Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

AGGRESSIVE BEHAVIOR

Volume 36, pages 28–44 (2010)

Could Mindfulness Decrease Anger, Hostility,


and Aggression by Decreasing Rumination?
Ashley Borders1, Mitch Earleywine2, and Archana Jajodia3
1
The College of New Jersey, Ewing, New Jersey
2
University at Albany, State University of New York, Albany, New York
3
Veterans Affairs San Diego Healthcare System, University of California at San Diego, San Diego, California

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
Research suggests that rumination increases anger and aggression. Mindfulness, or present-focused and intentional awareness,
may counteract rumination. Using structural equation modeling, we examined the relations between mindfulness, rumination, and
aggression. In a pair of studies, we found a pattern of correlations consistent with rumination partially mediating a causal link
between mindfulness and hostility, anger, and verbal aggression. The pattern was not consistent with rumination mediating the
association between mindfulness and physical aggression. Although it is impossible with the current nonexperimental data to test
causal mediation, these correlations support the idea that mindfulness could reduce rumination, which in turn could reduce
aggression. These results suggest that longitudinal work and experimental manipulations mindfulness would be worthwhile
approaches for further study of rumination and aggression. We discuss possible implications of these results. Aggr. Behav.
36:28–44, 2010. r 2009 Wiley-Liss, Inc.

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

Keywords: aggression; anger; hostility; rumination; mindfulness

INTRODUCTION trait rumination, for instance, Collins and Bell [1997]


randomly assigned low ruminators and high rumi-
Recent research documents important associa-
nators to provocation (negative feedback about
tions between rumination, hostility, anger, and
performance) or control conditions. Participants
aggression. Defined as repetitive, uncontrollable
then competed with an ‘‘opponent’’ and could
thoughts about negative internal or external experi-
choose to increase their own points, deduct points
ences [Ingram, 1990; Martin and Tesser, 1996],
from their opponent, or deliver a loud white noise to
rumination involves harping on something negative,
the opponent. High ruminators in the provocation
seemingly without end or control. These repetitive
condition exhibited more aggression (e.g., total
thoughts often focus on current feelings, related
number of noise blasts) than low ruminators in the
causes, consequences, and potential solutions
same condition. Similarly, Caprara [1986] and
[Lyubomirsky and Nolen-Hoeksema, 1995]. For
Caprara et al. [1987] found that high dispositional
instance, individuals might become fixated on
ruminators exhibited more hostility, deliberated
thoughts about why they cannot cope with things,
more over thoughts of retaliation, and delivered
why their partners treated them badly, or how they
stronger shocks following hostile experiences than
will perform in an upcoming speech. We will first
did low ruminators. Hostile rumination also longi-
briefly review the evidence suggesting that rumination
tudinally predicts self-reported aggressive and delin-
exacerbates and leads to anger, hostility, and aggres-
quent behavior in adolescents [Caprara et al., 2007].
sion. We will then turn to a discussion of how
mindfulness may counteract the effects of rumination.
Correspondence to: Ashley Borders, The College of New Jersey,
Social Sciences Building 121, 2000 Pennington Rd., Ewing 08628,
Rumination and Aggression-Related Variables New Jersey. E-mail: borders@tcnj.edu
Nonexperimental and experimental studies alike Received 5 October 2008; Accepted 20 July 2009
have found robust links between rumination, anger, Published online 22 October 2009 in Wiley InterScience (www.
hostility, and aggression. Examining the influence of interscience.wiley.com). DOI: 10.1002/ab.20327

r 2009 Wiley-Liss, Inc.


Mindfulness and Rumination 29

In addition, measures of rumination correlate with an associative network (e.g., a particular negative
increased vengefulness, or the disposition to seek emotion) can activate other items stored in that
revenge after suffering an interpersonal offense network. Miller et al. [2003] suggested that repetitive
[Barber et al., 2005; McCullough et al., 1998, 2001]. thoughts and elaborations about a provocation may
Experimental manipulations of rumination show maintain the activation of anger-related associative
similar links to anger and aggression. Bushman networks over time, making angry mood and hostile
[2002] instructed participants to hit a punching bag thoughts more accessible and subsequent aggressive
as many times as they wanted (venting) and at the behavior more likely. Some evidence now supports
same time to think about a partner who criticized this theory. Angered participants made to ruminate
their work (rumination). By contrast, participants in became angrier and exhibited more aggressive
the distraction condition focused their thoughts on responses following a minor trigger than did
becoming physically fit while they hit the punching participants who distracted themselves [Bushman
bag, and participants in the control condition sat et al., 2005; Rusting and Nolen-Hoeksema, 1998].
quietly for several minutes. Results indicated that Thus, rumination about perceived insults may
rumination while venting increased subsequent self- activate aggressive associative networks, exacerbat-
reported anger and aggressive behavior (again in the ing existing angry mood and hostile cognitions and
form of delivered noise blasts). Rusting and Nolen- resulting in aggressive behavior. Moreover, these
Hoeksema [1998] manipulated rumination by asking results reinforce the idea that anger, hostility, and
angered participants to focus their attention on aggression are unique constructs that can influence
thoughts that were emotion- and self-focused but each other [e.g., Bryant and Smith, 2001] and should
not explicitly about anger (e.g., ‘‘why people treat therefore be examined separately.
you the way they do’’). In the distraction condition, The above discussion strongly implies that coun-
participants focused their attention on nonemo- teracting rumination might decrease aggression,
tional thoughts (e.g., ‘‘the layout of the local post anger, and hostility. Research suggests that the
office’’). Following these manipulations, partici- active components of rumination are out-of-control
pants completed stories that were later rated for repetitive thoughts [Segerstrom et al., 2003] and
anger content. In all experiments, rumination negative judgments about current distress [Rude
increased anger, whereas distraction decreased or et al., 2007]. Moreover, Martin and Tesser [1996]
had no effect on pre-existing anger. theorized that rumination can be either about past
Researchers have also investigated the impact of events or future concerns. Several researchers have
rumination on triggered displaced aggression, or recently theorized that mindfulness may constitute a
instances in which a person who is provoked but potential antidote to this kind of thought process.
cannot retaliate directly against the source of
provocation subsequently becomes aggressive with
Mindfulness and its Impact
a person who merely provides a trivial annoyance
[Pedersen et al., 2000]. In two studies [Bushman Extensive research suggests that mindfulness is
et al., 2005], provoked participants who ruminated associated with a range of positive outcomes, from
were more aggressive after a trivial triggering event improved life satisfaction and positive mood to
than were those in a distraction condition. Rumina- decreased stress [Brown and Ryan, 2003; Shapiro
tion did not increase aggression in the absence of a et al., 2008]. Clinically, mindfulness interventions
trigger. Finally, provocation-induced negative affect have effectively treated depression, anxiety, eating
correlated with aggression only for participants in disorders, substance abuse, and borderline person-
the rumination condition. ality disorder [for reviews, see Baer, 2003; Grossman
In sum, research suggests that rumination exacer- et al., 2004]. Only recently studies have examined the
bates angry mood, increases hostility, interacts with mechanisms of mindfulness and its association with
perceived provocation to elicit more aggressive rumination and aggression. In this section, we will
behaviors, and may increase physiological arousal. first discuss the concept of mindfulness and then
The processes by which rumination causes aggres- differentiate it from rumination and related thought
sion are still being explored. Network models processes. We will also review the brief literature
propose that mood-congruent information is orga- examining associations between mindfulness and
nized in long-term memory around central nodes aggression-related variables.
[Berkowitz, 1990; Bower, 1981]. These associative Born out of traditional Eastern meditation prac-
networks link related memories, thoughts, feelings, tices, mindfulness involves a unique way of paying
and behavioral tendencies. Activation of one part of attention [Baer, 2003; Shapiro et al., 2006]. First, it

Aggr. Behav.
30 Borders et al.

