with a

5
manipulation similar to that used by Rusting and Nolen-Hoeksema (1998). As an
additional control, a third group of participants wrote about a happy incident in their
lives. Finally, all participants played a trivia game for which questions were read by
either a clumsy (trigger condition) or a composed (no-trigger condition) partner. In the
trigger condition, as well, participants were told that they did poorly compared to the
average student. Participants’ evaluations of the partner’s qualifications for a highly
coveted research job constituted the measure of aggression. Results indicated that
provoked participants who ruminated were more aggressive after a triggering event than
were those in the distraction and positive mood conditions. Rumination did not increase
aggression in the absence of a trigger. Distraction and positive mood did not
differentially impact triggered displaced aggression. Finally, provocation-induced
negative affect correlated with aggression only for participants in the rumination
condition.
Pedersen and colleagues (2005) compared the effects of self-focused versus
provocation-focused rumination on the accessibility of aggression-related words,
triggered displaced aggression, and physiological arousal. After receiving negative
feedback, participants either ruminated about themselves (self-focused rumination) or
about what had occurred so far in the experiment (provocation-focused rumination).
Results indicated that provocation-focused rumination increased hostility and the
accessibility of action words (e.g., hit, attack, shoot), whereas self-focused rumination
only differentially increased the accessibility of arousal words (e.g., aroused, tense).
Both forms of rumination increased the accessibility of anger-related words (e.g., mad,

6
upset, angry). Interestingly, only provocation-focused rumination increased
physiological arousal. Finally, high trait ruminators who engaged in provocation-focused
rumination generally exhibited more triggered displaced aggression than individuals who
engaged in self-focused rumination. One interpretation of these findings is that because
self-focused rumination turns individuals’ attention inward, individuals become more
aware of feelings and bodily sensations (Pedersen et al., 2005). However, this
introspective rumination does not necessarily affect the desire for aggressive behavior or
increase physiological arousal. Provocation-focused rumination, by contrast, involves a
more external focus on the provocation incident and likely involves retaliatory thoughts.
This form of rumination increases hostility, aggressive intentions, and arousal.
Moreover, the hostility that results from provocation-focused rumination may lead to
increased aggression.
In sum, research suggests that rumination exacerbates angry mood, increases the
accessibility of hostile thoughts, interacts with provocation to elicit more aggressive
behaviors, and may increase physiological arousal. The processes by which rumination
causes aggression are still being explored. Network models propose that mood-congruent
information is organized in long-term memory around central nodes (Berkowitz, 1990;
Bower, 1981). These associative networks link together related memories, thoughts,
feelings, and behavioral tendencies. Activation of one part of an associative network
(e.g., a particular negative emotion) can activate other items stored in that network.
Miller and colleagues (2003) suggest that repetitive thoughts and elaborations about a
provocation may maintain the activation of anger-related associative networks over time,

7
making angry and aggressive thoughts more accessible and subsequent aggressive
behavior more likely.
Some of the evidence discussed above supports this theory. Angered participants
made to ruminate became more angry and exhibited more aggressive responses following
a minor trigger than did participants who distracted themselves (Bushman et al., 2002;
Rusting & Nolen-Hoeksema, 1998). Moreover, in the study described above, hostility
mediated the association between rumination and aggression for high trait ruminators
who engaged in provocation-focused rumination (Pedersen et al., 2005). Specifically,
provocation-focused rumination led to increased accessibility of aggressive concepts and
aggressive behavior, compared to self-focused rumination. Thus, angry rumination may
activate aggressive associative networks, exacerbating existing angry mood and hostile
cognitions.
Finally, little is known about whether gender differences in the associations
between rumination and aggression. Depression researchers have found that women
generally ruminate more than men and that this difference accounts for women exhibiting
more depression than men (Butler & Nolen-Hoeksema, 1994; Nolen-Hoeksema, Parker,
& Larson, 1994). However, few aggression researchers have examined gender
differences in rumination. We found that rumination leads to increased alcohol-related
aggression for women but not for men (Borders, Smucker Barnwell, & Earleywine, In
Press). By contrast, Bushman and colleagues (2005) found no gender differences in
aggression for participants who received a rumination manipulation. Thus, the limited
evidence that exists on gender differences in anger rumination is mixed. In fact, studies

8
examining whether men exhibit more aggression than women after being provoked are
mixed (for a review, see Bettencourt & Miler, 1996). Rumination researchers should
start investigating gender differences in the association between rumination and
aggression.
Naturally-occurring rumination
Unfortunately, no laboratory studies have examined the impact of naturally-
occurring rumination on aggression. All of the experiments just reviewed first induced
rumination and then examined subsequent anger, hostility, and aggression. The
manipulations ranged from asking participants to think about their emotions and what
kind of people they are to having participants hit a punching bag and think about how
they had been treated by a partner. Research on the actual content of angry ruminative
thought does not yet exist. Thus, all the outcomes discussed above must be understood to
have occurred in the presence of experimenter-defined rumination. A better
understanding of what naturally leads to rumination is needed.
Miller and colleagues (2003) theorize that aggressive rumination likely arises
from anger. Indeed, most of the above experiments placed the rumination manipulation
after an angry mood manipulation, in the form of either a provocation or reflecting on a
negative memory. However, if rumination naturally follows from angry mood, then
manipulating both angry mood and rumination likely intensifies any naturally-occurring
rumination. Thus, the outcomes obtained in these laboratory studies may not reflect what
actually occurs for most people. Moreover, no experiment has actually tested the

9
hypothesis that rumination naturally arises from an angry mood. This is an important gap
in the field that should be addressed.
However, rumination may also naturally arise from sad mood. In the responses
style theory of depression, Nolen-Hoeksema (1991) proposed that depressed individuals
ruminate about past failures, current depressive feelings, and possible reasons for their
mood. A vast literature links rumination to sadness and depression (see Papageorgiou &
Wells, 2004). However, I know of no laboratory experiments that have only manipulated
sad mood and measured subsequent repetitive thought.
If both angry and sad moods give rise to rumination, how then does this naturally-
occurring rumination relate to aggression? Will rumination resulting from sadness show
similar effects on aggression as rumination resulting from anger? Existing research
suggests competing answers to these questions.
Emotion studies find that anger differs from sadness in important ways that may
contribute to different ruminative thoughts and subsequent outcomes. Specifically, anger
elicits approach-related, aggressive inclinations (Kenworthy et al., 2003; Tiedens, 2001),
whereas sadness typically elicits avoidance of interaction with others and decreased
initiative (Carlson & Miller, 1987; Dickson & MacLeod, 2004; Green & Sedikides,
1999). In research on learned helplessness, individuals who respond with anger to failure
at one set of problems subsequently show increased effort on a second, unrelated set of
problems (Mikulincer, 1988). By contrast, individuals who respond with depression
subsequently show decreased performance on the second set of problems (Mikulincer,
1988). In brain imaging research, anger activates the approach motivation regions of the

10
prefrontal cortex, whereas sadness activates the withdrawal regions (Harmon-Jones &
Sigelman, 2001). Moreover, several studies suggest that dysphoric participants who
ruminate show less willingness to problem-solve, implement solutions, and engage in
pleasant activities, even though they believe the solutions and pleasant activities would
be beneficial (Lyubomirsky, Tucker, Caldwell, & Berg, 1999; Nolen-Hoeksema &
Jackson, 2001; Nolen-Hoeksema et al., 1994; Nolen-Hoeksema et al., 1993). By
contrast, provocation-focused rumination increased physiological arousal (Pedersen et al.,
2005). In fact, some models of emotion suggest that anger is a high-activation negative
emotion, whereas sadness is a low-activation emotion (Larsen & Diener, 1992). These
results suggest that only anger-related rumination elicits aggressive-related thoughts,
setting the stage for aggressive responding. Conversely, sadness-related rumination may
elicit withdrawal-related thoughts, which should not lead to aggressive responding.
A contrary argument could also be made, however. Some research suggests that
rigid, repetitive thinking may be the common process behind all types of rumination.
Segerstrom and colleagues (2000) found that repetitive thinking explained the common
variance in measures of both depressive rumination and worry. Specifically, the shared
variance of worry and depressive rumination overlapped 64% in students and 99% in
patients who reported global repetitive thinking. Furthermore, these specific forms of
rumination related to pre-existing anxiety and depression only indirectly, through a
repetitive-thought latent variable. In another study, measures of thought rehearsal
correlated with both anger and sadness, whereas depressive rumination correlated only
with sad mood (Thomsen, Mehlsen, Christensen, & Zachariae, 2003). Thus, rumination

11
arising from different moods may reflect an underlying process of repetitive thought.
Cognitive inflexibility and perseverative thinking in general relate to more negative
outcomes (Channon, 1996; Davis & Nolen-Hoeksema, 2000; Newman, Patterson, &
Kosson, 1987; Shapiro et al., 1988). Therefore, the underlying process of repetitive and
rigid thinking may be more important for subsequent outcomes than the initial mood that
caused the rumination.
Interestingly, Berkowitz (1990) argued that both angering and saddening
experiences can stimulate anger-related associative networks and lead to aggression. In
his cognitive neoassociationistic model, he proposed that an associative network links
any kind of negative affect to anger-related feelings, thoughts, and memories, as well as
aggressive inclinations. Thus, sadness and depression (as well as foul odors, frustration,
or uncomfortable temperatures) will activate this large network and result in angry
feelings and aggressive behaviors. In support of his theory, some experimental research
has shown that induced sad feelings also produce anger and overt hostility (Finman &
Berkowitz, 1989). Although this model did not incorporate rumination, it seems possible
that repetitive and uncontrollable thinking might exacerbate any existing negative mood
and further activate the aggressive association network. Therefore, repetitive thoughts
resulting from any negative mood may lead to increased anger and aggression.
In sum, the literature suggests that both angry and sad moods should produce
rumination. However, opposing hypotheses can be made regarding the effect of this
naturally-occurring rumination on aggression. In line with emotion research, one
hypothesis states that only rumination arising from angry moods will lead to aggressive

12
thoughts and actions, whereas rumination arising from sad moods will result in decreased
motivation for aggression. However, an opposing hypothesis proposes that rumination
arising from any negative mood will trigger an aggressive associative network, resulting
in increased anger, hostility, and aggression.
The present research
The following two experiments expanded on previous research in several ways.
First, rather than manipulating rumination, this study manipulated mood only and
measured subsequent naturally-occurring rumination. This constituted the first test of
whether both angry and sad moods produce rumination. Moreover, this study also
compared the effects of ruminative thoughts arising from angry and sad moods on the
activation of aggressive association networks for men and women.
STUDY 1
Study 1 manipulated angry and sad moods in order to obtain naturally-occurring
rumination. I then examined the effect of this rumination on the activation of an
aggressive associative network. For the mood inductions, participants read mood-related
vignettes and wrote autobiographical narratives (Kenworthy et al., 2003). Both angry
and sad moods were expected to heighten self-reported ruminative thinking, compared to
neutral mood.
The activation of aggressive associative networks was measured in two ways.
Participants completed a word completion task that indirectly assessed the cognitive
accessibility of three types of aggression-related constructs: behavioral actions associated
with aggression, physiological arousal, and angry affect. Greater accessibility of these

13
constructs likely reflects activation of an aggressive association network. Participants
also provided open-ended responses to ambiguous interpersonal scenarios involving
provocation (MacBreyer, Milich, & Hundley, 2003). Their responses were coded for
hostile attributions and aggressive behavioral intentions. Perceiving hostile intent in
response to ambiguous provocations is associated with increased aggressive intentions
and behavior (Orobio de Castro et al., 2002; VanOostrom & Horvath, 1997).
If only rumination resulting from angry mood leads to activation of aggression
associative networks, then rumination should correlate with increased accessibility of
aggression-related constructs, hostile attributions, and aggressive intentions only for
participants in the angry mood condition. However, if rumination resulting from any
negative mood activates aggression networks, then rumination should correlate with
increased aggression measures for participants receiving both the angry and sad mood
manipulations. .
Finally, this study assessed self-focused attention, one variable that may
distinguish between angry and sad forms of ruminative thought. Specifically, anger
likely involves an external, other-oriented cognitive focus, whereas sad mood and
depression are associated with an internal self-focus (Mor & Winquist, 2002). Anger-
related thoughts often center on attributions of blame to others and revenge-related
actions (Averill, 1983; Baumeister, Stillwell, & Wotman, 1990). By contrast, sadness-
related thoughts often involve self-blame and self-deprecation (Lyubomirsky & Nolen-
Hoeksema, 1995). Moreover, sad mood led to significantly more self-focused attention
than neutral or happy moods (Salovey, 1992; Wood, Saltzberg, & Goldsamt, 1990),

14
whereas researchers suggest that anger may involve more external focus (Izard, 1991).
Thus, naturally-occurring ruminative thoughts may differ in internal versus external
focus, based on the initial mood. These differences might in turn contribute to divergent
behavioral tendencies. Specifically, internally-focused thoughts might relate to
withdrawal behaviors, whereas externally-focused thoughts might elicit more approach-
related behaviors. Thus, if angry and sad moods lead to different ruminative thoughts, I
would expect increased rumination to covary with less self-focused attention for angered
participants but with increased self-focused attention for saddened participants.
Moreover, differences in self-focused attention might relate to hostile attributions and
aggressive intentions.
Method
Participants and Design
One hundred students (76 women and 24 men) recruited from undergraduate
psychology courses participated in the study for course credit. Forty six percent of
participants were Caucasian, 33% were Asian, 14% were Hispanic, 3% were African-
American, and 4% were of other ethnicities. Seventy-five percent of participants were
native English speakers. Participants were randomly assigned to the angry, sad, or
neutral mood condition in a one-way, between-subjects design.
Measures
Mood inductions. Sad, angry, and neutral moods were induced using both vignettes
and autobiographical narratives (Kenworthy et al., 2003; Larsen & Ketelaar, 1991; Rusting,
1999). All participants first had 4 minutes to read two vignettes. The vignettes for angry

15
mood included (1) arriving late to an important interview because of a bad driver and (2)
having a bad roommate who moves out unexpectedly and steals many personal belongings.
The sad vignettes included passages about (1) attending to a younger sibling in the hospital
with a serious illness and (2) the unexpected death of one’s mother. The vignettes for the
neutral conditions included (1) having the day off and doing mundane things and (2)
coming home from work and then going out to eat dinner. All vignettes were
approximately 350 words long.
Based on previous research on mood inductions (Rusting, 1999), participants
were instructed to:
Imagine the situations as vividly as you can. Picture the events
happening to you. Picture in your “mind's eye” the surroundings as
clearly as possible. See the people or objects; hear the sounds;
experience the events happening to you. Think the thoughts you
would actually think in these situations. Feel the same feelings you
would feel in these situations. Let yourself react as if you were
actually there.
Following the vignettes, participants wrote autobiographical narratives about a time when
they experienced the same emotion. The instructions read: “Now, please think about
some experience in the past when you felt this same emotion. Really try to put yourself
back in that moment and write about it for the next ten minutes.”
Mood manipulation checks. A 15-item mood manipulation check was comprised
of positive, angry, and sad adjectives. Participants responded to all mood adjectives on a

16
seven-point scale. The adjectives glad, pleased, content, friendly, and happy were
averaged to create a positive mood index (Cronbach’s α = .93). The sad mood index
consisted of the words sad, mournful, sorrowful, somber, and gloomy (Cronbach’s α =
.91). Finally, an angry mood index was created from the words mad, angry, offended,
annoyed, and furious (Cronbach’s α = .91).
State rumination. Participants completed a 15-item measure of state rumination.
Ten items came from the rumination subscale of the Response Styles Questionnaire
(RSQ; Nolen-Hoeksema & Morrow, 1991) and assessed to what extent participants were
currently thinking negative thoughts like “why can’t I handle things better” and “why am
I so upset?” These items were altered from the original measure to assess state
rumination rather than trait rumination. The other 5 items were created for this study to
assess more general repetitive thinking. The items included “I am ruminating,” “The
same thought is bouncing around and around in my head,” “I can’t stop mulling thing
over,” “My thoughts are stuck on one thing,” and “My mind is harping over and over on
something.” Participants responded to all items on a 7-point scale. Higher averaged
scores indicated more rumination. Internal consistency for this measure was high
(Cronbach’s α = .93).
Self-focused attention. The measure of self-focused attention was disguised as a
measure of linguistic implications (Wegner & Giuliano, 1980, 1983). The measure
consisted of 20 sentences, each of which contained a blank. The cover story for this task
explained that the experimenter was interested in the idea that words in a sentence are
often redundant and can be guessed from knowledge of the remainder of the sentence. So,

17
to collect basic statistics on the redundancy in a series of sentences, participants were
asked to choose the most appropriate word from among three alternatives for each blank
(e.g., “It isn't easy to get lost in this town, but somehow (I, they, we) managed it”). All
three alternatives were grammatically correct, but participants were asked to pick the
alternative that seemed especially likely to be most appropriate in the context of the rest
of the sentence (Salovey, 1992). Sentences completed with the first-person singular
pronoun (e.g., I, me) received a score of 1, and sentences completed with any other
pronoun received a score of 0. Total summed scored indicated more self-focused
attention. Internal consistency for these 20 items was somewhat lower (Cronbach’s α =
.60), similar to other reports (Wegner & Giuliano, 1980; Wood, Saltzberg, & Goldstamt,
1990). Wegner and Giuliano (1980) suggest that low internal consistency is inherent in
such quasi-projective measures.
Cognitive accessibility of aggressive constructs. Participants also completed a
word-completion measure that implicitly assessed the accessibility of aggression-related
words (Anderson, 1997; Pedersen et al., 2005). The instructions indicated that each item
could be made into several possible words. Participants read 35 words with missing
letters (e.g., “_ell,” “_r_ _ sed,” “ha_e”), all of which could be completed as aggressive
or non-aggressive words (e.g., “yell or bell,” “aroused or bruised,” “hate or have”). The
measure consisted of 3 subscales: action words indicated a physically aggressive
response (e.g., hit), affect words indicated an aggressive emotional response (e.g., mad),
and arousal words indicated a cognitive representation of arousal (e.g., tense). There
were 14 action words, 11 negative affect words, and 9 arousal words.

18
Coders later scored each word as aggressive, non-aggressive, or incomplete (e.g.,
skipped words or fake words). Because the number of words in each subscale differed,
the total number of aggressive words in a subscale was divided by the total number of
words that that participant completed within that subscale, yielding a percentage. This
was done separately for each of the 3 subscales.
Hostile attributions and aggressive tendencies. Participants read 6 vignettes
describing ambiguous social interactions with adult peers (MacBreyer, Milich, &
Hundley, 2003). All of the vignettes involved subtle provocations. After reading each
vignette, participants wrote open-ended responses to two questions: “Why do you believe
the exchange happened?” and “How would you respond in this situation?” The first
question assessed hostile attributions, and the second question assessed aggressive
tendencies.
Two independent coders who were blind to participant condition separately
scored participants’ responses. Following a coding system developed by MacBrayer and
colleagues (2003), each response to the six hostile attribution questions received either a
hostile or a benign attribution rating. Hostile responses included attributions that another
person acted intentionally, with hostility, lack of sensitivity, or the intent to harm, (e.g.,
“She is trying to make a joke of me,” “He’s just a jerk”). Benign responses included
attributions that the negative event was due to an accident, outside circumstances, or to
the participant (e.g., “I wasn’t paying close enough attention,” “He didn’t mean to hurt
my feelings”).