involves bringing one’s full attention to the present in analyzing negative feelings. Because there is so
moment, rather than dwelling on past events or much to observe in any given moment, this
future possibilities. In fact, mindfulness may be seen intentional attention to the present moment
in part as a form of attention regulation, in that [Shapiro et al., 2006] necessitates flexibility and
individuals who are mindful are able to consciously prevents rigid self-absorption. In support of this
redirect their attention away from past negative theory, mindfulness is associated with increased
memories or future worries and back to present cognitive flexibility [Davis and Nolen-Hoeksema,
sensations [Baer, 2003; Bishop et al., 2004]. 2000], whereas rumination is associated with decreased
Mindfulness is also characterized by an attitude of flexibility on neuropsychological tests of executive
nonjudgmental acceptance [Shapiro et al., 2006, functioning [Philippot and Brutoux, 2007]. Thus, the
2008]. People practicing mindfulness can notice and intentional and flexible present-focus of mindfulness
experience current internal and external events in an contrasts sharply with the more rigid dwelling in the
accepting way. Teasdale et al. [1995, 2003] argue past or future that characterizes rumination.
that this acceptance permits people to stop brooding Research provides support for the negative
about their unpleasant thoughts and feelings. Rather associations between mindfulness and rumination.
than labeling pain as ‘‘bad’’ and something to avoid, Brown and Ryan [2003] found that their mind-
mindful people view pain as just one of many fulness measure correlated negatively with rumina-
sensations to experience. Hayes et al. [1999] suggests tion. Coffey and Hartman [2008] found a pattern of
that individuals who are mindful are more willing to correlations suggesting that rumination might med-
experience negative internal and external events. In iate the association between mindfulness and psy-
one study, participants engaged in mindful breath- chological distress (as measured by a latent factor
ing showed a greater willingness to view pictures combining depressive and anxious symptoms).
with negative emotional content [Arch and Craske, Recent experiments have examined the effect of
2006]. Thompson and Waltz [2008] also found links mindfulness meditation by novice meditators on
between mindfulness and experiential avoidance. rumination. Chambers et al. [2008] had 20 partici-
Mindful individuals were less likely to view negative pants engaged in a 10-day intensive mindfulness
feelings and unpleasant events as scary or unaccep- meditation retreat. Compared with a control group
table. This attitude of equanimity toward all kinds that did not attend the retreat, these participants
of emotions and thoughts may decrease people’s reported decreased depressive symptoms and reflec-
perceived need for rumination, as current distress is tive rumination. Similarly, two studies [Ramel et al.,
not judged so negatively. 2004; Shapiro et al., 2008] found that people who
participated in 8-week mindfulness courses showed
decreased rumination. Moreover, a change in
Distinguishing Between Mindfulness,
reported rumination accounted for decreases in
Rumination, and Related Constructs
depressive and anxious symptoms [Ramel et al.,
The intentional ‘‘here and now’’ focus of attention 2004]. In sum, initial evidence suggests that heigh-
contrasts with the seemingly uncontrollable mental tened mindfulness can decrease rumination, which in
circles people experience when they are ruminating. turn results in less depression and anxiety. Thus,
Some researchers have in fact conceptualized decreased rumination appears to be one of the
rumination as maladaptive self-focus [Ingram, mechanisms by which mindfulness leads to greater
1990; Mor and Winquist, 2002], a process char- well-being and less suffering.
acterized by internal, sustained, and rigid attention Theory and initial research suggest that mind-
to perceived discrepancies between current and fulness is a separate construct from distraction,
desired states. This rigid and repetitive self-focus which has traditionally been seen as a form of
elicits negative affect, particularly when a negative avoidant coping [Sexton and Dugas, 2008] and is a
discrepancy exists and when one fails to resolve such common alternative to rumination manipulations
a discrepancy [Carver and Scheier, 1981]. By [Bushman, 2002; Rusting and Nolen-Hoeksema,
contrast, mindfulness is thought to be characterized 1998]. Distraction is a coping skill that involves
by cognitive flexibility, so that it allows people to pushing away unpleasant thoughts and feelings,
disengage from rigid mental cycles [Bishop et al., often by focusing on other thoughts or activities.
2004; Shapiro et al., 2006]. Mindful people may Although moderate forms of distraction (such as
notice negative affect in one moment but will focusing on pleasant events or neutral thoughts)
inevitably notice other sensations or events in may be beneficial [Morrow and Nolen-Hoeksema,
subsequent moments, so they do not get swept up 1990; Nolen-Hoeksema et al., 1993], this strategy

Aggr. Behav.
Mindfulness and Rumination 31

only works for a short time and under certain (‘‘I’m trying to figure myself out’’) but not with
conditions [Hamilton and Ingram, 2001]. Moreover, awareness of internal states (‘‘I’m generally attentive
attempting to push specific thoughts and feelings out to my inner feelings’’), as measured by the Private
of awareness can paradoxically increase the fre- Self-Consciousness scale [Fenigstein et al., 1975]. By
quency of negative cognitions and produce greater contrast, mindfulness is associated with greater
emotional reactions [Wegner and Zanakos, 1994; awareness of internal states but not with increased
Wenzlaff and Wegner, 2000]. Mindfulness, by self-reflection [Brown and Ryan, 2003]. Moreover,
contrast, constitutes an alternative to distraction. mindfulness correlated with increased attention to
Present-centered awareness should in fact engender and clarity of feelings, as well as value-concordant
a deeper experience of one’s thoughts and emotions behavior [Brown and Ryan, 2003]. Thus, although
[Shapiro et al., 2006], rather than avoidance. Baer rumination involves repetitive and rigid thoughts
[2003] suggests that mindfulness might function in about negative thoughts and emotions, it is not
part by exposing individuals to their pain, resulting apparently associated with greater self-awareness.
in subsequent desensitization and decreased negative Mindfulness, by contrast, does appear to foster
reactions to pain. Other researchers suggest that more self-awareness of internal states and personal
mindfulness changes one’s relationship to thoughts values. As such, it should theoretically be associated
rather than changing the content of thought [Hayes with less aggression.
et al., 1999]. Unlike distraction, which temporarily
pushes away unpleasant cognitions or replaces them Mindfulness and Aggression
with other thoughts, mindfulness should foster the
Mindfulness covaries with aggression-related vari-
ability to fully experience negative feelings and
ables. Using their self-report measure of mind-
thoughts without being as emotionally upset by
fulness, Brown and Ryan [2003] found that greater
them. Thus, even though experimental distraction
mindfulness was associated with less impulsiveness
manipulations can decrease negative affect and
and hostility. Heppner et al. [2008] likewise found
aggression [Bushman, 2002; Rusting and Nolen-
that greater mindfulness correlated with less anger,
Hoeksema, 1998], mindfulness may provide a better
hostility, and verbal aggression. In a second study,
alternative. Broderick [2005] provided some initial
they had some participants engaged in a mindful
support for this hypothesis. Comparing 8-minute
eating exercise before receiving a social rejection.
rumination and distraction inductions to 8 minutes
Participants in the mindful condition showed less
of mindfulness meditation, they found that rumina-
subsequent aggression than did participants who
tion exacerbated sad mood, whereas distraction and
only received a rejection but did not engage in the
mindfulness meditation decreased sad mood.
mindfulness exercise. In a pair of multiple baseline
Importantly, however, mindfulness meditation re-
studies, Singh et al. [2007a,b] had three adults with
sulted in less sad mood than did distraction. Mind-
mental illness and three adolescents with conduct
fulness therefore appears to be a different construct
disorder participate in a 4-week course in mind-
than distraction and may have greater potential as an
fulness training. They were then asked to continue
alternative to ruminative thought processes.
their mindfulness practice on their own. Incidents of
Another interesting question is whether mind-
aggressive and delinquent behavior were recorded
fulness or rumination, or both, are associated with
for at least 25 weeks. For all participants, aggression
increased self-awareness. Self-awareness generally
decreased or was eliminated. Although these results
decreases physical and verbal aggression [Kinney
are promising, no data have been collected on the
et al., 2001; Prentice-Dunn and Rogers, 1982;
mechanisms by which mindfulness decreases aggres-
Scheier et al., 1974]. These studies typically induced
sive behavior. The above discussion strongly sug-
self-awareness by having participants look at
gests that rumination may be one such mechanism.
themselves in the mirror, take a self-esteem measure,
or focus primarily on themselves rather than others.
The Current Studies
Carver [1975] suggested that self-awareness makes
people more aware of their internal rules about In sum, although previous research suggests that
aggression, thus preventing the deindividuation that mindfulness decreases rumination and that mind-
might allow for aggressive actions that violate social fulness may decrease aggression, anger, and hosti-
norms [see also Prentice-Dunn and Rogers, 1982]. lity, no studies have examined the relationships
Interestingly, rumination may not be associated among all of these variables. We conducted two
with self-awareness. Trapnell and Campbell [1999] survey studies to examine whether rumination might
found that rumination correlated with self-reflection be one mechanism by which mindfulness is

Aggr. Behav.
32 Borders et al.

associated with less aggression, anger, and hostility. thoughts of retaliation. More recently, Sukhodolsky
Although nonexperimental research cannot prove a et al. [2001] created a measure specifically about
causal mediational model, we expected to find a ‘‘anger rumination,’’ which assesses memories of
pattern of correlations consistent with rumination past anger experiences, attention to immediate anger
mediating the association between mindfulness and experiences, and thoughts about the antecedents and
these aggression-related variables. These results consequences of an anger experience. Unfortu-
would provide an encouraging first step toward nately, these anger-rumination measures confound
establishing a rumination-mediated link between process with outcome. We would expect more
mindfulness and aggression. shared variance between anger-rumination measures
and aggression than between general measures of
rumination and aggression. Therefore, the strongest
evidence for the links between rumination and
STUDY 1
aggression-related variables will come from studies
Our first study examined the associations between that use measures of general rumination. Given the
mindfulness, rumination, and four aggression-related impressive existing literature, we expected rumina-
variables in an undergraduate sample. Specifically, tion to correlate positively with all four aggression
we examined the subscales of the Aggression subscales.
Questionnaire (AQ) [Buss and Perry, 1992] sepa-
rately to yield indices of physical aggression, verbal
Method
aggression, anger, and hostility. Given previous
studies [Brown and Ryan, 2003; Heppner et al., Participants and procedure. We collected
2008], we expected that mindfulness would be data from 464 undergraduate students at a major
strongly associated with hostility, anger, and verbal university in Los Angeles. They completed these
aggression. Although one previous study failed to three measures as part of a larger questionnaire
find significant correlations between mindfulness and given to students enrolled in psychology courses.
physical aggression using the BPAQ [Heppner et al., From information reported on the larger question-
2008], mindfulness-based training has been shown to naire packet, two-thirds of the undergraduate
reduce physical aggression [Singh et al., 2007a,b]. participants were female. Just over 50% of partici-
Thus, we had no clear hypothesis about the link pants were Caucasian-American, 19% were Asian-
between mindfulness and physical aggression. American, 9% were Hispanic-American, 7% were
We also used the Rumination and Reflection Scale African-American, and 13% rated themselves as
[Trapnell and Campbell, 1999], which was designed ‘‘other.’’ Participants’ mean age was 19.7. All
to differentiate between two types of private self- responses were anonymous and informed consent
consciousness [Fenigstein et al., 1975]. Reflection was obtained before completing the questionnaire
taps both self-reflection and awareness of internal packet.
states and is related to openness to experience and Measures
need for cognition [Trapnell and Campbell, 1999]. Mindfulness. We measured trait mindfulness
Rumination taps only self-reflection and is associated using the Mindful Attention Awareness Scale
with neuroticism, anxiety, depression, and negative [MAAS; Brown and Ryan, 2003]. This 15-item
affect. We used this measure because it allowed us to questionnaire inquires about participants’ typical
compare the associations between the two subscales level of present awareness and attention (e.g., ‘‘I
and other study variables. Consistent with prior rush through activities without being really attentive
work [Brown and Ryan, 2003], we expected mind- to them’’ and ‘‘I drive places on ‘automatic pilot’
fulness to correlate with rumination but not with and then wonder why I went there’’). Participants
reflection. Therefore, comparing the rumination and indicate how often they have these experiences on a
reflection subscales allowed us to further establish 6-point Likert scale from ‘‘almost always’’ to
the construct validity of mindfulness. ‘‘almost never.’’ Higher scores reflect more mind-
We also selected this measure because it constitu- fulness. Brown and Ryan [2003] showed that the
tes a general measure of rumination, rather than MAAS correlates with other mindfulness scales, as
symptom-specific rumination. There are several well as with emotional intelligence, openness to
measures of anger-related rumination. Caprara experience, and internal state awareness. The
[1986] created a measure of ‘‘dissipation–rumina- MAAS also reliably distinguished between medita-
tion,’’ or the storing in memory of various experi- tion practitioners and control participants [Brown
ences of provocation, expectations, judgments, and and Ryan, 2003]. One item from the MAAS (‘‘I find

Aggr. Behav.
Mindfulness and Rumination 33

myself preoccupied with the future or the past’’) Morren and Meesters, 2002; O’Connor et al.,
seemed to overlap conceptually with the rumination 2001]. Although some research confirms the original
construct. This item also showed low factor loadings factor structure of this measure [e.g., Harris, 1995],
during measure construction [l 5 .28, Brown and other results suggest that the original 29 items do
Ryan, 2003]. Therefore, this item was left out of our not load sufficiently onto four factors, especially
analyses. The other 14 items were used to create a across different populations [Bryant and Smith,
latent variable of mindfulness (see below). Internal 2001; Vigil-Colet et al., 2005; Williams et al.,
consistency of these 14 items was good, a 5 .86. 1996]. Bryant and Smith [2001] proposed a reduced
Rumination. Participants completed the Rumi- version of the AQ that maintains the four original
nation and Reflection Questionnaire [Trapnell and factors but includes only the three items with the
Campbell, 1999]. This 24-item measure assesses both highest loadings on each factor. They found that
rumination (e.g., ‘‘I always seem to be rehashing in across five diverse samples, this 12-item AQ fit the
my mind recent things I’ve said or done’’) and data better than the original 29-item measure.
reflection (e.g., ‘‘I love exploring my ‘inner self’’). Therefore, we used only three items to create each
Participants indicate their level of agreement with of the four latent variables of physical aggression
each item on a 5-point Likert scale from ‘‘strongly (‘‘Given enough provocation, I may hit another
disagree’’ to ‘‘strongly agree.’’ Higher subscale person,’’ ‘‘There are people who pushed me so far
scores indicate more rumination and reflection. that we came to blows,’’ ‘‘I have threatened people I
Trapnell and Campbell [1999] found that the know’’), verbal aggression (‘‘I often find myself
rumination subscale correlated highly with neuroti- disagreeing with people,’’ ‘‘I can’t help getting into
cism, whereas the reflection subscale correlated with arguments when people disagree with me,’’ ‘‘My
measures of openness to experience and need for friends say that I’m somewhat argumentative’’),
cognition. The rumination and reflection subscales anger (‘‘I flare up quickly but get over it quickly,’’
were minimally but significantly correlated in this ‘‘Sometimes I fly off the handle for no good reason,’’
sample, r 5 .18, Po.05. These results accord with ‘‘I have trouble controlling my temper’’), and
small correlations found by the Trapnell and hostility (‘‘At times I feel I have gotten a raw deal
Campbell [1999], suggesting that these subscales out of life,’’ ‘‘Other people always seem to get the
measure independent tendencies. Moreover, the breaks,’’ ‘‘I wonder why sometimes I feel so bitter
reflection subscale did not correlate significantly about things’’). Internal consistency values for these
with mindfulness or any aggression subscale.1 Thus, items of physical aggression (a 5 .72), verbal aggres-
only the 12 rumination items were used in testing the sion (a 5 .78), hostility (a 5 .75), and anger (a 5 .70)
measurement and mediational models (see below). were adequate.
Internal consistency of this measure was also good
in this sample, a 5 .90.
Plan of Analyses
Aggression, anger, and hostility. Finally, parti-
cipants completed the AQ [Buss and Perry, 1992]. Structural equation modeling (SEM) was used to
This 29-item measure assesses physical aggression, test whether it is plausible that rumination could
verbal aggression, hostility, and anger. Participants mediate the association between mindfulness and
indicate how much each item describes them on a aggression. We used the MPlus program 5.0
5-point Likert scale from ‘‘extremely uncharacter- [Muthén and Muthén, 1998–2007] with the missing
istic of me’’ to ‘‘extremely characteristic of me.’’ at random assumption [Little and Rubin, 1987] and
Higher scores reflect more aggression. This measure maximum likelihood estimation.
reliably differentiates between violent and nonvio- We first tested a measurement model to confirm
lent participants and correlates with responses to the relations of the observed variables with their
provoking scenarios, peer reports of aggression, and latent constructs [Anderson and Gerbing, 1988].
other self-reports of aggression [Harris, 1997; Based on CFAs done during construction of the
MAAS [Brown and Ryan, 2003] and the RRS
1
We initially tested a measurement model that included the reflection [Trapnell and Campbell, 1999] and the high
subscale, as well as rumination, mindfulness, and the four aggression Cronbach’s as obtained in these samples, we tested
variables. Allowing these seven factors to intercorrelate, the model fit single-factor models for the mindfulness and rumi-
the data well, w2 (1154) 5 1845.53, CFI 5 .85, RMSEA 5 .051. The nation items. Based on Bryant and Smith’s [2001]
12 reflection items all had high loadings (ls from .58 to .78).
However, the reflection scale showed small and nonsignificant
findings, we tested a four-factor (physical aggres-
correlations with all factors (r’so.10) other than rumination sion, verbal aggression, hostility, anger) model of
(r 5 .18, Po.05). the refined 12-item AQ. The measurement model

Aggr. Behav.
34 Borders et al.