19
Likewise, each response to the six aggressive tendencies questions received either
an aggressive or non-aggressive rating. Aggressive responses included retaliation of any
kind (e.g., verbal or physical force, active ignoring, hurting someone indirectly), whereas
non-aggressive responses included withdrawing, sharing a feeling, requesting an
explanation, or an emotional response with no action (e.g., “I’d feel hurt,” “I’d ask why
they did that,” “I’d walk away”).
The total number of hostile attributions and aggressive intentions across all six
scenarios constituted two dependent variables for each participant. Agreement between
raters was adequate. Intraclass correlation coefficients for hostile attributions and
aggressive tendencies were .82 and .78, respectively.
Dispositional measures. Finally, participants completed measures of trait
rumination and trait aggression. The 12-item rumination subscale of the Rumination and
Reflection Questionnaire (RRQ; Trapnell & Campbell, 1999) includes items like “I
always seem to be rehashing in my mind recent things I’ve said or done” and “My
attention is often focused on aspects of myself I wish I’d stop thinking about.” Our
research shows that this subscale correlates with aggression (Borders & Earleywine,
2004). Participants indicated their level of agreement with each item on a 5-point likert
scale from “strongly disagree” to “strongly agree.” Higher averaged subscale scores
indicated more rumination. Internal consistency in this sample was .89.
The 29-item Aggression Questionnaire (Buss & Perry, 1992) assesses physical
aggression, verbal aggression, hostility, and anger. This measure reliably differentiates
between violent and non-violent participants and correlates with responses to provoking

20
scenarios and peer reports of aggression (O’Connor, Archer & Wu, 2001; Morren &
Meesters, 2002). The measure employs a Likert-type scale from 1 (Extremely
uncharacteristic of me) to 5 (Extremely characteristic of me). Higher summed scores
indicated more dispositional aggression. Internal consistency with this sample was good
( α=.92).
Procedure
Upon arrival, participants were told that they were participating in a study that
included three separate experiments about language use. According to the cover story,
the first experiment explored how mood affected writing styles, the second experiment
involved providing feedback about two new measures of word choice, and the third,
“phrase structure” experiment was explained as an investigation of how different
sentence structures impacted people’s conclusions about ambiguous situations (see
Figure 1). Participants were run in small groups of 4-6 students. The entire procedure
took approximately one hour.
Participants first answered some demographic questions, including filler items
about the number of books they typically read each year and the age at which they first
started reading. The “first section” included the mood induction, 15-item mood
manipulation check, and the state rumination measure. The experimenter timed the mood
induction, allowing participants 4 minutes to read the mood-related vignettes and 10
minutes to write an autobiographical narrative. Participants were allowed to proceed
through the rest of the measures at their own pace. In the “second section” of the
experiment, participants completed the self-focused attention and word completion

21
Figure 1. Procedure for Study 1.

Section 1 instructions: Read two mood-related vignettes and then write an
autobiographical narrative about a related mood.

Mood manipulation:
angry, sad, neutral

Measured variables:
Angry, sad, and positive moods
State rumination

Section 2 instructions: Provide feedback on two new measures of word choice.

Measured variables:
Self-focused attention
Accessibility of aggression-related
constructs

Filler questions:
Opinions about measures

Section 3 instructions: Read 6 hypothetical scenarios involving ambiguous provocations
and answer “Why did this occur?” and “How would you respond?”

Aggression measures:
Hostile attributions
Aggressive intentions

Final measures instructions: Shake yourself of whatever mood the experiment put you in
and answer these questions about how you usually are.

Final measures:
Trait rumination
Trait aggression

22
measures. Following each of these measures, they answered some questions about the
quality of the measure (e.g., “This measure was clear”), to preserve the cover story.
Next, the “third section” included responding to the ambiguous scenarios.
Finally, to preserve the cover story that the experiment was about language use, the
questionnaires about dispositional rumination and aggression were administered at the
end of the experiment, rather than at the beginning. This way, participants would not
become aware of the actual variables of interest until the very end of the study. The sole
purpose of these measures was to determine whether random assignment successfully
matched participants on dispositional rumination and aggression. The instructions for
these questionnaires read: “Before completing these next measures, please try to shake
yourself of whatever mood the study put you in. These measures ask about how you
usually are, not how you’re feeling right now.” Only 64 participants completed these
dispositional measures.
Results
Based on the p < .001 criterion for Mahalanobis distance (Tabachnick & Fidell,
2001), one case was identified as a multivariate outlier. Three other cases had extremely
high scores on the measure of self-focused attention, and a fifth case had an unusually
high score on the word-completion measure. These five cases were excluded from
subsequent analyses. All variables met assumptions for normality except the angry mood
index. All reported statistical tests were performed using lognormal transformation of

23
this variable (Tabachnick & Fidell, 2000)
1
. However, for ease of interpretation, all
means and standard deviations reported below are from non-transformed variables.
Initial analyses revealed no gender interactions for any measure. Therefore, the
data were collapsed across gender and run in one-way (Mood Condition: Sad / Angry /
Neutral) between-subject analyses of variance. In addition, correlations between study
measures are reported for the entire sample (see Table 1) and separately by mood
condition (see Table 2).
Random assignment checks
As expected, there were no differences across groups in trait rumination, F(2,
60)=.50, ns, or aggression, F(2, 60)=.11, ns. Thus, random assignment successfully
matched participants on dispositional rumination and aggression. However, there were
differences in the number of men and women assigned to each condition, χ
2
(2, N = 95) =
10.57, p < .01. Specifically, approximately 63% of male participants were randomly
assigned to the sad condition, with the remaining 37% of men equally split between the
angry and neutral conditions. Thus, the sad condition had a significantly higher
proportion of men than did the other two conditions.
Men and women did not differ in the amount of dispositional rumination they
reported (sample M = 3.60, SD = .77). However, men reported higher dispositional
aggression (M=85.53, SD=10.88) than did women (M=69.01, SD=19.59), t(61)=3.01, p<
.01. These aggression levels are higher than reported norms for men (men M = 77.8;
women M = 68.2; Buss & Perry, 1992).

1
The skewness and kurtosis statistics of the angry mood index were 1.15 and .38, respectively, before
lognormal transformations and .51 and -1.07 after.
24
Table 1. Correlations between Study 1 variables for entire sample (N=95).

Measure 1 2 3 4 5 6 7 8 9 10 11 12

1. Sad mood —
2. Angry mood .26** —
3. Positive mood -.64^ -.38^ —
4. State rumination .51^ .51^ -.39^ —
5. Self-focused attention -.11 .08 .02 -.02 —
6. Action words -.04 .05 .08 .04 .18 —
7. Arousal words -.07 .25* -.03 .11 .03 .13 —
8. Affect words -.07 .19 -.07 .12 -.03 .13 .14 —
9. Hostile attributions -.11 .20* -.04 -.04 .06 .04 .32** .06 —
10. Aggressive tendencies .09 .11 -.15 .15 -.05 .10 .03 -.05 .27** —
11. Trait rumination .11 .08 .01 .28* -.02 .01 .06 .21 .11 .14 —
12. Trait aggression .11 .11 .02 .19 .00 .15 -.07 .25* .01 .30** .41** —

* p < 05. ** p < .01. ^ p < .001.

24
25
Table 2. Correlations between Study 1 variables by mood condition.

ANGRY CONDITION (N=29)

Measure 1 2 3 4 5 6 7 8 9 10

1. Sad mood —
2. Angry mood .36* —
3. Positive mood -.51** -.41* —
4. State rumination .30 .49** -.30 —
5. Self-focused attention -.11 .08 .02 .04 —
6. Action words -.03 .29 .03 .13 .19 —
7. Arousal words -.05 .35 -.30 -.06 .16 .12 —
8. Affect words -.01 .26 -.25 .28 -.15 .05 .13 —
9. Hostile attributions -.05 .19 -.15 -.03 -.25 -.22 .18 -.12 —
10. Aggressive tendencies -.12 -.13 .16 -.05 -.03 -.08 -.14 -.06 .14 —

SAD CONDITION (N=33)

Measure 1 2 3 4 5 6 7 8 9 10

1. Sad mood —
2. Angry mood .23 —
3. Positive mood -.65^ -.09 —
4. State rumination .54^ .37* -.24 —
5. Self-focused attention .00 .14 -.18 -.07 —
6. Action words .08 .04 -.22 .02 .34 —
7. Arousal words .17 .38* -.11 .43* .02 .09 —
8. Affect words -.02 .11 -.10 .29 .11 .26 .10 —
9. Hostile attributions .01 .35* -.05 .09 .11 .23 .31 .22 —
10. Aggressive tendencies .16 .06 -.19 .16 -.07 .35* .02 .10 .61^ —
25

26
NEUTRAL CONDITION (N=33)

Measure 1 2 3 4 5 6 7 8 9 10

1. Sad mood —
2. Angry mood .54** —
3. Positive mood -.51** -.22 —
4. State rumination .54^ .46** -.15 —
5. Self-focused attention -.03 .07 .01 .10 —
6. Action words -.27 .03 .41* .19 -.05 —
7. Arousal words -.29 .01 .22 -.11 -.07 .22 —
8. Affect words -.10 .04 .20 -.25 -.14 .13 .14 —
9. Hostile attributions -.18 .18 -.10 -.14 .15 .09 .42* .09 —
10. Aggressive tendencies .04 .23 -.14 .14 -.22 .03 .21 -.30 .03 —

* p < 05. ** p < .01. ^ p < .001.

26

27
angry and neutral conditions. Thus, the sad condition had a significantly higher
proportion of men than did the other two conditions.
Men and women did not differ in the amount of dispositional rumination they
reported (sample M = 3.60, SD = .77). However, men reported higher dispositional
aggression (M=85.53, SD=10.88) than did women (M=69.01, SD=19.59), t(61)=3.01, p<
.01. These aggression levels are higher than reported norms for men (men M = 77.8;
women M = 68.2; Buss & Perry, 1992).
Mood manipulation checks
As expected, the three conditions differed in sad mood, F(2, 92) = 20.00, p
<.0001, angry mood, F(2, 92) = 22.73, p<.0001, and positive mood, F(2, 92) = 20.20, p <
.0001. Specifically, participants in the sad condition reported more sad affect (M=3.92,
SD=1.68, range=1-6.2) than participants in the angry (M=2.41, SD=1.13, range=1-5.2)
and neutral (M=1.90, SD=1.12, range=1-5.4) conditions (both pairwise p-values < .0001).
Likewise, participants in the angry condition reported significantly more angry affect
(M=3.19, SD=1.47, range=1-6) than participants in both the sad (M=1.93, SD=1.00,
range=1-4.2) and neutral (M=1.38, SD=.57, range=1-3.2) conditions. Finally,
participants in the neutral condition reported significantly more positive affect (M=4.38,
SD=1.06, range=2.4-6.6) than participants in the sad (M=2.68, SD=1.35, range=1-5.6)
and angry (M=2.85, SD=1.13, range=1.2-5.2) conditions.
Across the entire sample as well as in each mood condition, positive mood
demonstrated strong negative correlations with sad mood (see Tables 1 and 2).
Interestingly, positive mood only inversely covaried with angry mood in the entire

28
sample and for participants in the angry mood condition. Finally, sad and angry moods
were positively correlated for the entire sample and participants in both angry and neutral
conditions.
State rumination
Participants completed a 15-item measure state rumination. Analyses yielded an
overall effect of mood on rumination, F(2, 92) = 9.18, p<.0001 (see Figure 2).
Specifically, participants in both the angry (M=3.62, SD=1.12, range=1.2-5.7) and sad
conditions (M=3.46, SD=1.38; range=1.3-6.0) reported more rumination than participants
in the neutral condition (M=2.38, SD=1.26; range=1.0-5.6) (both pairwise p-values <
.0001). There was no significant difference between the angered and saddened
participants.

Figure 2. Effect of mood manipulation on state rumination (Study 1).

1
2
3
4
5
6
7
Angry Sad Neutral
Rumination subscale of RSQ

F(2, 92) = 11.60, p < .0001

29
To determine whether participants’ self-reported moods following the mood
manipulation accounted for their reported level of rumination, I performed mediation
analyses. Traditionally, a test of mediation requires three separate regression analyses
(Baron & Kenny, 1986). First, the independent variable (IV: group assignment) must
predict the dependent variable (DV: state rumination). This first analysis establishes that
a relationship in fact exists to be explained, and the regression coefficient is called the
“total effect.” Second, the IV must predict the proposed mediator (state mood). Third,
the DV is regressed on both the IV and the mediator. The mediator must predict the DV,
even with the IV also entered into the equation. If these three tests are all significant,
then the coefficient for the IV in this third equation is examined. This regression
coefficient is called the “direct effect” and corresponds to the effect of the IV by itself,
after removing any variance explained by the mediator. Statistically, mediation occurs
when the there is no significant direct effect, or when the IV no longer affects the DV
after inclusion of the mediator in the regression equation. Conceptually, this result
suggests that the original effect of the IV is in fact explained by the mediating variable.
In other words, the IV has no direct effect on the DV, after accounting for the effect of
the mediator.
Researchers have designed ways to quantify the effect of mediation and
subsequently test for significance (for a review, see MacKinnon et al., 2002). One
common approach is to create an estimate of the “indirect effect,” or the effect of the
mediator (e.g., Bollen, 1987; Sobel, 1982). The indirect effect is defined as a product of
two regression coefficients: (1) the effect of the IV on the mediator and (2) the effect of

30
the mediator on the DV after controlling for the IV. A significance test of mediation
roughly determines whether the indirect effect differs significantly from zero. Traditional
significance tests involve dividing the indirect effect by an estimate of its standard error
and comparing the resulting value to the standard normal distribution (e.g., Sobel, 1982;
see MacKinnon et al., 2002). However, more recent analyses indicate that estimates of
indirect effects are not normally distributed, especially with small sample sizes (Bollen &
Stine, 1990; MacKinnon et al., 2002). Accordingly, these traditional tests generally have
low power and may lead to incorrect conclusions.
One promising new method of testing indirect effects in small samples without
making assumptions about normality uses bootstrapping methodology (Bollen & Stine,
1990; Lockwood & MacKinnon, 1998; Preacher & Hayes, 2004; Shrout & Bolger, 2002).
Specifically, a “bootstrap sample” consists of N individuals sampled randomly with
replacement from the original data set (Efron & Tibshirani, 1993). In other words, the
current sample of participants serves temporarily as a population, from which 95
individuals are randomly sampled with replacement to create a new, bootstrap sample.
One thousand bootstrap samples are created in this way, and each sample yields a
different value of the indirect effect of mediation. These 1000 values create an empirical
sampling distribution of the indirect effect. Accordingly, a 95% confidence interval for
the population indirect effect is defined simply by the values at the 5
th
and 95
th
percentiles
of the bootstrap distribution (Bollen & Stine, 1990; Efron & Tibshirani, 1993).
Moreover, the average of the 1000 indirect effects yields a bootstrap estimate of the
population indirect effect, and the standard deviation of these values yields a bootstrap

31
estimate of the standard error. In simulations in which the distributions of indirect effects
are skewed, this bootstrapping method has more power than traditional methods that
assume normality, even with small samples (Bollen & Stine, 1990; Efron & Tibshirani,
1993; Shrout & Bolger, 2002).
If evidence for mediation exists, reverse mediation should also be tested and ruled
out. If the IV, DV, and mediator are all intercorrelated, it is possible that the DV (state
rumination) in fact mediates the link between the IV (mood manipulation) and the
hypothesized mediator (state mood). Ruling this out is especially important for designs
where the mediator and DV are measured at the same time point. In these cases, the
specified “mediator” may actually be the true “dependent variable,” and vice versa
(Baron & Kenny, 1986). Thus, testing for reverse mediation requires performing the
mediation analyses as described above with the mediator and DV switched. In other
words, in the third regression equation the original mediator is regressed on both the IV
and the original DV. Reverse mediation can be ruled out if the coefficient between the
IV and original mediator (the new direct effect) is significant. Conceptually, ruling out
reverse mediation suggests that the proposed mediation works in only one direction and
occurs between conceptually distinct variables. This test therefore provides more robust
statistical evidence that the proposed mediator in fact constitutes a unique mechanism by
which the IV causes the specified DV. Failure to rule out reverse mediation and establish
directionality, on the other hand, could imply that the mediator and DV are conceptually
interchangeable variables. In fact, they may be seen as different manifest variables of
one underlying latent trait.

32
Entering group assignment into regression analyses required the creation of two
dummy variables (Aiken & West, 1991). The neutral condition was designated as the
comparison group, as a baseline against which to compare the sad and angry conditions.
In dummy coding, the comparison group is assigned a value of 0 in all dummy variables.
For each dummy variable, then, one other group is assigned a value of 1 and the third
group is assigned a value of 0. Thus, the sad dummy variable compared participants in
the sad condition to participants in the neutral condition. The angry dummy variable
compared participants in the angry and neutral conditions. Inclusion of both dummy
variables in a regression equation is equivalent to performing a one-way ANOVA.
However, rather than reporting the overall effect of both dummy variables, I report below
whether the raw regression coefficients for each dummy variable were significantly
different from zero.
The first set of regression analyses evaluated whether sad mood mediated the
effect of mood condition on rumination (see Figure 3a). First, the regression coefficient
for both the sad (b=1.08) and angry (b=1.24) dummy variables were significant (both p-
values < .0001), confirming that group assignment predicted state rumination. Second,
only the sad dummy variable predicted sad mood, b=1.27, p < .0001. The coefficient for
the angry dummy variable was not significant, b=.32, ns. Thus, sad mood could only
mediate the effect of the sad condition on rumination. Third, when rumination was
regressed on the sad dummy variable and sad mood together, sad mood predicted state
rumination, b=.80, p < .0001. Moreover, the effect of the sad mood condition was no
longer significant, b=-.39, ns. Using 1000 bootstrap samples, the bootstrap estimate of

33
the indirect effect was .89, with an estimated standard error (SE) of .21 and a 95%
confidence interval (CI) from .51 to 1.35. Because the confidence interval does not
include zero, I can conclude that sad mood significantly mediated the effect of the sad
mood condition on rumination. Finally, when sad mood and rumination were reversed in
the final equation, the effect of the sad mood condition on the sad mood remained
significant (b=.92, p < .0001), ruling out the possibility of reverse mediation.
A second set of regression analyses evaluated whether angry mood mediated the
effect of mood condition on rumination (see Figure 3b). Recall that both the angry
(b=1.24) and sad (b=1.08) dummy variables predicted state rumination. This time, both
the angry dummy variable (b=1.38) and the sad dummy variable (b=.50) predicted angry
mood (both p-values < .05). In the third equation, rumination was regressed on both
dummy variables and angry mood together. Even after controlling for mood condition,
angry mood predicted state rumination, b=.66, p < .0001. Moreover, the direct effect of
the angry condition on rumination was no longer significant (b=.33, ns), suggesting
mediation. Bootstrapping yielded an estimate of .86 for the indirect effect of mediation
(SE=.22; CI=.42-1.28). By contrast, the coefficient for the sad dummy variable remained
significant, b=.75, p < .01. Thus, angry mood mediated the effect of the angry mood
manipulation but not the sad mood manipulation on state rumination. Finally, when
angry mood and rumination were reversed in the final equation, the effect of the angry
mood condition on angry mood remained significant, b=1.04, p < .0001. Thus, there was
no evidence of reverse mediation.