allowed these six latent variables to intercorrelate significance [for a review, see MacKinnon et al.,
freely. 2002]. One common approach is to create an
To better confirm this measurement model, we estimate of the ‘‘indirect effect,’’ or the effect of
compared the above six-factor model to several the mediator [e.g., Sobel, 1982]. The indirect effect is
alternative models: (a) a five-factor model in which defined as a product of two regression coefficients:
the mindfulness and rumination items loaded onto a (1) the effect of the IV on the mediator and (2) the
single factor, (b) a three-factor model in which all 12 effect of the mediator on the DV after controlling
aggression items loaded onto a single factor, (c) a for the IV. A significance test of mediation roughly
two-factor model in which the mindfulness and determines whether the indirect effect differs sig-
rumination items loaded onto one factor and all nificantly from zero. Traditional significance tests
aggression items loaded onto one factor, and (d) a involve dividing the indirect effect by an estimate of
single-factor model with all measured variables its standard error and comparing the resulting value
loading onto one factor. w2 difference tests com- to the standard normal distribution [see MacKinnon
pared each alternative model to the proposed six- et al., 2002; e.g., Sobel, 1982]. However, analyses
factor model. In these tests, if the difference in w2 indicate that estimates of indirect effects are not
was significant for the number of degrees of freedom necessarily normally distributed [Bollen and Stine,
difference, then the model fit significantly less well 1990; MacKinnon et al., 2002]. Accordingly, these
than the proposed model. traditional tests generally have low power and may
We next used path models to test the plausibility lead to incorrect conclusions.
of mediation from trait mindfulness to the four One promising new method of testing indirect
aggression factors via rumination. First, we fit a effects without making assumptions about normal-
model where the aggression subscales were regressed ity uses bootstrapping methodology [Bollen and
on trait mindfulness (see Fig. 1A). We then Stine, 1990; MacKinnon et al., 2004; Preacher and
introduced the rumination factor as a mediator Hayes, 2004]. A ‘‘bootstrap sample’’ consists of N
(see Fig. 1B). individuals sampled randomly with replacement
We looked for evidence suggesting mediation in a from the original data set, where N is the size of
few ways. Traditionally, a test of mediation requires the original data set. Five hundred bootstrap
three separate regression analyses [Baron and Kenny, samples are created in this way, yielding an
1986]. First, the independent variable (IV; mind- empirical sampling distribution from which 95%
fulness) must predict the dependent variable (DV; confidence intervals can be estimated [Bollen and
aggression). This first analysis establishes that a Stine, 1990; MacKinnon et al., 2004]. In simulations
relationship in fact exists to be explained. Second, the in which the distributions of indirect effects are
IV must predict the proposed mediator (rumination). skewed, these resampling methods have more power
Third, the DV is regressed on both the IV and the than traditional tests that assume normality [Bollen
mediator. The mediator must predict the DV, even and Stine, 1990; MacKinnon et al., 2004; Shrout and
with the IV also entered into the equation. If these Bolger, 2002]. Recent simulation studies suggest that
three tests are all significant, then the coefficient for the bias-corrected bootstrap method produces the
the IV in this third equation is examined. Statistically, most accurate confidence intervals and has greater
mediation occurs when the IV no longer affects the power and more accurate Type I error rates than
DV after inclusion of the mediator in the regression other bootstrap methods [Cheung and Lau, 2008;
equation. Conceptually, this result suggests that the MacKinnon et al., 2004]. Therefore, our analyses
original effect of the IV is in fact explained by the below report both the indirect effects of mediation
mediating variable. In other words, the IV has no as well as bias-corrected bootstrap confidence
direct effect on the DV, after accounting for the effect intervals.
of the mediator. True tests of mediation require
experimental or longitudinal designs in order to
demonstrate that the IV in fact causes the mediator,
Results and Discussion
which in turn causes the DV. Nonexperimental
studies like this one cannot determine the actual Measurement model. A measurement model
temporal association between variables. Rather, was used to confirm the factor structures of trait
statistical tests of mediation with these data suggest mindfulness, rumination, and the four [shortened;
that the proposed mediation model is plausible. Bryant and Smith, 2001] aggression subscales. The
Researchers have designed ways to quantify the model fit the data well, w2(650) 5 1388.60, CFI 5 .89,
effect of mediation and subsequently test for RMSEA 5 .049.

Aggr. Behav.
Mindfulness and Rumination 35

Fig. 1. (A) Path model with mindfulness predicting aggression variables and (B) mediation model with rumination mediating the association between
mindfulness and the aggression variables.

As expected, the model confirmed a single latent the highest factor loadings. These results are in
factor structure for trait mindfulness. The factor accordance with earlier factor analyses of this
loadings for the individual mindfulness items ranged measure [Brown and Ryan, 2003]. We computed a
from .30 to .78. Items like ‘‘I could be experiencing composite reliability coefficient by the method
some emotion and not be conscious of it until some suggested by Werts et al. [1974] using the factor
time later’’ and ‘‘I tend not to notice feelings of loadings of individual items and residual variances
physical tension or discomfort until they really grab [see also Lattin et al., 2003]. The measure showed
my attention’’ had the lowest factor loadings. Items good composite reliability, rc 5 .86.
like ‘‘It seems I am running on automatic, without The model also confirmed a single latent factor
much awareness of what I’m doing’’ and ‘‘I find structure for rumination. The factor loadings for the
myself doing things without paying attention’’ had individual rumination items were high, ranging from

Aggr. Behav.
36 Borders et al.

.51 to .77. The composite reliability for this measure Table III presents intercorrelations between all
in this sample was .90. study variables. Rumination and trait mindfulness
The model confirmed a four-factor model of showed a strong negative correlation. The more
aggression, using only 12 items from the AQ. Factor mindful people were, the less they ruminated.
loadings and composite reliabilities were good for Higher mindfulness was related to less aggression
physical aggression (ls from .59–.73, rc 5 .73), on all aggression subscales. Rumination showed
verbal aggression (ls from .71–.77, rc 5 .78), anger strong positive correlations with the anger and
(ls from .49–.79, rc 5 .71), and hostility (ls from hostility subscales and weaker or nonsignificant
.69–.76, rc 5 .76). associations with physical (r 5 .08, ns) and verbal
To better confirm this measurement model, we aggression (r 5 .12, Po.05). Finally, the aggression
compared the above six-factor model to several subscales showed strong positive intercorrelations,
alternative models. Table I shows model fits for all as expected [Bryant and Smith, 2001; Buss and
of these models. A w2 difference test revealed that Perry, 1992].
the six-factor model provided a significantly better Mediation models. A path model with the
fit than combining rumination and mindfulness into aggression subscales regressed on trait mindfulness
a single factor, supporting the discriminant validity was fitted first. This model fit the data adequately,
of rumination and mindfulness as distinct con- w2(289) 5 610.19, Po.001, RMSEA 5 .049, CFI 5 .91.
structs. Combining the aggression factors into a Trait mindfulness was a significant predictor of all
single factor also created a significantly worse fit, four aggression subscales (see Table IV).
confirming previous work on the discriminant Next, we added rumination as a mediator
validity of these separate facets of aggression. The variable. The model fit was adequate, w2(650) 5
six-factor model also fit significantly better than 1388.59, Po.001, RMSEA 5 .049, CFI 5 .89. Trait
treating a two-factor model of rumination/mind- mindfulness significantly predicted rumination,
fulness and aggression. In addition, a single-factor b 5 .46, SE 5 .04, Po.001. Table IV shows all
model fit significantly less well than the proposed
six-factor model, revealing that the distinctions
among these constructs account for more data than TABLE II. Means, Standard Deviations, and Ranges for All
Study Variables
a single factor of self-report. Thus, theoretically
tenable alternative models fit the data less well than Study 1 Study 2
the proposed six-factor model.
Table II shows variable means, standard devia- Variable Mean SD Range Mean SD Range
tions, and ranges. Participants on average endorsed Mindfulness 4.05 .76 1.2–5.8 4.09 .70 2.2–5.9
a moderate level of trait rumination and trait Rumination 3.55 .71 1.4–5 3.19 .83 1.2–5
mindfulness. Participants in this sample also Physical aggression 1.81 .92 1–5 1.39 .66 1–4.3
reported moderate levels of verbal aggression, Verbal aggression 2.54 .94 1–5 2.42 .83 1–4.7
Hostility 2.54 .98 1–5 2.03 1.03 1–5
hostility, and anger, but they reported lower levels
Anger 2.21 .85 1–5 2.06 .85 1–4.7
of physical aggression.

TABLE I. Alternative Models to the Proposed six-factor Measurement Model

Model tested w2 (df) RMSEA CFI Dw2 (Ddf)

Study 1
Proposed six-factor model: mindfulness factor, rumination factor, four aggression factors 1388.60 (650) .049 .89 –
Five-factor model: one rumination/mindfulness factor, four aggression factors 2553.89 (655) .079 .71 1165.29 (5)
Three-factor model: rumination factor, mindfulness factor, one aggression factor 1973.57 (662) .065 .80 584.97 (12)
Two-factor model: one rumination/mindfulness factor, one aggression factor 3131.88 (664) .089 .62 1743.28 (14)
One-factor model: single rumination/mindfulness/aggression factor 3803.67 (665) .101 .52 2415.07 (15)
Study 2
Proposed six-factor model: mindfulness factor, rumination factor, four aggression factors 1069.17 (650) .059 .87 –
Five-factor model: one rumination/mindfulness factor, four aggression factors 1573.89 (655) .087 .71 504.72 (5)
Three-factor model: rumination factor, mindfulness factor, one aggression factor 1317.37 (662) .073 .79 248.20 (12)
Two-factor model: one rumination/mindfulness factor, one aggression factor 1819.33 (664) .097 .64 750.16 (14)
One-factor model: single rumination/mindfulness/aggression factor 2014.74 (665) .105 .58 945.57 (15)
Po.05; Po.01.

Aggr. Behav.
Mindfulness and Rumination 37

TABLE III. Zero-Order Correlations for All Study Variables

Variable 1 2 3 4 5 6

1. Mindfulness – .48 .24 .25 .43 .23


2. Rumination .46 – .22 .36 .63 .43
3. Physical aggression .14 .08 – .36 .40 .48
4. Verbal aggression .22 .12 .46 – .40 .59
5. Hostility .38 .56 .35 .30 – .50
6. Anger .37 .28 .51 .62 .52 –

Values below the diagonal are from Study 1. Values above the diagonal are from Study 2.
Po.05; Po.01; Po.001.

TABLE IV. Betas, Standard Errors, Indirect Path Estimates, and Bias-Corrected Bootstrap CIs for Mediational Models
Model tested PA b (SE) VA b (SE) Hostility b (SE) Anger b (SE)

Study 1
Mindfulness to aggression (SE) .14 (.06) .22 (.05) .38 (.05) .37 (.05)
Rumination to aggression (SE) .02 (.06) .02 (.06) .48 (.05) .14 (.06)
Mindfulness to aggression, via rumination (SE) .13 (.06) .21 (.06) .16 (.06) .31 (.06)
Indirect paths (SE) .01 (.03) .01 (.03) .22 (.03)a .06 (.03)a
Bias-corrected bootstrap 95% confidence intervals ( .18, .16) ( .16, 11) ( .91, .38) ( .22, .01)
Study 2
Mindfulness to aggression (SE) .21 (.09) .24 (.09) .41 (.08) .23 (.08)
Rumination to aggression (SE) .05 (.10) .33 (.10) .54 (.08) .41 (.09)
Mindfulness to aggression, via rumination (SE) .19 (.10) .08 (.10) .16 (.08) .03 (.10)
Indirect paths (SE) .02 (.05) .15 (.05)a .25 (.05)a .19 (.05)a
Bias-corrected bootstrap 95% confidence intervals ( .23, .11) ( .51, .08) ( 1.13, .57) ( .67, .11)
a
Zero is not included in the 95% confidence interval, indicating that the indirect path is significantly different from zero.
Po.05; Po.01; Po.001.

other betas in the mediational model. Mindfulness mediate the associations between mindfulness and
still significantly predicted all aggression subscales. anger and hostility.
Rumination only significantly predicted hostility In sum, Study 1 found support for many of our
and anger. hypotheses. Specifically, greater mindfulness was
To look for whether the pattern of relations was associated with less rumination. Mindfulness was
consistent with a mediating role for rumination, we also negatively associated with all four aggression
first examined whether the bs between mindfulness subscales. Finally, the pattern of correlations was
and the aggression subscales decreased with the consistent with rumination partially mediating the
inclusion of rumination in the model. The b links between mindfulness and both hostility and
estimates for the links between mindfulness and anger. Thus, it appears that more mindful people
hostility dropped from .38 to .16 with the may be less angry and hostile in part because they
inclusion of rumination, consistent with partial ruminate less. This finding constitutes the first
mediation. Likewise, the bs between mindfulness evidence that rumination may be one potential
and anger dropped from .37 to .31. The bs from mechanism by which mindfulness is associated with
mindfulness to physical and verbal aggression barely anger and hostility.
changed with the inclusion of rumination (for both, The absence of meaningful associations between
bs dropped from .02 to .01), indicating that rumination and physical and verbal aggression was
rumination did not account for these associations. surprising, given previous evidence that rumination
We also calculated indirect effect estimates and bias- produces increased behavioral aggression [Bushman,
corrected bootstrap estimates for the 95% confi- 2002; Bushman et al., 2005; Caprara et al., 1987].
dence intervals (see Table IV). Confidence intervals The absence of an observed association between
that do not include zero indicate a significant rumination and aggression could have resulted
indirect effect. Only the indirect effects for hostility from low levels of particularly physical aggression
and anger were meaningfully different from zero. reported in this undergraduate sample. The presence
These results suggest that our proposed mediation of a floor effect would make any existing relation-
model is plausible and that rumination may partially ships difficult to find. Replication with a different

Aggr. Behav.
38 Borders et al.

population would help clarify the association regular meditation practices. Of the participants
between rumination and behavioral aggression. who meditated regularly, 7% meditated a few times
Moreover, because we did not have access to a month, 45% meditated a few times a week to once
demographic data with this sample, we were unable a week, and 48% meditated once a day or more. We
to determine whether the associations between included the same measures of mindfulness, rumina-
mindfulness, rumination, and aggression reflected tion, and aggression as in Study 1.
any effects of age or gender. We addressed these
limitations in our second study.
Results and Discussion
Measurement models. Measurement models
STUDY 2
were again used to confirm the factor structure of
Study 2 was a replication study using a nonunder- the mindfulness, rumination, and aggression vari-
graduate population. The vast majority of research ables.2 The measurement model with trait mind-
on links between rumination and aggression has fulness, rumination, and aggression fit adequately,
employed undergraduate samples. Finding similar w2(650) 5 1069.17, CFI 5 .87, RMSEA 5 .059. The
results in a different population will help generalize factor loadings for the individual trait mindfulness
these important associations to people of different items again showed a wide range (from .25 to .82)
ages and life experiences. Moreover, we wanted to and a similar pattern as in Study 1. The factor
find a sample with increased exposure to and loadings for the rumination items ranged from .62 to
knowledge about mindfulness. Grossman [2008] .86. Aggression items loaded onto their subscales
suggests that because of their general inexperience similarly to Study 1 (ls from .43 to .90). The
with mindfulness and related concepts, undergrad- composite reliabilities for trait mindfulness
uates’ understanding of and subsequent reporting of (rc 5 .87) and rumination (rc 5 .94) were good.
‘‘mindfulness’’ likely differs from the construct that The composite reliabilities for physical aggression
authors of these measures intended. Therefore, we (.74), verbal aggression (.71), anger (.75), and
targeted adults who were presumably more knowl- hostility (.83) were also adequate.
edgeable about the construct of mindfulness, as We again compared the six-factor model to several
evidenced by their membership in mindfulness- alternative models (see Table I). A w2 difference test
related listservs. The meaning of mindfulness in this revealed that the six-factor model provided a
population should be more in line with the significantly better fit than combining rumination
conceptual basis of mindfulness measures. and mindfulness into a single factor, supporting the
discriminant validity of rumination and mindfulness
Method as distinct constructs. Combining the aggression
factors into a single factor also created a signifi-
Participants and procedure. To find partici-
cantly worse fit, confirming previous work on the
pants, we created an Internet survey and posted the
discriminant validity of these separate facets of
survey link on several mindfulness-related listservs.
aggression. The six-factor model also fit significantly
People were invited to participate through an
better than treating a two-factor model of rumina-
introductory email. Because of this method of
tion/mindfulness and aggression. In addition, a
recruitment, we cannot know what percentage of
single factor model fit significantly less well than
people responded to our request. All participants
the proposed six-factor model, revealing that the
had to read and agree to an informed consent before
distinctions among these constructs account for
participation. Study 2 had 211 participants from 26
more data than a single factor of self-report. Thus,
different states and 6 countries. The participants
theoretically tenable alternative models fit the data
ranged in age from 15 to 72 (mean 5 31.8,
less well than the proposed six-factor model.
SD 5 11.8), and 60% were female. On average,
these participants had graduated from college and 2
We again initially tested a measurement model that included the
taken some graduate courses and earned an average reflection subscale, as well as rumination, mindfulness, and the four
annual income of $40,000–60,000. Over 80% of aggression variables. Allowing these seven factors to intercorrelate,
these participants were Caucasian-American, 3% the model fit the data well, w2(1154) 5 1865.41, CFI 5 .84,
were Hispanic-American, and 5% each were Asian- RMSEA 5 .058. The twelve reflection items all had high loadings
(ls from .57 to .80). However, the reflection scale showed small and
American, African-American, and mixed. In this nonsignificant correlations with all factors (rso.15), including
sample, 57% of participants reported meditating at rumination (r 5 .06, ns). Therefore, the reflection subscale was
least once in their lives, and 22% currently had not included in subsequent models.