34
Figure 3. Mood indices mediate the effect of mood manipulations on state rumination
(Study 1).

(a) Sad mood mediates the effect of sad mood manipulation on state rumination

Note: Bolded coefficients are significant. Indirect effects are in parentheses
Mood manipulation:
angry vs. neutral
Sad mood
State
rumination
1.24
.70 .32
Bootstrap estimate of indirect effect = .89 (SE = .21)
95% confidence interval = .51 to 1.35
Mood manipulation:
sad vs. neutral
State
rumination
Sad mood
1.08 (-.39)
1.27 .70 (.80)

35
(b) Angry mood mediates the effect of angry mood manipulation on state rumination

Note: Bolded coefficients are significant. Indirect effects are in parentheses
Mood manipulation:
angry vs. neutral
Angry mood
State
rumination
1.24 (.33)
.72 (.66) 1.38
Bootstrap estimate of indirect effect = .86 (SE = .22)
95% confidence interval = .42 to 1.28
Mood manipulation:
sad vs. neutral
State
rumination
Angry mood
1.08 (.75)
.50 .72 (.66)

36
To further examine the effect of mood manipulation on state rumination, separate
rumination subscales were created: one subscale consisted of the 10 items from the
Response Styles Questionnaire (RSQ; Nolen-Hoeksema & Morrow, 1991) that measured
extent of negative thinking, and the other subscale consisted of the 5 items created for
this study to assess general repetitive thinking (RT). Participants in the angry (M=3.57,
SD=1.30) and sad (M=3.70, SD=1.57) conditions reported more repetitive thinking than
participants in the neutral mood condition (M=2.92, SD=1.64). However, this effect
failed to reach significance, F(2, 92)=2.43, p=.09. Analyses did yield a significant
overall effect of mood on the RSQ, F (2, 92) = 11.60, p<.0001. Specifically, participants
in both the angry (M=3.66, SD=1.27) and sad conditions (M=3.41, SD=1.51) reported
more rumination than participants in the neutral condition (M=2.13, SD=1.27) (both
pairwise p-values < .0001). There was no significant difference between the angered and
saddened participants. Because these two rumination subscales were highly correlated
(r=.61, p < .0001), only the total rumination score was used in subsequent analyses.
Across the entire sample, state rumination correlated with greater reported angry
and sad moods and with less positive mood (see Table 1). When analyzed separately by
group, however, the correlations were less consistent (see Table 2). For participants in
the sad and neutral conditions, state rumination covaried with increased angry and sad
moods. However, for participants receiving the angry mood manipulation, rumination
covaried with increased angry mood only.

37
Self-focused attention
A one-way (Mood Condition: Sad / Angry / Neutral) ANOVA showed no
significant effect of self-focused attention, F (2, 92) = 1.56, ns. Participants in all
conditions chose an average of 8.55 first-person pronouns out of 20 items (SD=2.40,
range=4-14). Across both the entire sample and separately by mood condition, self-
focused attention did not covary any other study variable (see Tables 1 and 2).
Accessibility of aggression-related constructs
The three word types (i.e., action, arousal, and affect) were initially combined to
form a single measure of the accessibility of aggression-related cognitions. A one-way
(Mood Condition: Sad / Angry / Neutral) ANOVA yielded no overall effect of mood
condition on the percentage of aggressive words, F(2, 92) = .69, ns. On average,
participants completed 26% of the word stems with aggressive-related words (SD=.08,
range=.06-.43).
To separately examine the 3 distinct word types, a 3 (Mood Condition: Sad /
Angry / Neutral) by 3 (Word Type: Action / Arousal / Affect) mixed ANOVA was
conducted. Aside from a main effect for word type, F(2, 91) = 15.41, p < .0001, no other
main effects or interaction emerged (all p-values > .41). In general, participants reported
greater accessibility of action words (M=.28, SD=.12) than arousal words (M=.19,
SD=.13), t(94) = 5.34, p<.0001. Participants also reported more affect words (M=.27,
SD=.14) than arousal words, t(94) = 4.42, p < .0001. Because the three subscales did not
significantly intercorrelate (see Table 1), they were not combined for any subsequent
analyses.

38
Across the entire sample, these three word types did not significantly covary with
state rumination (see Table 1). However, for participants in the angry mood condition,
increased rumination related to increased accessibility of arousal-related words (see Table
2). For all participants, greater angry mood correlated with more accessibility of arousal-
related words (see Table 1). Interestingly, when correlations were examined within mood
conditions, this correlation was only significant for participants in the sad mood condition
(see Table 2). In addition, increased positive mood correlated with more accessibility of
action-related words for participants in the neutral group.
Aggression measures
Participants’ responses to ambiguous social interactions with adult peers
(MacBrayer et al., 2003) yielded two dependent variables: hostile attributions and
aggressive intentions. Across all 6 scenarios, participants reported an average of 2.32
hostile attributions (SD=1.36, range=0-6) and 1.38 aggressive intentions (SD=1.12,
range=0-4.5). Thus, this particular sample exhibited relatively low levels of aggressive
intentions. Analyses showed no effects of mood condition on hostile attributions, F(2,
92) = .54, ns. Likewise, there was no difference between groups on aggressive
intentions, F(2, 92) = 1.52, ns. These two indices were significantly correlated for the
entire sample (see Table 1) and for participants in the sad group (see Table 2).
Across the entire sample and separately by groups, neither hostile attributions nor
aggressive intentions correlated with state rumination (see Tables 1 and 2). However,
hostile attributions did correlate with increased anger and accessibility of arousal-related
words in the whole sample. Interestingly, when examined separately by group, hostile

39
attributions only correlated with anger for participants in the sad group and with
accessibility of arousal words for participant in the neutral group. Finally, aggressive
intentions covaried with accessibility of action-related words for participants receiving
the sad manipulation.
Mediation analyses
The pattern of correlations between study variables for the entire sample
warranted a test of whether the accessibility of arousal-related words mediated the link
between angry mood and hostile attributions (see Figure 4). As before, I performed
mediation analyses using three separate regression equations (Baron & Kenny, 1986).
First, angry mood predicted hostile attributions, b=.28, p < .05. Second, angry mood
predicted the accessibility of arousal-related words, b=.26, p < .05. Third, when hostile
attributions were regressed on angry mood and accessibility of arousal-related words
together, the accessibility of arousal words predicted hostile attributions, b=.40, p < .01.
Moreover, the effect of angry mood was no longer significant, b=.17, ns. The bootstrap
estimate of the indirect effect was .10 (SE=.05; CI=.02-.23), suggesting that the
accessibility of arousal-related words mediated the effect of angry mood on hostile
attributions. Finally, when accessibility of arousal words and hostile attributions were
reversed in the third equation, the effect of angry mood on the accessibility of arousal
words was still significant (b=.20, p < .05), ruling out the possibility of reverse
mediation.

40
Figure 4. Accessibility of arousal words mediates the link between angry mood and
hostile attributions (Study 1).

Note: Bolded coefficients are significant. Indirect effects are in parentheses.

Angry mood
Accessibility
of arousal-
related words
Hostile
attributions
.28 (.17)
.44 (.40) .26
Bootstrap estimate of indirect effect = .10 (SE = .05)
95% confidence interval = .02 to .23

41
Table 2 presents correlations between study variables separately by mood
condition. The particular pattern of correlations for the sad group permitted a test of
whether state rumination mediated the association between angry mood and the
accessibility of arousal-related words (see Figure 5). I ran the same three regression
equations. Angry mood predicted accessibility of arousal words (b=.06, p < .05) and
state rumination (b=.44, p < .05). However, after controlling for angry mood, rumination
did not predict arousal-related words (b=.05, p = .06). Thus, although the direct effect of
angry mood was not reliable (b=.04, ns) when rumination was included in the final
equation, the conditions for mediation were not met. Accordingly, the bootstrap estimate
of the indirect effect was only .02 (SE=.01) and the confidence interval included zero
(CI=.00-.06). Thus, rumination did not mediate the link between angry mood and the
accessibility of arousal-related words.
Dispositional measures
Across the entire sample, participants who reported higher trait aggression
exhibited significantly more aggressive tendencies and greater accessibility of affect
words (see Table 1). Moreover, participants who reported higher dispositional
rumination endorsed more state rumination. The trait measures of rumination and
aggression were also positively correlated.

42
Figure 5. State rumination mediates the link between angry mood and accessibility of
arousal words, for participants in the sad mood condition (Study 1).

Note: Bolded coefficients are significant. Indirect effects are in parentheses
Angry mood
State
rumination
Accessibility of
arousal-related words
.06 (.04)
.06 (.05) .44
Bootstrap estimate of indirect effect = .02 (SE = .01)
95% confidence interval = .00 to .06

43
Gender analyses
Independent t-tests found no significant differences between men and women in
any variables except trait aggression (discussed above). Correlations between all study
variables were examined separately by gender (see Table 3). Because there were only 22
men, compared with 73 women, the significance tests for men had low power. Therefore,
any correlation larger than .40 will be interpreted as meaningful. For men, rumination
correlated with worse sad mood and less positive mood, but not increased angry mood.
Angry mood also did not correlate with sad mood. These results could be due to a small
range for men on angry mood. Greater accessibility of arousal words was associated with
worse angry mood, whereas greater accessibility of affect words correlated with more
rumination and less positive mood. In addition, trait rumination showed large
correlations with state rumination and hostile attributions. Finally, trait aggression
correlated with all mood indices and with rumination.
Women showed a somewhat different pattern of correlations (see Table 3). For
women, rumination and all the mood indices intercorrelated as expected. Increased
accessibility of action words was associated with a less positive mood. Furthermore,
increased accessibility of arousal words correlated with more hostile attributions. This
time, trait rumination was associated with the accessibility of affect words. Moreover,
trait aggression was not correlated with the mood indices or rumination. Rather, it
correlated with accessibility of affect words and with increased aggressive tendencies.
44
Table 3. Correlations between Study 1 variables by gender.
MEN (N=22)
Measure 1 2 3 4 5 6 7 8 9 10 11 12
1. Sad mood —
2. Angry mood .18 —
3. Positive mood -.73^ -.49^ —
4. State rumination .72^ .36 -.62^ —
5. Self-focused attention -.31 -.14 -.12 -.25 —
6. Action words .07 .05 -.31 .04 .22 —
7. Arousal words .13 .60** -.39 .26 .08 .07 —
8. Affect words .38 .31 -.41 .46* .12 .16 .30 —
9. Hostile attributions .00 .15 -.01 -.38 .17 .00 .02 .03 —
10. Aggressive tendencies .24 .07 -.36 .05 .29 .21 .10 .09 .14 —
11. Trait rumination .37 .23 -.34 .48 -.07 -.27 .25 .00 -.40 .07 —
12. Trait aggression .41 .53* -.47 .56* -.13 -.21 -.28 -.03 -.33 .00 .58* —

WOMEN (N=73)
Measure 1 2 3 4 5 6 7 8 9 10 11 12
1. Sad mood —
2. Angry mood .30* —
3. Positive mood -.60^ -.37** —
4. State rumination .45^ .55^ -.33** —
5. Self-focused attention -.03 .13 .05 .04 —
6. Action words -.10 .04 .24* .04 .15 —
7. Arousal words -.12 .18 .05 .08 .01 .14 —
8. Affect words -.22 .16 .03 .05 -.08 .12 .11 —
9. Hostile attributions -.17 .22 -.03 .02 .04 .07 .42^ .07 —
10. Aggressive tendencies .03 .13 -.07 .18 -.15 .06 .03 -.08 .31** —
11. Trait rumination -.01 .06 .14 .23 .01 .12 .07 .28* .24 .15 —
12. Trait aggression .03 .12 .18 .18 .07 .30* .00 .35* .00 .35** .39** —
* p < 05. ** p < .01. ^ p < .001.

44

45
Discussion
Previous research on the link between rumination and aggression manipulated
angry mood and rumination together and explored their combined effect on subsequent
anger and aggression (Bushman, 2002; Bushman et al., 2005; Pedersen et al., 2005;
Rusting & Nolen-Hoeksema, 1998). However, theory suggests that rumination should
naturally arise from both angry and sad moods (Miller et al., 2003; Nolen-Hoeksema,
1991). Moreover, opposing hypotheses exist about the effect of rumination arising from
angry and sad moods on subsequent aggression. Emotion research suggests that only
angry mood and subsequent rumination will trigger aggressive thoughts and behaviors
(Kenworthy et al., 2003; Tiedens, 2001). However, other research suggests that
rumination arising from both sad and angry moods should trigger aggressive thoughts,
feelings, and behaviors (Berkowitz, 1990; Segerstrom et al., 2003). Study 1 constitutes
the first work to examine the effects of experimentally induced angry and sad moods on
naturally-occurring rumination and the subsequent activation of aggression-related
associative networks. Several important results emerged.
State rumination
As expected, both angry and sad mood manipulations led to increased state
rumination, compared to neutral mood condition. There were no differences in the
amount of rumination exhibited by angered and saddened participants. Moreover,
mediation analyses indicated that sad and angry moods resulting from the mood
manipulations accounted for the increased rumination. This provides the first
experimental evidence that both angry and sad moods cause repetitive, uncontrollable

46
thinking. In addition to confirming untested theories that rumination follows from angry
(e.g., Miller et al., 2003) and sad moods (e.g., Nolen-Hoeksema, 1991), these findings
suggest that rumination is not associated with one particular negative emotion. Rather,
rumination likely arises in response to a range of negative emotions and experiences.
Future experiments should whether rumination also follows from other induced moods
such as fear and shame.
Interestingly, a significant effect of mood manipulation emerged only for the
subscale of rumination that assessed the degree of current negative thoughts, but not for
the subscale assessing the extent of repetitive and uncontrollable thinking. An evaluation
of means suggests that participants in the sad and angry conditions reported similar
amounts of repetitive thinking and negative thoughts. By contrast, participants in the
neutral condition reported somewhat more repetitive thinking than they did negative
thoughts, which likely accounts for the non-significant differences on the measure of
repetitive thinking. Considering the fact that the angry and sad mood manipulations
involved writing about personal memories involving sad or angry emotions, it makes
sense that these participants would report both negative thoughts and repetitive thoughts.
The neutral mood manipulation, by contrast, involved writing about a benign memory,
which might understandably lead to increased repetitive thinking but not increased
negative thoughts.
Self-focused attention
Contrary to expectations, there were no effects of the mood manipulations on self-
focused attention. Moreover, neither self-reported angry and sad moods nor increased

47
rumination were correlated with self-focused attention. These results conflicted with
previous findings that sad mood evokes internal self-focus (Mor & Winquist, 2002;
Salovey, 1992; Wood et al., 1990). Past research also suggested that angry rumination
would involve a more external, other-oriented cognitive focus (Izard, 1991). The low
internal consistency of the particular measure used in this study could account for the
non-significant results. Alternatively, no study has directly compared the effect of angry
and sad moods on self-focused attention. Perhaps angry and sad mood do not lead to
meaningful differences in self-focused attention. Moreover, in two recent internet
studies, Silvia and colleagues (2005) found that the Rumination and Reflection Scale
(Trapnell & Campbell, 1999) was unrelated to word-recognition or pronoun-selection
measures of self-focused attention. Thus, perhaps rumination does not correlate with
independent measures of self-focused attention. Clearly, self-focused attention in this
study cannot account for the differential effects of angry and sad moods on activation of
aggressive networks.
Accessibility of aggression-related constructs
According to one hypothesis discussed above, only angry mood and subsequent
rumination should activate aggressive associative networks. However, an opposing
hypothesis suggests that rumination arising from both sad and angry moods should
trigger aggressive thoughts, feelings, and behaviors. The associations between
rumination and the cognitive accessibility of aggression-related constructs failed to
convincingly support either hypothesis.

48
Contrary to expectations, for participants in the angry group, increased rumination
did not relate to the cognitive accessibility of aggressive actions, physiological arousal, or
angry affect. However, participants receiving the sad mood manipulation who reported
more rumination also reported greater accessibility of arousal-related words. This
correlation could support previous findings about sadness-related rumination. Some
researchers argue that depressive rumination is inherently self-focused, whereas angry
rumination is focused on other people and perceived provocations (Bushman et al., 2005;
Nolen-Hoeksema, 1991; Pedersen et al., 2005; Pyszczynski & Greenberg, 1987). In
support of this theory, Pedersen and colleagues (2005) found that participants induced to
ruminate about themselves reported increased accessibility of arousal-related words like
jittery and tense. Thus, the current results could be seen to confirm these previous
findings and suggest that self-focused, sadness-related rumination increases awareness of
internal arousal.
However, the correlation between rumination and awareness of arousal in the sad
group might be a spurious correlation. Angry mood also correlated with rumination and
the accessibility of arousal-related words in this group. Mediation analyses suggested
that rumination did not explain the link between angry mood and cognitive accessibility
of arousal words. Thus, it is possible that feelings of anger simply simultaneously
increased rumination and accessibility of arousal constructs for these participants. Future
studies are needed to clarify the direction of associations between negative moods,
rumination, and awareness of internal arousal.