Aggr. Behav.
Mindfulness and Rumination 39

Participants again endorsed a moderate level of confidence intervals (see Table IV). These data are
rumination and mindfulness (see Table II). This consistent with the idea that rumination mediates
sample showed a smaller range on the mindfulness the links between mindfulness and verbal aggres-
measure. Specifically, no one in this sample averaged sion, hostility, and anger.
below three on the 6-point scale, whereas under- In sum, this study served as a replication of
graduates in the first sample ranged from 1.2 to 5.8. Study 1 with a nonundergraduate sample. Even after
With this sample, physical aggression, anger and, including age and gender as covariates in the model,
hostility averages were significantly lower than the pattern of correlations was consistent with
values in Study 1 (using one-sample t-tests, rumination again partially mediating the associa-
Pso.05); only verbal aggression was endorsed at a tions between mindfulness and both anger and
similar level as in Study 1. The ranges of the hostility. This time, analyses also suggested that
aggression subscales and rumination measure were rumination may partially mediate the link between
similar across both samples. Rumination showed mindfulness and verbal aggression. Once again, the
strong negative correlations with trait mindfulness pattern was not consistent with rumination mediat-
(see Table III). Both rumination and trait mind- ing the relationship between mindfulness and
fulness showed strong correlations with all four physical aggression. Implications of these results
aggression subscales. will be discussed below.
Mediation models. The path model with the Why rumination correlated with physical and
aggression subscales regressed on trait mindfulness fit verbal aggression in this sample and not in the
the data adequately, w2(288) 5 468.88, Po.001, undergraduate sample is unclear. This particular
RMSEA 5 .060, CFI 5 .89. Specifically, trait mind- sample showed a slightly larger variance on the
fulness was a significant predictor of all four rumination measure, which might explain the larger
aggression subscales. When we added rumination as correlations. Alternatively, perhaps other factors
a potential mediator variable, the model fit was again than rumination best predict tendencies for aggres-
adequate, w2(649) 5 1047.85, Po.001, RMSEA 5 sive behavior in younger, undergraduate popula-
.058, CFI 5 .88. Finally, we added age and gender tions. For example, alcohol consumption and
to the models as covariates. The initial model with emotional vulnerability are associated with in-
mindfulness and aggression fit adequately, w2(330) 5 creased aggressive behavior [Caprara et al., 1987;
538.22, Po.001, RMSEA 5 .055, CFI 5 .87. The Chermack and Giancola, 1997]. These constructs
mediation model also fit the data adequately, may influence college-aged students more than
w2(713) 5 1152.37, Po.001, RMSEA 5 .054, CFI 5 .87. adults who are older and working. Although data
Trait mindfulness significantly predicted rumina- on aggression in children are plentiful, age-related
tion, b 5 .46, SE 5 .06, Po.001. Rumination differences in predictors of aggression have not been
significantly predicted verbal aggression, hostility, identified as frequently in adults [e.g., Hennessy and
and anger, but not physical aggression (see Wiesenthal, 2004]. Future research should investi-
Table IV). Trait mindfulness no longer significantly gate the predictors of behavioral aggression in
predicted any aggression subscale. Gender did adults of different ages.
not show any significant associations with mind-
fulness, rumination, or the aggression subscales
(all bso.13). Older participants reported less rumina-
GENERAL DISCUSSION
tion (b 5 .22, Po.01) and physical aggression
(b 5 .25, Po.01). This pair of studies explored whether rumination
To look for evidence consistent with a mediating might potentially mediate the association between
role for rumination, we first examined whether the mindfulness and several aggression-related vari-
bs between mindfulness and the aggression subscales ables. SEM analyses allowed us to confirm the
decreased with the inclusion of rumination in the factor structures of rumination and mindfulness and
model. The inclusion of rumination in the model to ensure that they are in fact distinct constructs.
caused b estimates to decrease to nonsignificance Moreover, we examined physical aggression, verbal
for the associations between trait mindfulness aggression, anger, and hostility separately, rather
and hostility (b dropped from .41 to .16), anger than as one composite measure of aggression. These
(a dropped from .23 to .03), and verbal studies therefore yield important new information
aggression (b dropped from .24 to .08). As about the potential of mindfulness to decrease both
before, we also calculated indirect effect estimates rumination and the behavioral, emotional, and
and bias-corrected bootstrap estimates for the 95% cognitive components of aggression.

Aggr. Behav.
40 Borders et al.

With two different samples, we found that people [Lopata, 2003; Tyson, 1998]. Thus, mindfulness
who are mindful are less angry and hostile. More- practice may decrease aggression partially through
over, our results are consistent with the proposed making individuals more relaxed.
model that the relationships between mindfulness Another possible mechanism between mindfulness
and anger and hostility are partly mediated by lower and behavioral aggression is emotion regulation, or
levels of ruminative thinking. Only two previous the ability to influence the experience and expression
studies have examined the links between mind- of emotions [Gross, 1998]. Common emotion
fulness, hostility, and anger [Brown and Ryan, 2003; regulation strategies include altering thoughts or
Heppner et al., 2008]. The current studies therefore behavior in order to better cope with negative
provide important initial evidence that mindfulness emotions. Two studies have found that mindfulness
may contribute to decreased anger and hostility. correlates with the ability to regulate negative affect
Moreover, this is the first study we know of to [Brown and Ryan, 2003; Coffey and Hartman,
suggest that mindfulness might reduce anger and 2008]. Moreover, Coffey and Hartman [2008] found
hostility through its effect on rumination. Only one that emotion regulation mediated the association
other study has found that decreased rumination between mindfulness and depressive and anxious
following mindfulness training accounted for de- symptoms. Likewise, research has found links
creased depressive and anxious symptoms [Ramel between aggressive behavior and poor emotion
et al., 2004]. Therefore, our results provide necessary regulation [e.g., Izard et al., 2008; McNulty and
additional evidence that rumination may be one Hellmuth, 2008]. Thus, mindfulness may decrease
mechanism of mindfulness. aggressive behavior in part because it promotes the
Mindfulness was also associated with less physical ability to cope with and even change negative
and verbal aggression in both of our studies. Two internal emotions.
previous case studies found that teaching mind- Mindfulness may also promote better cognitive
fulness decreased aggressive behavior [Singh et al., functioning and flexibility. As discussed above,
2007a,b]. Our studies therefore provide needed mindfulness is associated with increased cognitive
statistical support for the relationship between flexibility [Davis and Nolen-Hoeksema, 2000].
mindfulness and behavioral aggression. Experimen- A recent study also showed that participating in
tal work should be a focus of future studies. The mindfulness training led to increases in working
potential role of rumination in these relationships, memory and sustained attention [Chambers et al.,
however, is less clear. Results were consistent with 2008]. Some researchers in fact suggest that mind-
rumination mediating the association between mind- fulness is a form of ‘‘mental training’’ that involves
fulness and verbal aggression only for our non- self-regulation of attention and cognitive inhibition
undergraduate sample. Moreover, contrary to [Bishop et al., 2004]. On the other hand, aggression is
predictions, the pattern of correlations did not associated with decreased cognitive performance,
suggest that rumination might mediate the relation- especially on tests of planning, attentional control,
ship between mindfulness and physical aggression. and goal-directed behavior [Giancola, 2000]. Thus,
In the undergraduate sample, the lack of support for mindfulness may actually promote better executive
mediation might be due to weak and nonsignificant functioning, which may prevent some of the cogni-
correlations between rumination and overt aggres- tive errors that contribute to aggressive behavior.
sion. However, rumination was associated with both Relatedly, mindfulness may decrease aggression
physical and verbal aggression in the nonunder- because it decreases behaving in automatic or
graduate sample. Despite this, the results did not impulsive ways. Mindfulness should help make
support the proposed model that rumination would people aware of feelings and thoughts when they
mediate the link between mindfulness and physical arise, rather than after they are already acted upon.
aggression. Therefore, it seems that rumination may This may be especially important when emotional
not constitute a major mechanism through which association networks get activated, as in the case of
mindfulness affects verbal and especially physical a perceived provocation. As Ryan and Deci [2000]
aggression. suggest, an open awareness is necessary for choosing
Several other proposed mechanisms for mindfulness actions that coincide with one’s values, needs, and
may influence behavioral aggression. Relaxation is best interest. Initial evidence in fact suggests that
one common outcome of mindfulness practice [for a mindfulness is positively associated with value-
review, see Baer, 2003]. Studies have also found that concordant behavior and negatively associated
relaxation can decrease aggression, presumably be- with impulsivity [Brown and Ryan, 2003]. Future
cause relaxation decreases physiological arousal studies should explore these and other potential