49
In general, the measure of cognitive accessibility did not perform as expected.
There were no effects of the mood manipulations on the accessibility of aggression-
related constructs. This failure to find group differences conflicts with previous research
suggesting that anger results in increased aggressive inclinations and approach
motivation (e.g., Harmon-Jones & Sigelman, 2001; Kenworthy et al., 2003; Larsen &
Diener, 1992; Tiedens, 2001). It also conflicts with Berkowitz’s (1990) hypothesis and
limited experimental findings (Finman & Berkowitz, 1989) that both sad and angry, but
not neutral, experiences can stimulate anger-related associative networks. This lack of
results could be due to a small range on this measure. Thus, continuous measures of self-
reported state mood had the potential to explain more variance in cognitive accessibility
of aggressive constructs.
In no group did the angry mood index correlate with cognitive accessibility of
aggressive actions or angry affect. Thus, subjective awareness of anger was not
associated with implicit anger or aggression. Previous studies have also found small or
non-significant associations between implicit and explicit measures of anger (Ayduk,
Mischel, & Downey, 2002; Greenwald & Farnham, 2000). Researchers in fact argue that
people function in both explicit (e.g., controlled, aware, reflective) and implicit (e.g.,
automatic, intuitive) modes and that these two ways of processing social information
constitute separate constructs (see Greenwald & Farnham, 2000).
Reported angry mood did correlate with increased awareness of internal arousal,
although this appeared only in the sad group. The small range on arousal words for
participants in the angry condition might explain the lack of a similar pattern of

50
correlations for this group. Paradoxically, it is possible that that angry mood
manipulation was actually too effective in eliciting anger. Perhaps participants in this
group were able to keep their anger from affecting other measures because they knew that
the anger was in response to the manipulation. By contrast, the other two mood
conditions did not obviously pull for anger. Therefore, those participants who did report
anger likely did so because of individual tendencies to respond to stimuli with anger. For
these individuals, responding with anger may be associated with increased arousal and a
hostile attribution bias. The strength of the angry manipulation may have overshadowed
these relationships for participants in that condition.
Also surprising was the association between positive mood and increased
cognitive accessibility of aggressive actions exhibited by participants in the neutral
condition. In this group, participants who were more content exhibited greater
accessibility of words like “hit” and “fight.” Although counterintuitive, these results do
accord with research suggesting that happiness is an approach-oriented emotion
(Harmon-Jones & Seligman, 2001). In fact, several similarities have been found between
anger and happiness, such as reliance on heuristics, biased judgments, and a sense of
certainty (e.g., Bodenhausen, Sheppard, & Kramer, 1994; Tiedens & Linton, 2001).
Thus, it is possible that happy mood increases behavioral readiness for aggression.
However, the lack of any correlations between positive mood and aggression measures in
these data fail to support this conclusion. In fact, there is no evidence from this study that
angry and positive moods behaved in similar ways. Moreover, why this association
would emerge only for participants in the neutral group is unclear. Future studies should

51
more directly examine the association between positive mood, readiness for action, and
aggressive behavior.
Hostile attributions and aggressive tendencies
Contrary to hypotheses, rumination did not correlate with hostile attributions in
any group. This was the first study to examine the impact of rumination on hostile
attributions. Perhaps pre-existing rumination does not affect attributions about a
subsequent event. It is likely that participants were ruminating about the prior mood
vignettes or the personal memories they recounted, and not about the brief ambiguous
provocations in this particular measure. Perhaps ruminating participants would report
hostile attributions for the event about which they were ruminating. Future studies
should examine this hypothesis.
Rumination also did not result in greater aggressive intentions in any group. The
strong correlation between dispositional aggressiveness and aggressive intentions
suggests that this particular measure may have assessed more trait-based aggression,
rather than situational aggression. The measure of aggressive intentions also had a small
range across all participants, which prevents large correlations. Moreover, in this study
participants responded to hypothetical scenarios of ambiguous provocations. These
scenarios may not have seemed personally relevant to participants. Research suggests
that provocations that are personally threatening are most likely to produce aggression
(Bushman & Baumeister, 1998). Therefore, individuals who are ruminating may require
some personally relevant threat to ruminate about, before exhibiting activation of
aggressive association networks.

52
Interestingly, angry mood but not sad mood was associated with increased hostile
attributions for the entire sample. This result accords with findings that anger is often
associated with hostile inferences and attributions of blame to others (Averill, 1983;
Baumeister, Stillwell, & Wotman, 1990; Tiedens, 2001). In addition, implicit awareness
of arousal also related to more hostile attributions. Previous researchers have found an
attributional bias specifically in boys with reactive aggression, as opposed to
premeditated aggression (Dodge, Price, Bachorowski, & Newman, 1990). Thus,
participants who were feeling especially angry and aroused were more likely to interpret
ambiguous situations as hostile.
Furthermore, the link between angry mood and hostile attributions was mediated
by cognitive accessibility of arousal words. In other words, an increased angry mood
contributed to greater accessibility of words like “agitated” and “tense,” which in turn
contributed to interpreting ambiguous situations with greater hostility. These results
suggest that angry mood, but not sad mood, activated aggressive association networks.
This contradicts Berkowitz’s (1990) theory that both sad and angry moods can trigger
aggressive networks. Rather, the arousing nature of anger (Larsen & Diener, 1992) may
uniquely contribute to increased hostility and aggression.
Gender differences
There were no gender differences in mood, rumination, or activation of aggressive
networks. However, men and women exhibited a different pattern of associations
between several of these constructs. First, greater reported anger was related to more
internal arousal for men but not for women. This accords with previous findings that

53
women’s subjective reports of anger often do not match their levels of physiological
arousal (Labouvie-Vief et al., 2003; Levenson, Carstensen, & Gottman, 1994). By
contrast, men reporting anger also exhibit physiological arousal. Men in general appear
to be more attuned to changes in their bodily states (e.g., Harver, Katkin, & Bloch, 1993).
Pennebaker and Roberts (1992) suggest that men base their subjective affective states on
perceptions of internal states, whereas women are more likely to consider external
factors. Moreover, marriage researchers propose that men experience physiological
arousal in response to conflict as more aversive than women (Levenson, Carstensen, &
Gottman, 1994). Thus, even though men and women reported similar levels of anger in
response to the mood manipulation, men in this study might have been more likely to link
anger with internal arousal.
Moreover, increased rumination was also associated with greater implicit angry
affect for men but not for women. By contrast, rumination correlated with self-reported
angry mood for women but not for men. Previous research has shown that both men and
women who ruminated reported worse psychological distress and depression following
the death of a partner (Nolen-Hoeksema, McBride, & Larson, 1997; Nolen-Hoeksema,
Parker, & Larson, 1994). However, no researchers have yet examined gender differences
in the effect of rumination on anger. It is possible that the process and subsequent
outcomes of rumination differ for men and women. Perhaps rumination makes women
more aware of their internal states, thereby increasing their subjective experience and
reporting of anger. By contrast, men in this sample who ruminated were more implicitly
upset but did not necessarily report worse subjective experiences. Perhaps men’s

54
subjective experience of anger is primarily related to their perceived level of
physiological arousal, as discussed above. They may not explicitly make a connection
between repetitive thinking and anger. Women, on the other hand, may link repetitive
thinking by not physiological arousal to anger. No research to date has examined
whether the content and mechanisms of rumination differ by gender. This would be an
interesting direction for future studies.
Finally, more dispositional aggression was associated with greater rumination and
more self-reported anger for men but not for women. Thus, more aggressive men
responded to the mood manipulations with increased anger and rumination. This pattern
did not emerge for women. No previous research has examined the association between
trait aggression and state rumination, although several of our studies have found
correlations between trait aggression and trait rumination (Borders & Earleywine, 2004;
Borders, Smucker Barnwell, & Earleywine, In press). Future studies should investigate
why aggressive men might be more likely to ruminate in response to upsetting events.
Limitations
This study contained important methodological limitations. First, random
assignment was not entirely successful. A greater proportion of male participants
received the sad mood manipulation. Moreover, male participants in general reported
more trait aggression. However, there were no differences in trait aggression across the
three mood conditions. There were also no gender differences in mood, rumination, or
aggression. Therefore, the effect of this failure of random assignment does not likely
provide a meaningful alternative explanation for the obtained results. Nevertheless,

55
Study 2 ensured that equal proportions of men and women were randomly assigned to
each mood condition.
In addition, describing the study as three separate experiments might have
undermined the effects of mood and rumination on subsequent sections. This might
explain why group differences emerged only for the mood indices and measure of state
rumination, both of which were in the same “section” of the experiment as the mood
manipulation. If participants viewed the remaining two sections as truly separate
experiments, then they might have mentally re-started themselves at the beginning of
each section. This procedure was changed in Study 2.
Finally, Study 1 did not measure actual aggression. As suggested above,
individuals may require some personally relevant threat to ruminate about before
exhibiting increased aggression. All previous experiments have in fact presented
ruminating and non-ruminating participants with a provocation before measuring
aggression (Bushman, 2002; Bushman et al., 2005; Pedersen et al., 2005). Clearly, a
provocation is more personally relevant and threatening than hypothetical scenarios.
Study 2 provided participants with a mild provocation and assessed subsequent
aggressive behavior.
STUDY 2
The second experiment attempted to replicate and extend the results of Study 1.
This study again compared the effects of angry and sad mood manipulations on subsequent
rumination and the accessibility of aggression-related constructs. Based on the results of
Study 1, I expected that both angered and saddened participants would exhibit increased

56
rumination. Because Study 1 did not produce firm conclusions about the effects of angry
and sad rumination on the activation of aggressive networks, Study 2 was intended to test
the original two competing hypotheses again.
In contrast to the first experiment, Study 2 measured actual behavioral aggression in
response to a mild trigger. Rather than make inferences about ambiguous situations, all
participants received a negative evaluation by another student supposedly participating in
the study at the same time. This form of trigger manipulation has been used in several
studies of (Bushman et al., 2002, 2005; Pedersen et al., 2000).
Originally proposed by Dollard (1938), triggered displaced aggression refers to
circumstances in which a person experiences a strong initial provocation (an interpersonal
event that elicits anger) that precludes retaliation and then responds aggressively when
exposed to a second triggering provocation by another person (Miller et al., 2003). For
instance, after a man is publicly berated by his boss, he comes home to find that his dog has
made a mess on the floor and subsequently responds much more harshly than usual (by
kicking his dog, rather than cursing under his breath). Following some earlier provocation
(boss yelling) and related anger, the trigger (dog leaving a mess) elicits a much stronger
response than seems warranted. In contrast to other aggression paradigms, triggered
displaced aggression corresponds to some, if not many, real-life instances of violence.
Spousal and child abuse, riots and gang violence, and even road rage often involve
seemingly inexplicable aggressive outbursts in response to trivial events. Thus, the
ecological validity and applicability of triggered displaced aggression makes it a good
choice for laboratory investigation.

57
Research suggests that the interaction between an initial provocation and
subsequent minor trigger leads to disjunctively escalated aggressive responding
(Pedersen, Gonzales, & Miller, 2000). In the absence of an initial angering event, a mild
trigger will fail to produce any aggression. However, when preceded by a provocation,
that same trigger will elicit an aggressive response that seems incommensurate with
either the provocation or the trigger. Because the meaning and intentionality of a mild
trigger is ambiguous, it is more susceptible to attribution distortions (Duncan, 1976). The
initial provocation likely primes an individual to make these attribution distortions of low
intensity, trivial triggers (Higgins & King, 1981).
As reviewed above, participants who ruminate exhibit more displaced aggression in
response to the mild trigger than participants who are distracted (Bushman et al., 2005;
Pedersen et al., 2005). However, all previous studies manipulated rumination after
presenting participants with an initial provocation. As discussed above, this double-
manipulation may be stronger than that which occurs naturally and may mask important
individual differences in how individuals respond to provocation. Moreover, no studies of
triggered displaced aggression have examined the effect of a negative mood induction
rather than an initial provocation. This situation might correspond to an individual simply
being in a bad mood for whatever reason, ruminating about it, and subsequently becoming
disjunctively angry and aggressive about a minor annoyance. The initial bad mood need
not result from some personal attack or threat, but perhaps from an upsetting memory,
underlying work pressure, or general annoyance with any number of daily activities. Thus,
in this study the negative mood manipulations substituted for an initial provocation.

58
In line with previous research, I expected that angered participants would respond
to a minor trigger with more aggression than saddened or neutral participants. In addition,
participants who were already ruminating should respond to a mild trigger with increased
anger and aggression. If only anger-related rumination leads to the activation of aggression
associative networks, then rumination should correlate with aggression only for participants
in the angry condition. However, if rumination resulting from any negative mood activates
aggression networks, then rumination should predict aggression for participants receiving
both the angry and sad mood manipulations.
Finally, participants reported on their current mood twice during the experiment.
As in the first study, they reported on their mood immediately following the mood
manipulation procedure. In addition, they reported on their mood again following the
mild trigger. This design allowed for an examination of whether rumination initially
following the mood manipulation led to worse moods at the end of the experiment.
Method
Participants and Design
Sixty students (43 women and 17 men) recruited from undergraduate psychology
courses participated in the study for course credit. Participants ranged in age from 17 to
43 (M=20.13, SD=3.51). Sixty-three percent of participants were Caucasian, 8% were
African-American, 8% were Hispanic, 16% were Asian, and 5% were of another
ethnicity. Approximately 82% of participants were native English speakers. Participants
were randomly assigned to the angry, sad, or neutral mood condition in a between-
subjects design.

59
Procedure
Participants were run individually. Participants were told that they were
participating in a language experiment. Identical to Study 1, the first section involved
mood and writing style, and the second section involved evaluating a language measure.
This time, the third section was described as a “decision making” experiment that
investigated how people make decisions in varied contexts. Participants engaged in
several decision-making tasks, one of which involved interacting with another student
who was supposedly participating in the same study down the hall. The trigger
manipulation and aggression measures were embedded in this third section (see Figure
6).
First, angry, sad, and neutral moods were induced in the same manner as Study 1,
using both vignettes and autobiographical writing. Immediately following the mood
manipulation (Time 1), participants completed shortened mood manipulation and
rumination measures. Unlike Study 1, the mood manipulation checks consisted of single
items (happy, angry, sorrowful). Moreover, participants only answered two rumination
items (“My mind is harping over and over again on something” and “I am playing out in
my mind how and why a negative event happened”). These measures were shortened to
help ensure that the manipulated moods did not dissipate before the end of the study.
Participants also completed the same word-completion measure used in Study 1.

60
Figure 6. Procedure for Study 2.

Section 1 instructions: Read two mood-related vignettes and then write an
autobiographical narrative about a related mood.

Mood manipulation:
angry, sad, neutral

Measured variables:
Angry, sad, and positive moods (Time 1)
State rumination

Section 2 instructions: Provide feedback on two new measures of word choice.

Measured variables:
Accessibility of aggression-related
constructs

Filler questions:
Opinions about measure

Section 3 instructions: Participate in a joint decision-making task with other participant.
Think of 6 characteristics of a NASA astronaut and then evaluate each other’s responses.

Trigger manipulation:
Negative feedback by partner on NASA task

Aggression measures:
Behavioral aggression (number of seconds
partner should hold hand in ice water)
Evaluation of partner
Emotional reaction to trigger

Final measures:
Angry, sad, and positive moods (Time 2)
Trait rumination
Trait aggression

61
Next, participants completed a “joint decision-making” task, during which they
and their alleged partner brainstormed independently about the six most useful
characteristics for a NASA astronaut, after which they had the opportunity to see each
other’s answers. Each of them was then asked to select, from the 12 total answers
proposed, the 3 best characteristics for an astronaut.
First, the experimenter instructed participants that they had three minutes to think
of 6 useful characteristics of an astronaut. Participants were instructed to write their
answers on an answer sheet and then place the answer sheet in an envelope. After three
minutes, the experimenter collected participants’ answers and took them to their alleged
partner. The experimenter returned with their partner’s answers and gave participants
both the envelope with the partner’s answers and a separate evaluation sheet. The
evaluation sheet asked participants to consider their own answers as well as the other
participant’s answers and select the 3 best characteristics for an astronaut. The form also
contained several questions asking participants to evaluate their partner’s answers.
Finally, the form contained several blank lines for indicating an overall impression of
their partner and any additional feedback or evaluation. After completing the evaluation
form, participants were instructed to put both the form and their alleged partner’s answers
back into the envelope.
After a few minutes, the experimenter returned, took the envelope with
participants’ evaluation of their partner’s answers, and handed participants an envelope
containing their partner’s evaluation. The experimenter explained that he or she needed
to leave and prepare for the final part of the study and that participants could read the

62
partner’s evaluations while waiting. The partner's evaluation of participants ranged from
neutral to negative on the specific items. In addition, none of the partner’s three choices
for best astronaut characteristics included the participants’ suggestions. However, the
key feature occurred in the section in which the partner wrote his or her overall
evaluation: “Boring answers. Didn’t put much thought into it.” This evaluation by the
alleged partner constituted the mild trigger manipulation.
After reading their partner’s evaluation, participants were told that the last section
of the experiment involved completing a final language task while being distracted. The
experimenter went on to explain that for the final decision-making task, the participant
and partner would decide the length of distraction for each other. The experimenter
explained that distractions can apply to several sensory modalities (auditory, visual, taste,
or touch) and that for this study both the participant and the partner were randomly
assigned to one kind of these distractions. The experimenter then explained that the
participant was randomly assigned to the visual distraction, which would involve
watching a video of nature scenes, and that the partner would receive the touch
distraction, which involved placing his or her hand in ice water. Therefore, the
experimenter presented the participant with a bucket of ice water and asked the
participant to “try it out” for a few seconds, to see what it would feel like for the partner.
The experimenter then left with the ice water and the participant privately indicated on a
sheet of paper how long the partner should leave his or her hand in the water (0-80
seconds). The length of time the participant indicated constituted a behavioral measure
of aggression.

63
Participants then completed a final packet of questionnaires. The packet included
an evaluative measure assessing their attitude toward their partner (e.g., friendliness,
competence, intelligence). Participants also completed a brief questionnaire about their
emotional reaction (e.g., annoyed, pleased) to their partner’s evaluation of their work.
The packet also contained a measure of state mood that was identical to the questionnaire
used in Study 1. Finally, participants completed the same measures of dispositional
rumination (Trapnell & Campbell, 1999) and aggression (Buss & Perry, 1982) used in
Study 1.
At the end of the study, the experimenter explained that the final language task
was not necessary. Therefore, participants never actually experienced any distraction.
The experimenter debriefed all participants and probed for suspicion.

Results
There were no multivariate or univariate outliers. All variables met assumptions
for normality except for the final sad mood index. A log transformation of this variable
was used in all statistical analyses
2
. For ease of interpretation, all means and standard
deviations reported below are from non-transformed variables. Initial analyses revealed
no main effects or interactions with gender for any dependent measure. Therefore, the
data were collapsed across gender and run in one-way (Mood Condition: Sad / Angry /
Neutral) between-subject analyses of variance. Correlations between study variables for
the entire sample (Table 4) and for separate mood conditions (Table 5) are also reported.

2
The skewness and kurtosis statistics of the sad mood index were 2.32 and 5.77, respectively, before
lognormal transformations and 1.02 and .84 after.
64
Table 4. Correlations between Study 2 variables, for entire sample (N=60).

Measure 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1. Sad mood T1 —
2. Angry mood T1 .41^ —
3. Positive mood T1 -.50^ -.46^ —
4. State rumination .56^ .37** -.51^ —
5. Action words .19 .01 -.08 -.03 —
6. Arousal words .27* .06 -.22 .21 .15 —
7. Affect words -.07 -.16 .04 -.13 .30* -.15 —
8. Behavioral aggression .12 -.01 -.04 .29* -.04 -.19 -.05 —
9. Evaluation of partner .07 .16 -.21 .33** -.12 -.08 .12 .44^ —
10. Emotional response .18 .27* -.10 .27* .04 -.10 .27* .24 .48^ —
11. Sad mood T2 .44^ .26* -.14 .39** .10 .19 -.07 .04 .16 .17 —
12. Angry mood T2 .23 .33** -.07 .32** .25* .06 .20 .34** .37** .64^ .30* —
13. Positive mood T2 -.21 -.11 .23 -.30* -.07 -.09 -.21 -.08 -.29* -.35** -.18 -.36** —
14. Trait rumination .20 .35** -.41** .36** -.10 .16 -.07 .07 .21 .33** .26* .33* -.32* —
15. Trait aggression .19 .23 -.47^ .31* -.10 .10 .07 .15 .23 .19 .11 .25 -.21 .48^ —

* p < 05. ** p < .01. ^ p < .001
64

65
Table 5. Correlations between rumination and Study 2 variables by mood condition.

ANGRY CONDITION (N=18)

Measure 1 2 3 4 5 6 7 8 9 10 11 12 13

1. Sad mood T1 —
2. Angry mood T1 .65** —
3. Positive mood T1 -.35 -.71^ —
4. State rumination .16 .41 -.42 —
5. Action words .24 .32 .02 -.24 —
6. Arousal words .39 .48* -.40 .42 -.17 —
7. Affect words -.25 -.30 .23 .02 .01 -.54* —
8. Behavioral aggression .17 .19 -.02 .27 .12 -.44 .27 —
9. Evaluation of partner .02 .28 -.02 .34 -.05 -.17 .36 .57* —
10. Emotional response .11 .31 .14 .47* .17 -.01 .33 .47* .58* —
11. Sad mood T2 .47* .38 .03 .19 .27 .30 .07 -.03 .11 .28 —
12. Angry mood T2 .24 .51* -.10 .43 .34 .09 .25 .49* .49* .80^ .60** —
13. Positive mood T2 .16 -.13 -.07 .01 -.01 -.07 -.21 -.02 -.48* -.27 -.28 -.37 —

* p < 05. ** p < .01. ^ p < .001.
65
66
SAD CONDITION (N=21)

Measure 1 2 3 4 5 6 7 8 9 10 11 12 13
1. Sad mood T1 —
2. Angry mood T1 .17 —
3. Positive mood T1 -.27 -.41 —
4. State rumination .74^ .14 -.29 —
5. Action words -.15 -.31 -.10 -.10 —
6. Arousal words .37 -.13 -.20 .22 .49* —
7. Affect words -.30 -.17 .08 -.38 .30 .19 —
8. Behavioral aggression .16 -.37 .15 .15 -.02 .17 -.03 —
9. Evaluation of partner .18 .13 -.01 .20 -.45* -.11 -.04 .45* —
10. Emotional response .45* .24 -.09 .03 -.34 -.13 .20 .05 .31 —
11. Sad mood T2 .13 .23 .09 .30 -.41 -.08 -.22 -.06 .27 .10 —
12. Angry mood T2 .39 .08 .01 .12 .12 .27 .17 .15 .15 .47* .19 —
13. Positive mood T2 -.66** -.17 .50* -.49* -.08 -.36 .00 -.01 -.27 -.45* -.20 -.29 —

66
67
NEUTRAL CONDITION (N=21)

Measure 1 2 3 4 5 6 7 8 9 10 11 12 13
1. Sad mood T1 —
2. Angry mood T1 .19 —
3. Positive mood T1 -.31 .10 —
4. State rumination .17 .11 -.54* —
5. Action words .57** .30 -.10 .17 —
6. Arousal words .34 -.08 -.11 .14 .19 —
7. Affect words .43* .05 -.19 .06 .53* -.07 —
8. Behavioral aggression -.35 .01 -.12 .51* -.26 -.22 -.44* —
9. Evaluation of partner -.05 .19 -.53* .57* -.01 .02 .05 .30 —
10. Emotional response .19 .32 -.37 .44* .26 -.18 .32 .13 .55* —
11. Sad mood T2 .34 -.02 -.11 .38 .49* .50* .04 .17 .07 .27 —
12. Angry mood T2 .06 .49* .00 .43* .27 -.12 .19 .33 .42* .62* .19 —
13. Positive mood T2 -.21 .04 .23 -.37 -.12 .11 -.39 -.20 -.17 -.32 -.09 -.41 —

67

68
Random assignment checks
As expected, there were no differences across groups in trait rumination, F(2,
53)=.52, ns, or aggression, F(2, 53)=.37, ns. Moreover, men and women were
proportionally assigned to each condition, χ
2
(2, N = 60) = 1.53, ns. Thus, random
assignment successfully matched participants on gender, dispositional rumination, and
aggression.
Mood manipulation checks
As expected, the three conditions differed in sad mood, F (2, 57) = 35.33, p
<.0001, angry mood, F (2, 57) = 11.29, p<.0001, and positive mood, F (2, 57) = 6.78, p <
.01. Specifically, participants in the sad condition reported more sad affect (M=4.95,
SD=1.50, range=2-7) than participants in the angry (M=2.50, SD=1.46, range=1-5) and
neutral (M=1.67, SD=.91, range=1-4) conditions (both pairwise p-values < .0001). By
contrast, participants in the angry condition reported significantly more angry affect
(M=3.28, SD=1.64, range=1-6) than participants in the neutral condition (M=1.33,
SD=.66, range=1-3) but not significantly more than participants in the sad condition
(M=2.71, SD=1.52, range=1-5). Finally, participants in the neutral condition reported
significantly more positive affect (M=4.19, SD=1.37, range=1-7) than participants in the
sad condition (M=2.67, SD=1.28, range=1-5) but not more than participants in the angry
condition (M=3.61, SD=1.42, range=1-6).
Across the entire sample, positive mood exhibited significant negative
correlations with sad mood and angry mood (see Table 4). Sad and angry moods were
positively correlated. For angered participants only, angry mood related to more sad

69
mood and less positive mood (see Table 5). The correlation between sad and positive
mood was large but not significant. For participants in the sad and neutral conditions,
none of the intercorrelations between initial moods reached significance.
State rumination
As predicted, analyses yielded an overall effect of state rumination, F (2, 57) =
7.07, p<.01 (see Figure 7). Participants in the sad (M=4.26, SD=1.88) and angry
(M=3.64, SD=1.51) conditions reported more rumination than participants in the neutral
condition (M=2.45, SD=1.30) (both pairwise p-values < .05). The pairwise difference
between the sad and angry conditions was not significant.

Figure 7. Effect of mood manipulation on state rumination (Study 2).
1
2
3
4
5
6
7
Angry Sad Neutral
State rumination

F(2, 57) = 7.07, p < .01

As in Study 1, I performed mediation analyses to determine whether participants’
self-reported moods following the mood manipulation accounted for their reported levels

70
of state rumination. Dummy variables were created as before to represent group
assignment. Thus, the sad dummy variable compared participants in the sad condition to
participants in the neutral condition. The angry dummy variable compared participants in
the angry and neutral conditions.
The first set of regression analyses evaluated whether sad mood mediated the
effect of mood condition on rumination (see Figure 8a). First, the regression coefficients
for both the sad (b=1.81) and angry (b=1.19) dummy variables were significant (both p-
values < .05), confirming that group assignment predicted state rumination. Second, only
the sad dummy variable predicted sad mood, b=1.70, p < .0001. The coefficient for the
angry dummy variable was not significant, b=.43, ns. Thus, sad mood could only
mediate the effect of the sad condition on rumination. Third, when rumination was
regressed on the sad dummy variable and sad mood together, sad mood predicted state
rumination, b=1.14, p < .0001. Moreover, the effect of the sad mood condition dropped
to non-significance, b=-.45, ns. The bootstrap estimate of the indirect effect was 1.71
(SE=.49, CI=.83-2.74), suggesting that sad mood mediated the effect of the sad mood
condition on rumination. Finally, when sad mood and rumination were reversed in the
last equation, the sad mood manipulation still predicted sad mood, b=1.39, p < .0001.
There was therefore no evidence that rumination mediated the link between sad mood
condition and sad mood.
A second set of regression analyses evaluated whether angry mood mediated the
effect of mood condition on rumination (see Figure 8b). Recall that both the angry
(b=1.19) and sad (b=1.81) dummy variables predicted state rumination. This time, both

71
Figure 8. Mood indices mediate the effect of mood manipulations on state rumination
(Study 2).

(a) Sad mood mediates the effect of sad mood manipulation on state rumination

Note: Bolded coefficients are significant. Indirect effects are in parentheses
Mood manipulation:
angry vs. neutral
Sad mood
State
rumination
1.19
.98 .43
Bootstrap estimate of indirect effect = 1.71 (SE = .49)
95% confidence interval = .83 to 2.74

Mood manipulation:
sad vs. neutral
State
rumination
Sad mood
1.81 (-.45)
1.70 .98 (1.14)

72
(b) Angry mood mediates the effect of angry mood manipulation on state rumination

Note: Bolded coefficients are significant. Indirect effects are in parentheses
Mood manipulation:
angry vs. neutral
Angry mood
State
rumination
1.19 (.66)
.64 (.42) 1.26
Bootstrap estimate of indirect effect = .54 (SE = .29)
95% confidence interval = .06 to 1.23
Mood manipulation:
sad vs. neutral
State
rumination
Angry mood
1.81 (1.44)
.89 .64 (.42)

73
the angry dummy variable (b=1.26) and the sad dummy variable (b=.89) predicted angry
mood (both p-values < .01). With rumination regressed on both dummy variables and
angry mood together, however, angry mood failed to predict state rumination, b=.42, p =
.09. Interestingly, although the coefficient for the sad dummy variable remained
significant (b=1.44, p < .01), the regression coefficient for the angry dummy variable was
no longer reliable (b=.66, ns). Moreover, the bootstrap estimate of the indirect effect was
.54 (SE=.29, CI=.06-1.23), suggesting a significant effect of mediation. Furthermore,
when angry mood and rumination were reversed in the final equation, the effect of the
angry manipulation on angry mood remained (b=1.11, p < .0001), ruling out the
possibility of reverse mediation. Thus, the overall pattern of results suggests that angry
mood did mediate the effect of the angry mood manipulation on state rumination. The
one non-significant analysis could reflect the small range in reported initial angry mood
in the neutral group.
For the entire sample, more reported rumination correlated with more sad and
angry moods and less positive mood (see Table 4). When separated by mood condition,
these correlations all remained in the same direction although few reached significance
(see Table 4). Specifically, rumination covaried with increased sad mood for saddened
participants and with decreased positive mood for neutral participants.
Accessibility of aggression-related constructs
The three word types (i.e., action, arousal, and affect) were initially combined to
form a single measure of the accessibility of aggression related cognitions. A one-way
(Mood Condition: Sad / Angry / Neutral) ANOVA yielded no overall effect, F(2, 57) =

74
.91, ns. On average, participants completed 32% of the word stems with aggressive words
(SD=.12, range=.09-.60). To separately examine the 3 distinct word types, a 3 (Mood
Condition: Sad / Angry / Neutral) by 3 (Word Type: Action / Arousal / Affect) mixed
ANOVA with repeated measures on the last factor was conducted. No main effects or
interaction emerged (all p-values > .35).
Across all participants, accessibility of action- and affect-related words were
correlated (see Table 4). Moreover, more reported sad mood correlated with greater
accessibility of arousal-related words. No associations emerged between rumination and
accessibility of aggression-related constructs. This pattern of correlations varied,
however, between different mood groups (see Table 5). For angered participants, greater
accessibility of arousal- and affect-related words were negatively correlated. Moreover,
angry mood covaried with more accessibility of arousal-related words. The correlations
between arousal words and sad affect, positive affect, and rumination were large but non-
significant. For participants in the sad condition, by contrast, action and arousal words
were positively correlated. Moreover, there were no significant correlations between
words, rumination, or initial mood. Finally, action and affect words showed a significant
positive correlation for the neutral group. In addition, greater sad mood covaried with
accessibility of action- and affect-related words.
Aggression variables
Participants’ rating of how long their partner should keep his or her hand in ice
water served as a behavioral index of aggression. Participants selected a number from 1
(no distraction at all) to 9 (80 seconds ice water; very strong distraction). Analyses

75
showed no effect of mood condition on behavioral aggression, F(2, 57) = .32, ns. Across
all three conditions, participants selected an average score of 5.32 (SD=2.87, range=0-9).
Participants also evaluated their partner on several dimensions (e.g., friendliness,
competence, intelligence). Participants responded to all items on a seven-point scale.
These evaluation items were averaged to create a composite variable, with higher scores
indicating worse evaluations (Cronbach’s α = .85). There was no difference between
groups on evaluation of the partner, F(2, 57) = .06, ns. Participants across all groups
scored an average of 3.57 (SD=.81, range=1.7-5.4).
Finally, participants rated their emotional reactions to their partner’s evaluation of
their work (e.g., “I felt annoyed”), again on a Likert scale from 1 to 7. Averaged higher
scores indicated more negative emotional reactions (Cronbach’s α =.86). There was no
difference between groups on this measure, F(2, 57)=.23, ns. Participants in all
conditions rated an average of 5.06 (SD=1.08, range=2.6-7).
For the whole sample and participants in only the angry group, all three
aggression measures intercorrelated (see Tables 4 and 5). For participants in the sad and
neutral conditions, the correlations were in the same direction but did not all reach
significance. Across the entire sample, higher state rumination covaried with increased
aggression on all three aggression measures. Moreover, a more negative emotional
response to the partner related to increased initial anger and accessibility of affect words.
For the angry group, however, only a negative emotional response correlated with
rumination. Moreover, for the sad group, greater initial sad mood also related to a more
negative emotional response to the trigger. Finally, for the neutral mood group,

76
rumination was again correlated with increased aggression across all three measures.
Moreover, greater initial positive mood led to more positive evaluations of the partner.
A few unexpected correlations emerged for participants in specific groups. In the
angry group, for instance, accessibility of arousal words was associated with less
behavioral aggression. Moreover, accessibility of action-related words related to less
negative evaluations of the partner for saddened participants. Finally, in the neutral
condition, accessibility of affect words correlated with less behavioral aggression.
Scatterplots of these unexpected associations did not reveal any outliers,
heteroscedasticity, or non-linear relationships.
Final mood measures
At the end of the experiment (Time 2), the three conditions differed only in sad
mood, F(2, 57)=4.48, p < .05. Participants in the sad condition reported more sad affect
(M=2.36, SD=1.49) than participants in the neutral condition (M=1.48, SD=.52).
Participants in the angry condition (M=1.57, SD=.59) reported less sad mood than
participants in the sad condition and more than participants in the neutral condition,
although these pairwise differences were not significant. There were no effects of
condition on angry, F(2, 57) = .40, or positive mood, F(2, 57) = .06, ns. Across all three
conditions, participants reported a positive mood score of 3.20 (SD=1.14, range=1-5.8)
and an angry score of 2.86 (SD=1.30, range=1-5.4).
To examine the change in mood from post-manipulation to post-trigger, I ran
paired t-tests for each mood index separately by mood condition. To control for
increased Type 1 error with multiple tests, alpha was set to .01 for these analyses. Sad

77
mood decreased over time for participants in both the angry, t(17)=3.05, and sad
conditions, t(20)=6.52, p < .01, but not for the neutral condition, t(20)=.99, ns. Angry
mood increased for participants in the neutral condition, t(20)=-5.28, p < .01, but not for
participants in the angry, t(17)=.89, or sad conditions, t(20)=-.69, ns. Finally, positive
mood did not change for any of the three mood conditions (all p-values > .01).
For all the participants (see Table 4), rumination immediately following the mood
manipulation related to positive, sad, and angry moods following the trigger (Time 2).
Moreover, angry mood at Time 2 correlated with accessibility of action-related words
and all three aggression measures. Positive mood at Time 2 also related to less negative
evaluations and emotional responses to the trigger.
Likewise for participants in the angry mood condition, angry mood at Time 2
related to all three aggression measures, and positive mood at Time 2 related to more
positive evaluations of the partner (see Table 5). In the sad group, more negative
emotional responses to the trigger covaried with more anger and less positive mood at
Time 2. In addition, less rumination after the sad mood manipulation led to greater
positive mood at Time 2. Finally, in the neutral group, angry mood at Time 2 again
covaried with more negative evaluations and emotional responses. Moreover, rumination
correlated with increased angry mood at Time 2. Interestingly, sad mood at Time 2
related to increased accessibility of action- and arousal-related words.
I also ran partial correlations for the entire sample between rumination and mood
at Time 2, controlling for the same mood at Time 1. When controlling for sad mood at
Time 1, rumination was no longer correlated with sad mood at Time 2, partial r = .19, ns.

78
Likewise, when controlling for initial angry mood, rumination no longer covaried with
angry mood at Time 2, partial r = .23, ns. Finally, rumination no longer correlated with
positive mood at Time 2 after controlling for positive mood at Time 1, partial r = -.22, ns.
When performing the same analyses separately by group, many of the zero-order
correlations were not significant due to small sample size. However, I compared any
zero-order correlations larger that .40 between rumination and mood at Time 2 with their
corresponding partial correlations, controlling for the same mood at Time 1. In the angry
group, the zero-order correlation between rumination and anger at Time 2 was .43 (p =
.07), whereas the partial correlation after controlling for anger at Time 1 decreased to .28
(p = .27). In the sad group, the zero-order correlation of -.49 (p = .03) between
rumination and positive mood at Time 2 became a partial correlation of -.41 (p = .07)
after controlling for positive mood at Time 1. In the neutral group, the zero-order
correlation of .43 between rumination and anger at Time 2 became a partial correlation of
.43 after controlling for anger at Time 1 (both p-values = .05).
Mediation analyses
The pattern of results between study variables for the entire sample did not permit
a test of whether rumination mediated the link between initial angry mood and
subsequent aggression. However, it did warrant a test of whether angry mood at Time 2
mediated the link between rumination and behavioral aggression (see Figure 9a). To do
this, I performed three regression analyses. First, rumination predicted behavioral
aggression, b=.84, p < .05. Second, rumination predicted angry mood at Time 2, b=.32, p
< .05. Third, angry mood at Time 2 also predicted behavioral aggression, b=.80, p < .05,

79
even after controlling for rumination. Finally, the regression coefficient for rumination
(b=.58) dropped to non-significance when entered with angry mood at Time 2, suggesting
mediation. However, the bootstrap estimate of the indirect effect (.27) was not reliably
different from zero (SE=.27, CI=.00-.69). Moreover, when behavioral aggression and
angry mood were reversed in the last equation, the effect of rumination on angry mood
was no longer significant, b=.24, p = .06. However, the bootstrap estimate of the indirect
effect for reverse mediation (.17) was also not reliably different from zero (SE=.15, CI=-
.11-.50). Thus, angry mood at Time 2 and behavioral aggression were not statistically
distinguishable by these analyses and the direction of mediation could not be determined.
Participants’ evaluation of the partner also correlated with rumination and angry
mood at Time 2. I repeated the same regression analyses to determine mediation (see
Figure 9b). Again, rumination predicted negative evaluations (b=.27) and angry mood at
Time 2 (b=.32) (both p-values < .05). Moreover, angry mood predicted negative
evaluations (b=.24, p < .05) after controlling for rumination. However, the direct effect
of rumination on evaluations was almost significant (b=.19, p=.06). Accordingly,
bootstrap estimate of the indirect effect (.08) was not reliably different from zero
(SE=.04, CI=.00-.18). In addition, with the evaluation and angry mood reversed in the
last analysis, the effect of rumination on angry mood again dropped from significance
(b=.23, p=.08), suggesting possible reverse mediation. However, the bootstrap estimate
of the indirect effect for reverse meditation (.06) was again not reliably different from
zero (SE=.04, CI=.00-.16).