Aggr. Behav.
Mindfulness and Rumination 41

mechanisms of the relationship between mindfulness individuals to engage in high levels of mindfulness
and behavioral aggression. and low levels of rumination. However, substantial
The current studies also suggest important informa- experimental evidence suggests that higher levels of
tion about the construct of mindfulness. Our use of rumination in fact cause increased behavioral
SEM analyses allowed us to confirm the single-factor aggression, compared with lower levels of rumina-
structure of the MAAS. In both samples, the mind- tion [Bushman, 2002; Bushman et al., 2005; Rusting
fulness items assessing awareness of internal states and Nolen-Hoeksema, 1998]. We know of no studies
showed the weakest factor loadings, whereas items showing that manipulated aggression increases sub-
tapping intentional attention had the strongest factor sequent rumination. Likewise, meditation training
loadings. These findings concur with a recent item leads to decreased aggressive behavior [Heppner
response analysis suggesting that the MAAS most et al., 2008; Singh et al., 2007a,b]. Proposed
clearly measures automaticity [Van Dam et al., 2009; mechanisms of mindfulness include intentional
see also MacKillop and Anderson, 2007]. Addition- attention, cognitive and behavioral flexibility, self-
ally, mindfulness shared no variance with reflection, regulation, and values clarification [Shapiro et al.,
suggesting that mindfulness is not related to self- 2006]. It seems much more likely that these mechan-
analysis or philosophical introspection [see also Brown isms result in low aggressive tendencies than the
and Ryan, 2003]. Interestingly, both samples showed other way around. Therefore, both empirical and
the same mean level of mindfulness and similar theoretical work strongly suggest that aggression-
correlations between mindfulness and other study related variables should be the DVs in our model.
variables. These similarities across samples suggest However, only longitudinal or experimental studies
that the MAAS functions similarly across populations can rule out alternative possibilities.
with and without extensive knowledge about mind- In addition, we included only self-report measures,
fulness. Brown and Ryan [2004] in fact suggest that which allowed the possibility of reporter bias. As
mindfulness is an inherent, individual capacity, rather with any study of aggression, participants may
than a product of any behavioral practice like underreport aggressive attitudes or behaviors due
meditation. Future research should investigate the to social norms. Both of our samples reported less
meaning of self-reported mindfulness and the associa- physical aggression than in other published research
tions between mindfulness, spiritual beliefs, and [e.g., Buss and Perry, 1982]. Given that under-
meditation experience in different populations. reporting tendencies would diminish the likelihood
This study possessed some methodological limita- of finding correlations, these data may actually
tions that should be addressed in future research. underestimate true effects. Future studies should
As stated above, statistical tests of mediation therefore include behavioral measures of aggression.
with cross-sectional data cannot illuminate the However, this limitation does not pose a barrier to
temporal associations between study variables. interpretation of our data.
Thus, although we proposed that mindfulness In sum, our studies contribute important new
contributes to decreased rumination, which in turn information about how mindfulness relates to
contributes to decreased anger, hostility, and ag- rumination and aggression-related variables and
gression, our results cannot rule out other possibi- suggest several hypotheses for future research. First,
lities. For instance, high levels of rumination could our results suggest theoretical hypotheses for how
make people less mindful, which might lead to mindfulness might help a person respond to a
increased subsequent anger, hostility, and aggres- provocation or triggering event. Specifically, pre-
sion. It does seem theoretically plausible that people sent-focused and intentional attention may prevent
who abstain from the repetitive and inflexible the destructive cognitive and emotional cycles that
process of rumination have more available attention occur following a perceived provocation. Associa-
for the present moment. However, we know of no tive network theories suggest that when a particular
experimental studies that have examined this possi- provoking trigger is experienced, related emotions,
bility. Moreover, experimental studies suggest that beliefs, and memories are brought to mind
increased mindfulness causes reductions in rumina- [Anderson and Bushman, 2002; Berkowitz, 1990].
tion [Ramel et al., 2004; Shapiro et al., 2008], Some research suggests that anger in particular
indicating the same directionality that we proposed. produces a high amount of arousal and cognitive
Future research could examine the temporal asso- processing [Baumeister et al., 1990]. It is likely that
ciation between mindfulness and rumination. people with a tendency to ruminate will harp on
Alternatively, our results cannot rule out the perceived provocations and possible revenge, thus
possibility that low aggressive tendencies might cause keeping their aggressive association networks active

Aggr. Behav.
42 Borders et al.

and increasing their likelihood of reporting hostile Arch JJ, Craske MG. 2006. Mechanisms of mindfulness: Emotion
thoughts and angry feelings [Berkowitz, 1990; regulation following a focused breathing induction. Behav Res
Rusting and Nolen-Hoeksema, 1998]. Mindfulness Ther 44:1849–1858.
Baer RA. 2003. Mindfulness training as clinical intervention:
may help stop or even prevent this activation by A conceptual and empirical review. Clin Psychol Sci Pract
keeping a person attentive to the present moment 10:125–143.
and preventing rumination about past provocations. Barber L, Maltby J, Macaskill A. 2005. Angry memories and
Therefore, a mindful person may actually experience thoughts of revenge: The relationship between forgiveness and
less anger and hostility following a triggering event, anger rumination. Pers Individ Differences 39:253–262.
Baron RM, Kenny DA. 1986. The moderator-mediator variable
in part because of decreased rumination. Although distinction in social psychological research: Conceptual,
the current studies suggest that rumination does not strategic, and statistical considerations. J Pers Soc Psychol
mediate the link between mindfulness and disposi- 51:1173–1182.
tional physical aggression, it is possible that Baumeister RF, Stillwell A, Wotman SR. 1990. Victim and
rumination might be more related to situationally perpetrator accounts of interpersonal conflict: Autobiographical
narratives about anger. J Pers Soc Psychol 59:994–1005.
induced aggression. Thus, perhaps a mindful person
Berkowitz L. 1990. On the formation and regulation of anger and
would respond to an acute provocation with less aggression: A cognitive-neoassociationistic analysis. Am Psychol
rumination and subsequently less behavioral aggres- 45:494–503.
sion. Alternatively, mindfulness might help a person Bishop SR, Lau M, Shapiro S, Carlson L, Anderson ND, Carmody J,
respond to a provocation with less behavioral Segal ZV, Abbey S, Speca M, Velting D, Devins G. 2004.
aggression through other mechanisms like relaxa- Mindfulness: A proposed operational definition. Clin Psychol Sci
Pract 11:230–241.
tion or self-regulation. Many more experimental Bollen KA, Stine R. 1990. Direct and indirect effects: Classical and
studies are needed to test these specific hypotheses. bootstrap estimates of variability. Sociol Methodol 20:115–140.
Our nonexperimental findings also have important Bower GH. 1981. Mood and memory. Am Psychol 36:129–136.
implications for future anger management treat- Broderick PC. 2005. Mindfulness and coping with dysphoric
ments. Interventions for anger problems typically mood: Contrasts with rumination and distraction. Cogn Ther
Res 29:501–510.
include contingency management, stimulus control, Brown KW, Ryan RM. 2003. The benefits of being present:
or social skills training [Deffenbacher et al., 2002; Mindfulness and its role in psychological well-being. J Pers Soc
Edmondson and Conger, 1996]. Mindfulness train- Psychol 84:822–848.
ing may provide a good complement to these other Brown KW, Ryan RM. 2004. Perils and promise in defining and
important skills. By increasing awareness of current measuring mindfulness: Observations from experience. Clin
Psychol Sci Pract 11:242–248.
emotions, thoughts, and situations, mindfulness
Bryant FB, Smith BD. 2001. Refining the architecture of aggression:
may help promote acceptance, distress tolerance, A measurement model for the Buss–Perry Aggression Ques-
and intentional attention. These emotion regulation tionnaire. J Res Pers 35:138–167.
skills provide a solid foundation for skills training. Bushman BJ. 2002. Does venting anger feed or extinguish the flame?
Recent clinical interventions have incorporated Catharsis, rumination, distraction, anger and aggressive respond-
ing. Pers Soc Psychol Bull 28:724–731.
mindfulness training in treatments for depressive
Bushman BJ, Pederson WC, Vasquez EA, Bonacci AM, Miller N.
relapse, chronic pain, anxiety, borderline personality 2005. Chewing on it can chew you up: Effects of rumination on
disorder, and relapse prevention in substance triggered displaced aggression. J Pers Soc Psychol 38:969–983.
abusers [see Baer, 2003 for a review]. The non- Buss AH, Perry M. 1992. The aggression questionnaire. J Pers Soc
experimental results presented here suggest the Psychol 63:452–459.
hypothesis that mindfulness training may also Caprara GV. 1986. Indicators of aggression: The dissipation-
rumination scale. Pers Individ Differences 7:763–769.
enhance treatments for anger problems. Future Caprara GV, Gargaro BT, Pastorelli C, Prezza M, Renzi P, Zelli A.
research studies should empirically test the benefit 1987. Individual differences and measures of aggression in
of adding mindfulness training to anger manage- laboratory studies. Pers Individ Differences 8:885–893.
ment interventions. Caprara GV, Paciello M, Gerbino M, Cugini C. 2007. Individual
differences conducive to aggression and violence: Trajectories
and correlates of irritability and hostile rumination through
adolescence. Aggr Behav 33:359–374.
Carver CS. 1975. Physical aggression as a function of objective self-
awareness and attitudes toward punishment. J Exp Soc Psychol
REFERENCES 11:510–519.
Carver CS, Scheier MF. 1981. Attention and Self-Regulation:
Anderson CA, Bushman BJ. 2002. Human aggression. Annu Rev A Control Theory Approach to Human Behavior. New York:
Psychol 53:27–51. Springer.
Anderson JC, Gerbing DW. 1988. Structural equation modeling in Chambers R, Lo BCY, Allen NB. 2008. The impact of intensive
practice: A review and recommended two-step approach. Psychol mindfulness training on attentional control, cognitive style, and
Bull 103:411–423. affect. Cogn Ther Res 32:303–322.