80
Participants’ emotional reactions to the negative evaluation from the partner
correlated with initial angry mood, rumination, accessibility of negative affect words, and
angry and positive moods at Time 2. The same mediation tests were warranted, given
this pattern of correlations (see Figure 9c). Once again, rumination predicted a negative
emotional response (b=.29, p < .05) and angry mood at Time 2 (b=.32, p < .05).
Moreover, angry mood predicted negative evaluations (b=.67, p < .0001) after controlling
for rumination. This time, the direct effect of rumination was small and not reliable
(b=.07, ns) after controlling for angry mood at Time 2. Moreover, the bootstrap estimate
of the indirect effect was .21 (SE=.10, CI=.04-.41), suggesting that angry mood at Time 2
did mediate the association between rumination and participants’ emotional reaction to
the partner. When emotional reaction to the trigger and angry mood were reversed in the
last equation, rumination no longer predicted angry mood, b=.16, ns. However, the
bootstrap estimate of the indirect effect for reverse meditation (.02) was not reliably
different from zero (SE=.04, CI=-.07-.11). Thus, although reverse meditation was
unlikely, anger and a negative emotional response to the trigger may not be completely
separate constructs.
I performed one final set of mediation analyses to determine whether participants’
evaluations of the partner mediated the association between rumination and behavioral
aggression (see Figure 10). Rumination predicted both behavioral aggression (b=.83, p <
.05) and negative evaluations (b=.33, p < .05). Negative evaluations also predicted
behavioral aggression after controlling for rumination (b=1.11, p < .05). With behavioral
aggression regressed on both rumination and evaluation, the coefficient for rumination

81
Figure 9. Angry mood at Time 2 mediates the link between state rumination and
aggression measures (Study 2).

(a) Behavioral aggression

Note: Bolded coefficients are significant. Indirect effects are in parentheses
State
rumination
Angry mood
Time 2
Behavioral
aggression
.84 (.58)
.99 (.80) .32
Bootstrap estimate of indirect effect = .27 (SE = .27)
95% confidence interval = .00 to .69

82
(b) Evaluation of partner

(c) Emotional response to trigger

State
rumination
Angry mood
Time 2
Emotional reaction
to trigger
.29 (.07)
.70 (.67) .32
Bootstrap estimate of indirect effect = .21 (SE = .10)
95% confidence interval = .04 to .41
State
rumination
Angry mood
Time 2
Evaluation of
partner
.27 (.19)
.30 (.24) .32
Bootstrap estimate of indirect effect = .08 (SE = .04)
95% confidence interval = .00 to .18

83
decreased, b=.47, ns. The bootstrap estimate of the indirect effect (.37) was reliably
different from zero (SE=.17, CI=.08-.75), suggesting that rumination led to more negative
evaluations of the partner, which in turn contributed to greater behavioral aggression.
When behavioral aggression and evaluations of the partner were reversed in the final
equation, the effect of rumination on negative evaluations barely dropped from
significance, b=.18, p = .07. However, the bootstrap estimate of the indirect effect for
reverse meditation (.16) was not reliably different from zero (SE=.14, CI=-.09-.46).

Figure 10. Evaluation of partner mediates the link between state rumination and
behavioral aggression (Study 2).

Note: Bolded coefficients are significant. Indirect effects are in parentheses

State
rumination
Evaluation of
partner
Behavioral
aggression
.83 (.47)
1.27 (1.11) .33
Bootstrap estimate of indirect effect = .37 (SE = .17)
95% confidence interval = .08 to .75

84
Table 5 presents the correlations between study variables for separate mood
groups. Because of small sample sizes, I did not perform mediation tests separately by
group.
Dispositional measures
The average amount of dispositional rumination reported was 3.42 (SD=.85).
Men (M=71.50, SD=13.64) and women (M=66.98, SD=18.47) reported comparable
amounts of dispositional aggression, t(54)=.89, ns. Correlations between dispositional
rumination and aggression and the study variables are presented in Table 4. In general,
participants high in trait rumination reported more sad and angry moods and less positive
mood. They also reported more state rumination. These individuals reported a more
negative emotional response to the partner’s evaluation of their work. Aggressive
individuals reported more state rumination and less happy mood at Time 1. Interestingly,
dispositional aggression did not correlate with any of the dependent aggressive measures.
Gender analyses
Independent t-tests yielded no significant gender differences in any of the study
variables. The study correlations were examined separately for males and females (see
Table 6). Again, because the sample of men (n=17) was smaller than the sample of
women (n=43), any correlations larger than .40 will be interpreted. For men, rumination
and the mood indices at Time 1 were generally intercorrelated. The accessibility of
action and affect words were positively correlated. No significant correlations emerged
between behavioral aggression and any predictor variables. However, negative
85
Table 6. Correlations between rumination and Study 2 variables by gender.

MEN (N=17)

Measure 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1. Sad mood T1 —
2. Angry mood T1 .37 —
3. Positive mood T1 -.58* -.45 —
4. State rumination .64**.45 -.41 —
5. Action words .25 .29 -.27 .04 —
6. Arousal words .38 -.15 -.20 .31 .00 —
7. Affect words .03 -.07 -.05 .00 .47* -.36 —
8. Behavioral aggression-.25 -.24 .22 -.04 -.04 -.01 -.22 —
9. Evaluation of partner .20 -.04 -.35 .41 -.02 .11 .03 .43 —
10. Emotional response .02 .24 -.21 .31 .21 -.10 .38 .18 .42 —
11. Sad mood T2 .28 .27 -.08 .20 -.09 .23 .04 -.50** -.06 -.23 —
12. Angry mood T2 -.03 .24 .20 .12 .29 .06 .17 .24 .21 .40 .40 —
13. Positive mood T2 -.18 .05 .15 -.06 -.09 -.23 .10 .04 -.16 .07 -.13 .02 —
14. Trait rumination .25 .58* -.32 .18 .13 -.02 .01 -.22 .02 .35 .37 .32 .21 —
15. Trait aggression .05 .63** -.49* .26 .26 -.17 -.09 -.23 -.12 .17 .20 .04 .08 .63** —

85

86
WOMEN (N=43)

Measure 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1. Sad mood T1 —
2. Angry mood T1 .42** —
3. Positive mood T1 -.48**-.50** —
4. State rumination .54^ .34* -.58^ —
5. Action words .18 -.06 .02 -.06 —
6. Arousal words .24 .12 -.30 .19 .21 —
7. Affect words -.11 -.19 .11 -.19 .25 -.06 —
8. Behavioral aggression.28 .10 -.17 .45** -.07 -.22 -.01 —
9. Evaluation of partner .02 .24 -.12 .30* -.16 -.13 .15 .45** —
10. Emotional response .24 .28 -.06 .26 .00 -.13 .26 .30* .50** —
11. Sad mood T2 .50**.27 -.16 .46** .14 .22 -.13 .21 .25 .32* —
12. Angry mood T2 .31* .37* -.19 .39** .23 .08 .20 .38* .43** .72^ .26 —
13. Positive mood T2 -.22 -.18 .25 -.39** -.05 -.10 -.31* -.08 -.34* -.51^ -.17 -.47** —
14. Trait rumination .17 .27 -.45** .44** -.17 .24 -.11 .19 .30* .34* .22 .34* -.51** —
15. Trait aggression .23 .14 -.48** .34* -.18 .19 .11 .26 .34* .22 .08 .30 -.27 .43** —
86

87
evaluations of the partner were associated with increased rumination, negative emotional
responses to the trigger, and increased behavioral aggression. Furthermore, angry mood
at Time 2 was associated with more negative emotional responses to the trigger and more
sad mood at Time 2. Oddly, sad mood at Time 2 was associated with less behavioral
aggression. Finally, both trait rumination and aggression covaried with angry mood at
Time 1.
The pattern of correlations for women more closely resembled the pattern of
correlations for the entire sample (see Table 6). Once again, rumination and mood
indices at Time 1 were all intercorrelated. Rumination was also associated with mood
indices at Time 2. Furthermore, increased rumination covaried with behavioral
aggression and negative evaluations of the partner. As expected, all three measures of
aggression intercorrelated. As found with the general sample, angry mood at Time 2
correlated with all three measures of aggression. Increased sad mood only correlated
with more negative emotional responses to the trigger. Finally, trait rumination and
aggression correlated with more rumination, less positive mood at Time 2, and more
negative evaluations of the partner.
Discussion
State rumination
Replicating the results of Study 1, both angry and sad mood manipulations
produced increased rumination. Once again, mediation analyses indicated that sad and
angry moods resulting from the mood manipulations accounted for the increased
rumination.

88
Across the entire sample, increased rumination in turn related to increased angry
and sad moods at the end of the experiment. In both the angry and neutral conditions, as
well, rumination correlated with angry mood following the trigger. However, these
results generally disappeared when controlling for the same mood at Time 1. Thus, the
associations between rumination and mood at the end of the experiment may have
predominantly reflected initial mood. One notable exception occurred in the neutral
group, where the correlation between rumination and anger remained after controlling for
initial angry mood. Clearly, participants in the angry group felt more initial anger than
did participants in the neutral group in response to the manipulations. In this instance,
initial anger explained a large proportion of anger at the end of the experiment.
However, for participants in the neutral group who did not report such strong initial
negative anger, pre-existing rumination made a unique contribution to anger following
the trigger. Unfortunately, because all participants received the mild trigger, it is
impossible to determine whether anger at the end of the experiment resulted from initial
anger, rumination, or the effect of the trigger. Future studies should include a no-trigger
condition in a 3 (mood) by 2 (trigger / no trigger) design, so that the effect of initial mood
and rumination can be separated from the effect of the trigger.
Accessibility of aggression-related constructs
Across the entire sample, rumination showed no links with accessibility of action,
arousal, or negative affect constructs. For participants in the angry group, rumination
was directionally related to increased accessibility of arousal words, although the
correlation failed to reach significance due to small sample size. The general lack of

89
correlations between rumination and this measure is unexpected. Past research suggests
that all three word-types should be more accessible for participants who ruminate
(Pedersen et al., 2005). The low power in this particular study likely prevented more
significant results from emerging. Moreover, past studies have used the same word
completion measure only following an initial provocation and rumination manipulation
(Pedersen et al., 2005). Perhaps rumination following a personally relevant initial
provocation activates aggression-related constructs more powerfully than mood
manipulations evoking a personal memory.
In this experiment, again, the word completion measure showed an inconsistent
pattern of associations with mood. As in Study 1, there was no effect of mood
manipulation on cognitive accessibility of aggressive constructs. Moreover, correlational
analyses were again inconsistent. Across the entire sample, the accessibility of action-
related words related to increased anger at the end of the experiment. Moreover,
replicating Study 1, initial sad mood led to increased accessibility of arousal-related
words. However, in the angry group, more initial anger was related to greater awareness
of arousal. Moreover, sad mood correlated with action, arousal, and affect words in the
neutral condition. Thus, no clear pattern emerged between mood and the accessibility of
aggression-related constructs. It does seem that both angry and sad moods resulted in
increased accessibility of action words like shoot and injure and arousal words like tense
and jittery. However, these results should be treated with caution, as they were not
reliable across groups and across both experiments.

90
Even more surprising, in each mood group one of the word completion subscales
negatively correlated with an aggression measure. This suggests that activation of an
aggression network following the mood manipulation related to less aggression and more
positive evaluations following a mild trigger. Perhaps for social desirability reasons,
participants who felt more internally aroused and aggressive suppressed their emotions
when responding to someone who provoked them. However, any firm conclusions are
unwarranted because the variables involved differed in each group and no consistent
pattern emerged.
The inconsistent pattern of results with this measure in both samples could also
reflect poor psychometric properties. First, internal consistency on the subscales could
not be calculated because so many of the items exhibited zero variance. Previous studies
using this measure did not report internal consistency or correlations among the three
subscales (e.g., Pedersen et al., 2005). However, an inconsistent pattern of correlations
emerged between the three accessibility subscales in these studies, further casting doubt
on the internal consistency of the measure. In general, participants completed only 30%
of the word stems with aggressive words, suggesting low cognitive accessibility of
aggression constructs. Post-hoc item analyses indicated that the percentage of individual
incomplete words completed with aggressive words ranged from 2% to 72% across all 35
items. This suggests that not all of the aggressive word completion options were equally
accessible to participants. Moreover, words that were either not completed or completed
with non-words were excluded when calculating the total percentages used for analyses.
Across all 35 words, the percentage of blank words generally ranged from 0% to 30%,

91
with two words left blank by over 50% of participants. Thus, some incomplete words
may have been more difficult to complete than others. Anecdotal evidence also
suggested that many students, especially those for whom English was not their first
language, had difficulty completing this measure. Thus, the measure may have been too
difficult for many students, which could have interfered with the implicit purpose of the
measure. Future research should better examine the psychometric properties of this
measure.
Aggression
Across the entire sample, rumination predicted increased aggression.
Specifically, participants who ruminated more exhibited more behavioral aggression,
more negative evaluations of their partner, and more negative emotional reactions in
response to the trigger. These results accord with previous studies that manipulated
rumination and found increased aggression (Bushman et al., 2002; Pedersen et al., 2005).
Mediation analyses tested whether angry mood following the trigger mediated the link
between rumination and aggression. Results were unclear about the direction of causality
between angry mood and aggression. It is possible that angry mood following the trigger
led to increased aggression. However, the results could not rule out the possibility that
aggressive responding contributed to more reported anger. Mediation analyses also
suggested that rumination led more negative evaluations of the partner, which in turn
contributed to increased behavioral aggression. Again, however, the direction of
causality between the negative evaluation and behavioral aggression was not confirmed.
These unclear directional effects likely resulted from the strong intercorrelations between

92
measures of aggression, negative emotional responses, and angry affect. In fact, these
various anger and aggression variables may constitute a single underlying construct of
related reactions, all of which increased following rumination. This then constitutes
strong evidence that rumination paired with a trigger can activate aggressive association
networks.
Interestingly, this pattern of results was driven by participants in the neutral and
angry mood conditions but not by participants who received the sad mood induction.
Rumination only predicted aggression measures for the angry group and especially the
neutral group. In fact, whereas increased anger following the trigger correlated with all
aggression measures for the angry and neutral groups, anger only related to reports of
more negative emotional responses for participants in the sad condition. These results
could imply that sad-related rumination does not as easily activate anger and aggression
in response to a minor provocation. Participants in the sad mood condition did report
more sad affect following the trigger than other groups, further suggesting differences
between sad and angry forms of rumination. Emotion researchers in fact argue that
sadness-related thoughts increase withdrawal tendencies (Carlson & Miller, 1987;
Dickson & MacLeod, 2004; Green & Sedikides, 1999). Moreover, sad thoughts are often
more past-focused than anxious or neutral thoughts (Papageorgiou and Wells, 1999).
Thus, perhaps rumination resulting from sad mood turns individuals’ attention away from
the current situation, resulting in attenuated reactions to triggers. Future researchers will
need to test these hypotheses.

93
Unlike previous research of triggered displaced aggression, participants in the
angry mood condition did not exhibit significantly more aggression or more negative
evaluations in response to a mild trigger (Bushman et al., 2002, 2005; Pedersen et al.,
2005). Even studies without rumination manipulations found that an initial provocation
leads to increased anger, hostile thoughts, and aggressive behavior in response to a mild
trigger (Pedersen, Gonzales, & Miller, 2000). One difference is that other studies used a
direct initial provocation, in the form of negative feedback by the experimenter on a
difficult task. By contrast, participants in this study read two vignettes involving
angering events and then wrote about an angering memory of their own.
Miller and colleagues’ (2003) theoretical model of triggered displaced aggression
specifies that the initial angering event must be stronger than the following trigger.
Moreover, provocations that are especially ego-threatening are theorized to elicit greater
aggression following a mild trigger. Because participants in these experiments wrote
about a personal emotional memory, it is unclear whether the participants in the angry
condition experienced an ego-threatening manipulation. Thus, the anger manipulation in
this study may not have been as strong as initial provocations in previous studies (e.g.,
Pedersen et al., 2000). It is also possible that the mild trigger was more salient than the
mood manipulation, especially because the trigger was an immediate event that happened
directly to the participants, whereas the mood manipulations involved theoretical
scenarios and an event from the past. In fact, analyses indicated that the trigger
manipulation erased any angry mood differences that existed from the mood
manipulations. Specifically, participants in the neutral conditions became just as angry

94
after the trigger as participants in the angry and sad conditions already were from the
mood manipulation. Therefore, the salience of the mild trigger may have been greater for
participants in this study than in previous research, leading to no effect of initial mood on
aggression.
Gender differences
Once again, men and women exhibited different patterns of correlations between
study variables. Unlike Study 1, rumination was associated with initial levels of anger
for both men and women. However, rumination correlated with reported moods after the
perceived provocation for women only. By contrast, men who ruminated did not report
increased anger or sadness at the end of the study. Moreover, men who ruminated also
did not exhibit increased behavioral aggression, whereas women who ruminated
evidenced more aggression. This again suggests that rumination may function differently
for men and women. In general, rumination did not impact men’s subjective affect or
aggressive responding as much as it did for women. In a previous study, we found that
gender moderated the association between rumination and alcohol-related aggression
(Borders, Smucker Barnwell, & Earleywine, In Press). The interaction suggested that
rumination leads to increased aggression for women but not for men. By contrast,
Bushman and colleagues (2005) found no gender differences in aggression for
participants who received a rumination manipulation. Clearly, more research is needed
on whether gender moderates the effect of rumination on subsequent anger and
aggression.

95
Interestingly, men who felt sad following the trigger exhibited less behavioral
aggression then men not reporting sadness. This result did not appear for women.
Moreover, anger following the trigger was less strongly associated with aggression for
men than it was for women. In fact, men who were angry only reported a more negative
emotional response to the trigger. By contrast, women who were angrier after the trigger
exhibited more aggression, negative evaluations of the confederate, and negative
emotional responses to the trigger. Finally, angry and sad moods at Time 2 were
correlated for men but not for women. Taken together, these results suggest an
interesting conclusion. Men who responded to the trigger with anger also responded with
sadness, and this sadness in turn related to less behavioral aggression. In other words, the
sadness may have inhibited men from acting out their anger. By contrast, women who
responded to the trigger with anger also exhibited increased aggression but did not report
sadness. Researchers have shown that sad mood typically elicits avoidance of
interaction, withdrawal, and decreased initiative (Carlson & Miller, 1987; Dickson &
MacLeod, 2004; Green & Sedikides, 1999; Harmon-Jones & Sigelman, 2001). However,
to my knowledge no studies have investigated whether a simultaneous sadness inhibits
the expression of anger. This would be an interesting direction for future research, as sad
and angry feelings often co-occur.
Limitations
This study shared some methodological limitations with Study 1. First,
participants completed dispositional measures of rumination and aggression at the end of
the study, rather than at the beginning. The rationale for this procedure was to prevent

96
participants from becoming aware of the study variables of interest. However,
completing personality measures after an entire procedure that included mood
manipulations and aggressive triggers likely affected how participants responded on these
trait measures. Thus, any results with these measures are difficult to interpret.
Specifically, the significant correlations between trait rumination, state rumination, and
state mood could reflect the impact of underlying tendencies for rumination on present
rumination and mood. However, the correlations could also reflect the impact of
situational rumination and mood on how participants perceived themselves generally.
The same possibilities exist for the correlations found between trait aggression, state
mood, and state rumination. Future studies should obtain measures of trait rumination
and aggression before the experimental procedure, preferably not in the same session.
Second, several questions exist about the mood manipulations used in these
experiments. In both this study and Study 1, participants in the neutral mood condition
scored higher on positive mood than participants in the sad and angry conditions. These
results are relatively common with neutral mood manipulations (Kenworthy et al., 2003;
Park & Banaji, 2000). Moreover, participants in the neutral mood condition scored on
average just above the mid-point of the seven-point scale, suggesting that they were not
actively happy. However, because mood was not measured prior to the mood
manipulation, it is possible that the neutral manipulation in fact increased participants’
positive moods. Future studies should include baseline measures of mood to ensure that
participants did not differ from each other before the mood manipulations.

97
The external validity of the neutral manipulation is also less obvious than the
other mood conditions. Numerous descriptive studies have found that people think
repeatedly about sad and angry memories. It is less clear whether individuals naturally
relive neutral memories, although some evidence suggests that people do ruminate about
positive memories (Segerstrom et al., 2003). Thinking about a day off of work would
likely not constitute a strong positive memory for most people, however. Thus, the
external validity of the neutral manipulation in these studies is unclear. To better answer
these issues, it would be helpful to examine exactly what people wrote in their personal
narratives. In the future, the data from both of these studies will be coded along a variety
of content and affect domains. This should help highlight the content and process of
participants’ thoughts in all mood groups.
Furthermore, the particular mood manipulations used in these studies are similar
to rumination manipulations used in previous research. Specifically, rumination
manipulations often involve asking participants to write about a series of self-referential
statements or about a particular provocation for either 8 or 20-25 minutes (Ayduk,
Mischel, & Downey, 2002; Bushman et al., 2005; Pedersen et al., 2005; Rusting &
Nolen-Hoeksema, 1998). These writing exercises generally follow either an actual
provocation or an angry mood manipulation that involved reading an angering story.
These procedures are similar to the mood manipulation used here of reading a story and
then writing about a similar personal experience for ten minutes. Thus, although these
experiments intended to manipulate mood and measure subsequent naturally-occurring

98
rumination, it could be argued that they in fact manipulated rumination. I will address
this issue in more detail below.
Finally, the large number of analyses reported for both studies likely inflated the
risk of Type 1 error. Moreover, the small sample size in Study 2 in particular limited the
power of all analyses. Thus, in many instances above, the size of the observed effect was
considered as much as the significance of the statistic. Obviously, those effects
significant at an alpha level of .01 or below are clearly the most robust results.
GENERAL DISCUSSION
A pair of studies examined the impact of induced negative moods on state
rumination and the subsequent effect of rumination on the activation of aggressive
association networks. Both studies found that angry and sad mood manipulations
produced increased rumination, compared to neutral mood. Both studies also found that
self-reported angry mood generally led to increased aggression. Although Study 1 did
not find an association between increased rumination and hostile attributions or
aggressive intentions, Study 2 showed that increased rumination predicted increased
behavioral aggression, more negative evaluations, and greater emotional negativity in
response to a minor trigger. Furthermore, although rumination led to increased angry and
sad moods following a mild trigger, only angry mood mediated the association between
rumination and increased aggression. Across both studies, inconsistent results emerged
about the relationship between rumination, mood, and the implicit awareness of
aggression-related constructs. Moreover, a measure of self-focused attention failed to
distinguish between anger- and sadness-related rumination.

99
These results provide several important contributions to the literature. This is the
first study to manipulate both sad and angry moods and measure subsequent rumination.
Researchers have previously just assumed or theorized that situational rumination arises
from negative moods. However, to study the effects of rumination in a laboratory setting,
researchers have relied exclusively upon rumination manipulations (for a review, see Mor
& Winquist, 2002). Unfortunately, rumination manipulations vary dramatically from
study to study (for a review, see Borders & Earleywine, 2006). Moreover, because most
studies pair a rumination manipulation with a mood manipulation, the subsequent effects
are confounded and likely larger than occur in everyday life. Thus, the external validity
of these manipulations is doubtful. Therefore, the ability to manipulate mood and simply
measure the resulting rumination is an important advancement in improving how
rumination is studied in laboratory-based studies.
This study was also the first study to my knowledge to directly compare the
effects of rumination on sad and angry moods and related outcomes. Other correlation
studies have compared how angry and sad moods related to various measures of
rumination (e.g., Thomsen, Mehlsen, Christensen, & Zachariae, 2003). However,
rumination researchers have generally studied the effect of rumination on either
depression or aggression, but seldom on both outcomes together (for reviews, see Borders
& Earleywine, 2006; Papageorgiou & Wells, 2004). The current studies showed that
rumination generally increased both sad and angry moods. However, because rumination
and the presence of a trigger were not manipulated, it was impossible to disentangle the
effects of initial mood, rumination, and the trigger on subsequent mood. The moods that

100
followed from rumination did appear to have different consequences. Specifically, only
angry mood mediated the link between rumination and aggression. Rumination that led
to sad mood did not relate to most aggression dependent variables. These results suggest
that important insights will be learned from comparing the impact of rumination on
separate negative moods and related outcomes.
This study also provided some interesting results about gender differences in the
association between rumination, anger, and aggression. In both studies, women’s
rumination was associated with self-reported negative mood and subsequent aggression.
By contrast, rumination in men was associated only with implicit angry affect but not
with self-reported mood or aggression. These results accord with some previous findings
that rumination increased alcohol-related aggression for women but not men, even after
controlling for dispositional aggression (Borders et al., In Press). It seems possible,
therefore, that rumination functions differently for men and women. Future studies
should compare the content of men’s and women’s rumination, to determine whether
different content might explain these differences.
The current results also provide important information about triggered displaced
aggression. First, an angry mood induction may not provide a strong enough initial event
to produce the hypothesized escalated aggression in response to a mild trigger (see Miller
et al., 2003; Pedersen et al., 2000). It may be that triggered displaced aggression occurs
only following a strong personal attack, rather than a general angry mood. Given the
number of angry moods that an individual may experience in the course of any given day,

101
this may be good news. Future studies should compare an angry mood induction to a
strong provocation to reach more firm conclusions.
However, these results also showed that people who were already ruminating but
had not experienced an initial strong provocation still responded to a mild trigger with
more anger and aggression than people who reported lower amounts of rumination. These
results conflict with previous findings that induced rumination in the absence of an initial
provocation did not lead to increased aggression (Bushman et al., 2005; Pedersen et al.,
2005). One possible explanation for these divergent results could be differences in
rumination manipulations. Previous studies have asked people to write about what has
occurred so far in the experiment (Pedersen et al., 2005) or about what kind of person they
are (Bushman et al., 2005). In the absence of an initial provocation, this kind of rumination
manipulation appears to lack external validity. Researchers argue that rumination generally
occurs following some negative mood, blocked goal, or perceived threat (Carver & Scheier,
1981; Martin & Tesser, 1996; Pyszczynski & Greenberg, 1987; Watkins & Mason, 2002).
Thus, the meaning of induced rumination without some initial negative event is unclear. I
suggest that the results of this study more closely approximates what happens in real life:
individuals who engage in rumination for any reason are more likely to respond to a minor,
ambiguous provocation with anger, hostility, and aggression.
These two studies also challenge Berkowitz’s (1990) argument that both angry and
sad moods can trigger an aggressive association network, leading to increased anger,
hostility, and aggression. Both current studies found that only self-reported angry mood
correlated with increased aggression. Although Study 1 suggested that the accessibility of

102
arousal-related constructs mediated this link, these results were not replicated in Study 2.
Thus, no firm conclusions should be made about the mechanisms by which angry mood
increases aggression. Nevertheless, the results suggest that individuals in a sad mood are
not likely to respond to provocations with anger and aggression.
Finally, this paper proposed to compare two competing hypotheses about the impact
of rumination arising from angry and sad moods on aggression. Emotion research implies
that only rumination arising from anger will lead to increased aggression, due in part to the
arousing nature of anger. An opposing hypothesis states that rumination in response to any
negative mood will produce more anger in response to a perceived provocation, resulting in
increased aggression. Which hypothesis do the current data support? In fact, neither of
these hypotheses is supported by the current results. In neither study did rumination predict
aggression only for participants receiving the angry mood manipulation. Although Study 1
did suggest that anger-related rumination led to greater accessibility of action words, these
results are not robust and should be interpreted with caution. Thus, there is no evidence
from these experiments that angry mood produces a specific form of rumination that leads
to aggression. However, the data also do not suggest that rumination resulting specifically
from sad mood produces increased aggression. On the contrary, Study 2 found that
rumination did not relate to any aggression measures for participants who received the sad
mood condition. Oddly, many of the links between rumination and aggression appeared in
the neutral mood condition. Moreover, rumination did not consistently correlate with sad
and angry moods in the neutral condition.

103
Thus, the most logical interpretation for the current results may be that repetitive
thoughts, evoked for whatever reason, prime individuals to respond to perceived
provocations with increased aggression. Contrary to rumination theory (Carver & Scheier,
1981; Ingram, 1990; Martin & Tesser, 1996), perhaps ruminative thoughts are not always
precipitated by a negative mood or event. The correlations between state and trait
rumination in these studies suggest that perhaps individuals with tendencies to ruminate
exhibit greater amounts of state rumination, even in non-negative moods. Thus, rumination
may describe more of a mental state than any particular thought content. Perhaps
individuals engaged in a repetitive and rigid thought style respond to perceived
provocations by fixing their thoughts on the event, thus more readily triggering angry
feeling and aggressive behaviors. Future researchers should further examine the
relationship between trait and state rumination. Moreover, researchers could develop
rumination manipulations that induce a repetitive thought style, rather than negative
thoughts (Bushman, 2002; Bushman et al., 2005).
One potential limitation of these interpretations involves the nature of the mood
manipulations used in these studies. Specifically, the mood manipulations could have been
confounded with rumination. The procedure asked participants to read two mood-related
vignettes about other people and then to write for 10 minutes about a personal memory in
which they felt the same emotion. Results from both studies showed that the manipulation
increased both negative moods and state rumination. Although the goal of this procedure
was to induce mood and measure resulting rumination, the writing part of the procedure
could be seen as a rumination manipulation. The most common rumination manipulation in

104
fact asks participants to read and think (or write) about a series of self-referential
statements for several minutes (Lyubomirksy & Nolen-Hoeksema, 1993; Morrow & Nolen-
Hoeksema, 1990; Nolen-Hoeksema & Morrow, 1993). The statements are emotion-
focused, symptom-focused, and self-focused but do not specify any particular emotions
(e.g., “how active/passive you feel,” “the kind of person you are,” “the physical sensations
in your body,” and “why you react the way you do”). Although the current manipulations
did not instruct participants to focus on themselves or any other topic, the mere fact of
writing for 10 minutes may mimic the standard rumination manipulation of thinking about
the same topic for several minutes.
An alternative view can be taken, however. It is in fact possible that most existing
rumination manipulations are merely mood inductions. No researchers have directly
measured whether participants are engaged in a repetitive thought process after the
manipulations. If researchers do include rumination checks, they generally assess how
much participants are thinking about themselves or something that happened (e.g.,
Pedersen et al., 2005). It is clear that thinking about negative events or personal
characteristics exacerbates negative mood (e.g., Morrow & Nolen-Hoeksema, 1990;
Rusting & Nolen-Hoeksema, 1998). However, it is not clear that these manipulations in
fact induce a rigid, repetitive thought process. Rumination researchers must develop less
confounded methods of inducing ruminative thinking.
Researchers in this field should also address the difference between rumination and
cognitive or emotional processing. Whereas rumination is generally seen as an unhealthy
coping strategy, cognitive and emotional processing can be adaptive forms of coping (see

105
Segerstrom et al., 2003). Researchers have shown that expressive writing about traumatic
events or losses can benefit some people’s health (Pennebaker, Zech, & Rime, 2001;
Stanton, Kirk, et al., 2000; Stroebe, Schut, & Stroebe, 2006). Stroebe and colleagues
(2006) argued that writing aimed at improving a negative sense of self, reconstructing an
event, or confronting avoided memories would especially benefit individuals who have
poor coping strategies. By contrast, they argue that passively focusing on negative
representations of self is detrimental to health and adjustment. Researchers have just
started exploring the difference between rumination and more healthy forms of repetitive
thought (e.g., Segerstrom et al., 2003). These different forms of repetitive thought likely
differ along certain content dimensions (Borders & Earleywine, 2006; Segerstrom et al.,
2003). Unfortunately, previous rumination manipulations, including the manipulation used
in these experiments, do not adequately specify or code the content of participants’
ruminations. Thus, participants could actually be engaging in more productive forms of
repetitive thinking. Future studies will have to address this important issue.
In addition to suggesting the need for methodological improvements, the current
results suggest important clinical implications and directions for future research.
Specifically, researchers might focus on developing cognitive strategies for counteracting
and preventing situational rumination. The most promising advances in the treatment of
rumination involve attention training (Wells, 1990), which focuses on developing flexible
control over processing in an attempt to disrupt a rigid, repetitive thought process.
Patients are taught to attend flexibly to a range of different external stimuli. This
treatment has effectively treated anxiety and depression (for a review, see Wells &

106
Papageorgiou, 2004). Researchers could adapt attention training to help individuals who
ruminate to respond less aggressively to perceived provocations. Even more importantly,
however, attention (Wells, 1990) or mindfulness (Segal, Williams, & Teasdale, 2002)
training might help individuals respond to negative emotions with less automatic, rigid
thinking patterns. Given the links between situational rumination and aggressive
responding, preventing rumination from occurring following a negative mood might offer
the most benefits to individuals and society as a whole. In earlier research, we found that
individuals who fear anger are more likely to engage in rumination (Borders &
Earleywine, unpublished data). Thus, researchers may want to focus on individuals who
fear negative emotions (see also Roemer et al., 2005; Taylor, 1995; Taylor & Rachman,
1991; Wegner & Zanakos, 1994) and find techniques for preventing these individuals
from responding to negative emotions with increased rumination.
In conclusion, these studies showed that manipulated mood does lead to increases
in rumination, which in turn relates to increased angry and sad moods. Moreover, this
naturally-occurring rumination resulted in increased aggressive responding to a mild
trigger. Although the mechanisms by which rumination leads to aggression could not be
established from the current results, there was evidence that rumination paired with a
minor provocation activates a whole network of angry and aggressive responses. These
results highlight the importance of finding ways to prevent and counteract the effects of
situational rumination.

107
References
Aiken, L. S., & West, S. G. (1991). Multiple Regression: Testing and interpreting
interactions. Newbury Park, CA: Sage.

Anderson, C. A. (1997). Effects of violent movies and trait hostility on hostile feelings
and aggressive thoughts. Aggressive Behavior, 23, 161-178.

Averill, J. R. (1983). Studies on anger and aggression: Implications for theories of
emotion. American Psychologist, 38, 1145-1160.

Ayduk, O., Mischel, W., & Downey, G. (2002). Attentional mechanisms linking
rejection to hostile reactivity: The role of "hot" versus "cool" focus. Psychological
Science, 13, 443-448.

Barber, L., Maltby, J., & Macaskill, A. (2005). Angry memories and thoughts of
revenge: The relationship between forgiveness and anger rumination. Personality and
Individual Differences, 39, 253–262.

Baron, R. M., & Kenny D. A. (1986). The moderator-mediator variable distinction in
social psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology, 51, 1173-1182.

Baumeister, R., Stillwell, A., & Wotman, S. R. (1990). Victim and perpetrator accounts
of interpersonal conflict: Autobiographical narratives about anger. Journal of
Personality and Social Psychology, 59, 994-1005.

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

Bettencourt, B. A., & Miller, N. (1996). Gender differences in aggression as a function of
provocation: A metaanalysis. Psychological Bulletin, 119, 422–447.

Bodenhausen, G. V., Sheppard, L. A., & Kramer, G. P. (1994). Negative affect and
social judgment: The differential impact of anger and sadness. European Journal of
Social Psychology, 24, 45-62.

Bollen, K. A. (1987). Total direct and indirect effects in structural equation models. In C.
C. Clogg (Ed.), Sociological methodology (pp. 37–69). Washington, DC: American
Sociological Association.

Bollen, K. A., & Stine, R. (1990). Direct and indirect effects: Classical and bootstrap
estimates of variability. Sociological Methodology, 20, 115–140.

108
Borders, A., & Earleywine, M. (2004, November). Rumination and Acceptance Mediate
the Link Between Mindfulness and Aggression. Poster session presented at the
Association for Advancement of Behavior Therapy, New Orleans, LA.

Borders, A., & Earleywine, M. (2006). A conceptual review of rumination: Causes,
process, content, and outcomes. Manuscript in preparation.

Borders, A., Smucker Barnwell, S., & Earleywine, M. (In Press). Alcohol-Aggression
Expectancies and Dispositional Rumination Moderate the Effect of Alcohol
Consumption on Alcohol-related Aggression and Hostility. Aggressive Behavior.

Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129-136.

Bushman, B. J. (2002). Does venting anger feed or extinguish the flame? Catharsis,
rumination, distraction, anger and aggressive responding. Personality and Social
Psychology Bulletin, 28, 724-731.

Bushman, B. J. & Baumeister, R. F. (1998). Threatened egotism, narcissism, self-esteem,
and direct and displaced aggression: Does self-love or self-hate dead to violence?
Journal of Personality and Social Psychology, 75, 219-229.

Bushman, B.J., Pederson, W.C., Vasquez, E.A., Bonacci, A M., & Miller, N. (2005).
Chewing on it can chew you up: Effects of rumination on triggered displaced
aggression. Journal of Personality and Social Psychology, 38, 969-983.

Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of Personality
and Social Psychology, 63, 452-459.

Butler, L. D., & Nolen-Hoeksema, S. (1994). Gender differences in responses to
depressed mood in a college sample. Sex Roles, 30, 331-346.

Carlson, M., & Miller, N. J. (1987). Explanation of the relation between negative mood
and helping. Psychological Bulletin, 102, 91-108.

Caprara, G. V. (1986). Indicators of aggression: The dissipation-rumination scale.
Personality & Individual Differences, 7, 763-769.

Caprara, G. V. (1985). Effect of insult and dissipation-rumination on delayed aggression
and hostility. Archivio di Psicologia, Neurologia e Psichiatria, 46, 130-139.

Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A control theory
approach to human behavior. New York: Springer-Verlag.

109
Channon, S. (1996). Executive dysfunction in depression: The Wisconsin Card Sorting
Test. Journal of Affective Disorders, 39, 107-114.

Collins, K., & Bell, R. (1997). Personality and aggression: The dissipation-rumination
scale. Personality and Individual Differences, 22, 751-755.

Davis, R. N., & Nolen-Hoeksema, S. (2000). Cognitive inflexibility among ruminators
and non-ruminators. Cognitive Therapy and Research, 24, 669-711.

Dickson, J. M., & MacLeod, A. K. (2004). Approach and Avoidance Goals and Plans:
Their Relationship to Anxiety and Depression. Cognitive Therapy and Research, 28,
415-432.

Dodge, K. A., Price, J. M., Bachorowski, J., & Newman, J. P. (1990). Hostile
Attributional Biases in Severely Aggressive Adolescents. Journal of Abnormal
Psychology, 99, 385-392.

Dollard, J. (1938). Hostility and fear in social life. Social Forces, 17, 15–25.

Duncan, B. L. (1976). Differential social perception and attribution of intergroup
violence: Testing the lower limit of stereotyping blacks. Journal of Personality and
Social Psychology, 34, 590–598.

Efron, B., & Tibshirani, R. (1993). An introduction to the bootstrap. New York:
Chapman & Hall/CRC.

Finman, R. & Berkowitz, L. (1989). Some factors influencing the effect of depressed
mood on anger and overt hostility toward another. Journal of Research in
Personality, 23, 70-84.

Green, J. D., & Sedikides, C. (1999). Affect and self-focused attention revisited: The
role of affect orientation. Personality and Social Psychology Bulletin, 25, 104-119.

Greenwald, A.G., & Farnham, S.D. (2000). Using the Implicit Association Test to
measure self-esteem and self-concept. Journal of Personality and Social Psychology,
79, 1022–1038.

Harmon-Jones, E., & Sigelman, J. (2001). State anger and prefrontal brain activity:
evidence that insult-related relative left-prefrontal activation is associated with
experienced anger and aggression. Journal of Personality and Social Psychology, 80,
797-803.

110
Harver, A., Katkin, E. S., & Bloch, E. (1993). Signal-detection outcomes on heartbeat
and respiratory resistance detection tasks in male and female subjects.
Psychophysiology, 30, 223–230.

Higgins, E., & King, G. (1981). Accessibility of social constructs: Information processing
consequences of individual and contextual variability. In N. Cantor & J. Kihlstrom
(Eds.), Personality, cognition, and social interaction (pp. 69–122). Hillsdale, NJ:
Lawrence Erlbaum Associates, Inc.

Ingram, R. E. (1990). Self-focused attention in clinical disorders: Review and a
conceptual model. Psychological Bulletin, 107, 156-176.

Izard, C. E. (1991). The Psychology of Emotions. New York : Plenum Press.

Kenworthy, J. B., Canales, C. J., Weaver, K. D., & Miller, N. (2003). Negative
incidental affect and mood congruency in crossed categorization. Journal of
Experimental Social Psychology, 39, 195-219.

Labouvie-Vief, G., Lumley, M. A., Jain, E., & Heinze, H. (2003). Age and gender
differences in cardiac reactivity and subjective emotion responses to emotional
autobiographical memories. Emotion, 3, 115-126.

Larsen, R. J., & Diener, E. (1992). Promises and problems with the circumplex model of
emotion. Review of Personality and Social Psychology, 13, 25-59.

Larsen, R. J.,& Ketelaar, T. (1989). Extraversion, neuroticism, and susceptibility to
positive and negative mood induction procedures. Personality and Individual
Differences, 10, 1221-1228.

Levenson, R. W., Carstensen, L. L., & Gottman, J. M. (1994). The influence of age and
gender on affect, physiology, and their interrelations: A study of long-term marriages.
Journal of Personality and Social Psychology, 67, 56–68.

Lockwood, C. M., & MacKinnon, D. P. (1998, March). Bootstrapping the standard error
of the mediated effect. Paper presented at the 23rd Annual Meeting of SAS Users
Group International, Cary, NC.

Lyubomirsky, S., & Nolen-Hoeksema, S. (1993). Self-perpetuating properties of
dysphoric rumination. Journal of Personality and Social Psychology, 65, 339–349.

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

111
Lyubomirsky, S., Tucker, K. L., Caldwell, N. D., & Berg, K. (1999). Why ruminators are
poor problem-solvers: Clues from the phenomenology of dysphoric rumination.
Journal of Personality and Social Psychology, 77, 1041–1060.

MacBrayer, E. K., Milich, R., & Hundley, M. (2003). Attributional biases in aggressive
children and their mothers. Journal of Abnormal Psychology, 112, 698-708.

MacKinnon, D. P., Lockwood, C.M., Hoffman, J.M., West, S.G., & Sheets, V. (2002). A
comparison of methods to test mediation and other intervening variable effects.
Psychological Methods, 7, 83-104.

Martin, L. L., & Tesser, A. (1996). Some ruminative thoughts. In, R. S. Wyer, Jr. (Ed.),
Ruminative Thoughts, Advances in Social Cognition, Volume IX (pp. 1-48). New
Jersey: Lawrence Erlbaum Associates.

McCullough, M. E., Bellah, C.G., Kilpatrick, S. D., & Johnson, J. L. (2001).
Vengefulness: Relationships with forgiveness, rumination, well-being, and the big
five. Personality and Social Psychology Bulletin, 27, 601-610.

McCullough, M. E., Rachal, K. C., Sandage, S. J., Worthington, E. L., Brown, S. W., &
Height, T. L. (1998). Interpersonal forgiving in close relationships: II. Theoretical
elaboration and measurement. Journal of Personality and Social Psychology, 75,
1586–1603.

Metcalfe, J., & Mischel, W. (1999). A hot/cool-system analysis of delay of gratification:
Dynamics of willpower. Psychological Review, 106, 3-19.

Mikulincer, M.(1988). Reactance and helplessness following exposure to unsolvable
problems: The effects of attributional style. Journal of Personality and Social
Psychology, 54, 679-686.

Miller, N., Pederson, W. C., Earleywine, M., & Pollock, V. E. (2003). A theoretical
model of triggered displaced aggression. Personality and Social Psychology Review,
7, 75-97.

Mor, N., & Winquest, J. (2002). Self-focused attention and negative affect: A meta-
analysis. Psychological Bulletin, 128, 638-662.

Morren, M., & Meesters, C. (2002). Validation of the Dutch version of the aggression
questionnaire in adolescent male offenders. Aggressive Behavior, 28, 87-96.

Morrow, J., & Nolen-Hoeksema, S. (1990). Effects if responses to depression on the
remediation of depressive affect. Journal of Personality and Social Psychology,58,
519-527.

112

Newman, J. P., Patterson, C. M., & Kosson, D. S. (1987). Response perseveration in
psychopaths. Journal of Abnormal Psychology, 96, 145-148.

Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration
of depressive episodes. Journal of Abnormal Psychology, 100, 569-582.

Nolen-Hoeksema, S., & Jackson, B. (2001). Mediators of the gender difference in
rumination. Psychology of Women Quarterly, 25, 37-47.

Nolen-Hoeksema, S., McBride, A., & Larson, J. (1997). Rumination and psychological
distress among bereaved partners. Journal of Personality and Social Psychology, 72,
855–862.

Nolen-Hoeksema, S., & Morrow, J. (1991). A prospective study of depression and
distress following a natural disaster: The 1989 Loma Prieta earthquake. Journal of
Personality and Social Psychology, 61, 105-121.

Nolen-Hoeksema, S., & Morrow, J. (1993). Effects of rumination and distraction on
naturally-occurring depressed mood. Cognition and Emotion, 7, 561-570.

Nolen-Hoeksema, S., Morrow, J., & Fredrickson, B. L. (1993). Response styles and the
duration of episodes of depressed mood. Journal of Abnormal Psychology, 102, 20-
28.

Nolen-Hoeksema, S., Parker, L. E., & Larson, J. (1994). Ruminative coping with
depressed mood following loss. Journal of Personality and Social Psychology, 67,
92–104.

O’Connor, D. B., Archer, J., & Wu, F. W. C. (2001). Measuring aggression: Self-reports,
partner reports, and responses to provoking scenarios. Aggressive Behavior, 27, 79-
101.

Orobio de Castro, B., Veerman, J. W., Koops, W., Bosch, J. D., & Monshouwer, H. J.
(2002). Hostile attribution of intent and aggressive behavior: A meta-analysis. Child
Development, 73, 916-934.

Papageorgiou, C., & Wells, A. (1999). Process and meta-cognitive dimensions of
depressive and anxious thoughts and relationships with emotional intensity. Clinical
Psychology and Psychotherapy, 6, 156-162.

Papageorgiou, C., & Wells, A., Eds. (2004). Depressive Rumination: Nature, Theory, and
Treatment. England: Wiley.

113
Park, J. & Banaji, M. R. (2000). Mood and heuristics: The influence of happy and sad
states on sensitivity and bias in stereotyping. Journal of Personality and Social
Psychology, 78, 1005-1023.

Pedersen, W. C., Denson, T. F., Goss, R. J. Vasquez, E. A. Earleywine, M., & Miller, N.
(2005). The impact of rumination on aggressive thoughts, feelings, behavior, and
arousal. Manuscript in preparation.

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

Pennebaker, J. W, & Roberts, T. A. (1992). Toward a his and hers theory of emotion:
Gender differences in visceral perception. Journal of Social and Clinical Psychology,
11, 199-212.

Pennebaker, J. W., Zech, E., & Rime´, B. (2001). Disclosing and sharing emotion:
Psychological, social, and health consequences. In M. Stroebe, R. O. Hansson, W.
Stroebe, & H. Schut (Eds.), Handbook of bereavement research: Consequences,
coping, and care (pp. 517– 543). Washington, DC: American Psychological
Association Press.

Preacher, K. J., & Hayes, A. (2004). SPSS and SAS procedures for estimating indirect
effects in simple mediation models. Behavior Research Methods, Instruments, and
Computers,36, p. 717-731.

Pyszczynski, T., & Greenberg, J. (1987). Self-regulatory perseveration and the depressive
self-focusing style: A self-awareness theory of reactive depression. Psychological
Bulletin, 102, 122–138.

Pyszczynski, T., Holt, K., & Greenberg, J. (1987). Depression, self-focused attention, and
expectancies for positive and negative future life events for self and others. Journal of
Personality & Social Psychology, 52, 994-1001.

Roemer, L., Salters, K., Raffa, S. D., & Orsillo, S. M. (2005). Fear and avoidance of
internal experiences in GAD: Preliminary tests of a conceptual model. Cognitive
Therapy and Research, 29, 71–88.

Rusting, C. L. (1999). Interactive effects of personality and mood on emotion-congruent
memory and judgment. Journal of Personality and Social Psychology, 77, 1073-
1086.

114
Rusting, C. L., & Nolen-Hoeksema, S. (1998). Regulating responses to anger: Effects of
rumination and distraction on angry mood. Journal of Personality and Social
Psychology, 74, 790-803.

Salovey, P. (1992). Mood-induced self-focused attention. Journal of Personality and
Social Psychology, 62, 699-707.

Segal, Z. V., Williams, J., M., & Teasdale, J. D. (2002). Mindfulness Based Cognitive
Therapy for Depression: A New Approach for Preventing Relapse. New York:
Guilford Press.

Segerstrom, S. C., Shortridge, B. E., Stanton, A. L., & Alden, L. E. (2003). A
multidimensional structure for repetitive thought: What’s on your mind, and how, and
how much? Journal of Personality and Social Psychology, 85, 909-921.

Segerstrom, S. C., Tsao, J. C. I., Alden, L. E., & Craske, M. G. (2000). Worry and
rumination: Repetitive thought as a concomitant and predictor of negative mood.
Cognitive Therapy and Research, 24, 671–688.

Shapiro, S. K., Quay, H. C., Hogan, A., E., & Schwartz. K. P. (1988). Response
perseveration and delayed responding in undersocialized aggressive conduct disorder.
Journal of Abnormal Psychology, 97, 371-373.

Silvia, P. J. Eichstaedt, J., & Phillips, A. G. (2005). Are rumination and reflection types
of self-focused attention? Personality and Individual Differences, 38, 871-881.

Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental
studies: New procedures and recommendations. Psychological Methods, 7, 422-445.

Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations
models. In S. Leinhart (Ed.). Sociological methodology 1982 (pp. 290-312). San
Francisco: Jossey-Bass.

Stanton, A. L., Kirk, S. B., Cameron, C. L., & Danoff-Burg, S. (2000). Coping through
emotional approach: Scale construction and validation. Journal of Personality and
Social Psychology, 78, 1150–1169.

Stroebe, M., Schut, H., & Stroebe, W. (2006). Who benefits from disclosure?
Exploration of attachment style differences in the effects of expressing emotions.
Clinical Psychology Review, 26, 66– 85.

Sukhodolsky, D. G.; Golub, A., & Cromwell, E. N. (2001). Development and validation
of the anger rumination scale. Personality & Individual Differences, 31, 689-700.

115
Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4
th
Ed.).
Boston: Allyn and Bacon.

Taylor, S. (1995). Anxiety sensitivity: Theoretical perspectives and recent findings.
Behaviour Research and Therapy, 33, 243–258.

Taylor, S. & Rachman, S. J.(1991). Fear of sadness. Journal of Anxiety Disorders, 5,
375-381.

Tiedens, L. Z. (2001). The effect of anger on the hostile inferences of aggressive and
nonaggressive people: Specific emotions, cognitive processing, and chronic
accessibility. Motivation & Emotion, 25, 233-251.

Tiedens, L. Z., & Linton, S. (2001). Judgment under emotional certainty and uncertainty:
The effects of specific emotions on information processing. Journal of Personality
and Social Psychology, 81, 973-988.

Thomsen, D. K., Mehlsen, M. Y., Christensen, S., & Zachariae, R. (2003). Rumination—
relationship with negative mood and sleep quality. Personality and Individual
Differences, 34, 1293–1301.

Trapnell, P. D., & Campbell, J. D. (1999). Private self-consciousness and the five-factor
model of personality: Distinguishing rumination from reflection. Journal of
Personality and Social Psychology, 76, 284-304.

VanOostrum, N., & Horvath, P. (1997). The effects of hostile attribution on adolescents'
aggressive responses to social situations. Canadian Journal of School Psychology, 13,
48-59.

Watkins, E., & Mason, A. (2002). Mood as input and rumination. Personality and
Individual Differences, 32, 577–587.

Wegner, D. M. & Giuliano, T. (1980). Arousal-induced attention to self. Journal of
Personality and Social Psychology, 38, 719-726.

Wegner, D. M. & Giuliano, T. (1983). On sending artifact in search of artifact: Reply to
McDonald, Harris, and Maher. Journal of Personality and Social Psychology, 44,
290-293.

Wegner, D. M., & Zanakos, S. (1994). Chronic thought suppression. Journal of
Personality, 62, 615–640.

Wells, A. (1990). Panic disorder in association with relaxation induced anxiety: An
attentional training approach to treatment. Behavior Therapy, 21, 273-280.

116

Wells, A., & Papageorgiou, C. (2004). Metacognitive therapy for depressive rumination.
In C. Papageorgiou & A. Wells (Eds). Depressive Rumination: Nature, Theory, and
Treatment. England: Wiley.

Wood, J. V., Saltzberg, J. A., & Goldsamt, L. A. (1990). Does affect induce self-focused
attention? Journal of Personality and Social Psychology, 58, 899-908.

Abstract (if available)

Abstract Experimental studies show that manipulated rumination increases aggressive responding to perceived provocations. This study expands on previous work by instead manipulating negative mood and measuring naturally-occurring rumination and subsequent aggression. A pair of studies found that both angry and sad moods increased self-reported rumination, compared to neutral mood. Moreover, angry mood generally led to increased aggression. Study 1 failed to show an association between rumination, hostile attributions, and aggressive intentions. However, Study 2 found robust correlational links between increased rumination and aggressive responding to a mild trigger. Furthermore, although rumination led to increased angry and sad moods following a mild trigger, only angry mood mediated the association between rumination and increased aggression. Interpretations and implications of these results are discussed.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

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

Implicit theory ideology about human attributes and aggression

PDF

The moderating effect of triggering provocation on direct aggression

PDF

Frustration within the triggered displaced aggression paradigm

PDF

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

PDF

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

PDF

Family aggression exposure and community violence exposure associated with brain volume in late adolescence: a comparison of automated versus manual segmentation

PDF

An investigation of the factors associated with electronic aggression among college students

PDF

Reducing gang involvement through employment: a pilot intervention

PDF

The link between the expression of negative emotion and future well-being: the role of disclosure modality in an East Asian American population

PDF

The role of framing in health social comparisons by multiple sclerosis patients

PDF

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

PDF

Perceived discrimination and Latino youth adjustment: examining the influence of relinquished control and immigration status

PDF

Adolescent life stress and the cortisol awakening response: the moderating roles of emotion regulation, attachment, and gender

PDF

Opportunity beliefs and behavioral outcomes in Latinx youth: an exploration of Ogbu’s cultural-ecological model

PDF

Genetic correlates of the timing of natural menopause: the breast and prostate cancer cohort consortium

PDF

The impact of statistical method choice: evaluation of the SANO randomized clinical trial using two non-traditional statistical methods

PDF

Proposition 209: a case study on the impact of race-based legislation on student affairs at the University of California

PDF

Pregnancy in the time of COVID-19: effects on perinatal mental health, birth, and infant development

Asset Metadata

Creator Borders, Cathryn Ashley (author)

Core Title The impact of naturally-occurring rumination on aggression

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 04/16/2007

Defense Date 05/17/2006

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

Tag aggression,mood,OAI-PMH Harvest,rumination

Language English

Advisor Miller, Norman (committee chair), Earleywine, Mitch (committee member), Huey, Stanley J., Jr. (committee member), Margolin, Gayla (committee member)

Creator Email cborders@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m393

Unique identifier UC1411887

Identifier etd-Borders-20070416 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-325783 (legacy record id),usctheses-m393 (legacy record id)

Legacy Identifier etd-Borders-20070416.pdf

Dmrecord 325783

Document Type Dissertation

Rights Borders, Cathryn Ashley

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