Aggr. Behav.
Mindfulness and Rumination 43

Chermack ST, Giancola PR. 1997. The relation between alcohol and Little RJA, Rubin DB. 1987. Statistical Analysis With Missing Data.
aggression: An integrated biopsychosocial conceptualization. New York: Wiley.
Clin Psychol Rev. Special Issue: Biopsychosoc Conceptualiza- Lopata C. 2003. Progressive muscle relaxation and aggression among
tions of Hum Aggression 17:621–649. elementary students with emotional or behavioral disorders.
Cheung G, Lau R. 2008. Testing mediation and suppression effects Behav Disord 28:162–172.
of latent variables: Bootstrapping with structural equation Lyubomirsky S, Nolen-Hoeksema S. 1995. Effects of self-focused
models. Organ Res Methods 11:296–325. rumination on negative thinking and interpersonal problem-
Coffey KA, Hartman M. 2008. Mechanisms of action in the inverse solving. J Pers Soc Psychol 69:176–190.
relationship between mindfulness and psychological distress. MacKillop J, Anderson EJ. 2007. Further psychometric validation of
Complement Health Pract Rev 13:79–91. the Mindful Attention Awareness Scale (MAAS). J Psychopathol
Collins K, Bell R. 1997. Personality and aggression: The dissipation- Behav Assess 29:289–293.
rumination scale. Pers Individ Differences 22:751–755. MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V.
Davis RN, Nolen-Hoeksema S. 2000. Cognitive inflexibility among 2002. A comparison of methods to test mediation and other
ruminators and non-ruminators. Cogn Ther Res 24:669–711. intervening variable effects. Psychol Methods 7:83–104.
Deffenbacher JL, Oetting ER, DiGiuseppe RA. 2002. Principles of MacKinnon DP, Lockwood CM, Williams CM. 2004. Confidence
empirically supported interventions applied to anger manage- limits for the indirect effect: Distribution of the product and
ment. Couns Psychol 30:262–280. resampling methods. Multivariate Behav Res 39:99–128.
Edmondson CB, Conger JC. 1996. A review of treatment efficacy for Martin LL, Tesser A. 1996. Some ruminative thoughts. In:
individuals with anger problems: Conceptual, assessment and Wyer Jr RS (ed.). Ruminative Thoughts, Advances in Social
methodological issues. Clin Psychol Rev 16:251–275. Cognition, Vol. IX. New Jersey: Lawrence Erlbaum Associates,
Fenigstein A, Scheier MF, Buss AH. 1975. Public and private self- pp 1–48.
consciousness: Assessment and theory. J Consult Clin Psychol McCullough ME, Rachal KC, Sandage SJ, Worthington EL,
43:522–527. Brown SW, Height TL. 1998. Interpersonal forgiving in close
Giancola PR. 2000. Executive functioning: A conceptual framework relationships: II. Theoretical elaboration and measurement.
for alcohol-related aggression. Exp Clin Psychopharmacol J Pers Soc Psychol 75:1586–1603.
8:576–597. McCullough ME, Bellah CG, Kilpatrick SD, Johnson JL. 2001.
Gross JJ. 1998. The emerging field of emotion regulation: An Vengefulness: Relationships with forgiveness, rumination, well-
integrative review. Rev Gen Psychol. Special Issue: New Dir Res being, and the big five. Pers Soc Psychol Bull 27:601–610.
Emotion 2:271–299. McNulty JK, Hellmuth JC. 2008. Emotion regulation and intimate
Grossman P. 2008. On measuring mindfulness in psychosomatic and partner violence in newlyweds. J Fam Psychol 22:794–797.
psychological research. J Psychosom Res 64:405–408. Miller N, Pederson WC, Earleywine M, Pollock VE. 2003.
Grossman P, Niemann L, Schmidt S, Wallach H. 2004. Mindfulness- A theoretical model of triggered displaced aggression. Pers Soc
based stress reduction and health benefits: A meta-analysis. Psychol Rev 7:75–97.
J Psychosom Res 57:35–43. Mor N, Winquest J. 2002. Self-focused attention and negative affect:
Hamilton NA, Ingram RE. 2001. Self-focused attention and coping: A meta-analysis. Psychol Bull 128:638–662.
Attending to the right things. In: Snyder CR (ed.). Coping Morren M, Meesters C. 2002. Validation of the Dutch version of the
With Stress: Effective People and Processes. London: Oxford aggression questionnaire in adolescent male offenders. Aggr
University Press, pp 178–195. Behav 28:87–96.
Harris JA. 1995. Confirmatory factor analysis of the aggression Morrow J, Nolen-Hoeksema S. 1990. Effects of responses to
questionnaire. Behav Res Ther 33:991–993. depression on the remediation of depressive affect. J Pers Soc
Harris JA. 1997. A further evaluation of the aggression questionnaire: Psychol 58:519–527.
Issues of validity and reliability. Behav Res Ther 35:1047–1053. Muthén LK, Muthén BO. 1998–2007. Mplus User’s Guide, 5th
Hayes SC, Strosahl KD, Wilson KG. 1999. Acceptance and edition. Los Angeles, CA: Muthén & Muthén.
Commitment Therapy: An Experiential Approach to Behavior Nolen-Hoeksema S, Morrow J, Fredrickson BL. 1993. Response
Change. New York: Guilford Press. styles and the duration of episodes of depressed mood. J Abnorm
Hennessy D, Wiesenthal D. 2004. Age and vengeance as predictors Psychol 102:20–28.
of mild driver aggression. Violence Vict 19:469–477. O’Connor DB, Archer J, Wu FWC. 2001. Measuring aggression:
Heppner WL, Kernis MH, Lakey CE, Campbell WK, Goldman BM, Self-reports, partner reports and responses to provoking scenar-
Davis PJ, Cascio EV. 2008. Mindfulness as a means of reducing ios. Aggr Behav 27:79–101.
aggressive behavior: Dispositional and situational evidence. Aggr Philippot P, Brutoux F. 2007. Induced rumination dampens
Behav 34:486–496. executive processes in dysphoric young adults. J Behav Ther
Ingram RE. 1990. Self-focused attention in clinical disorders: Review Exp Psychiatry 39:219–227.
and a conceptual model. Psychol Bull 107:156–176. Preacher KJ, Hayes AF. 2004. SPSS and SAS procedures for
Izard CE, King KA, Trentacosta CJ, Morgan JK, Laurenceau J, estimating indirect effects in simple mediation models. Behav Res
Krauthamer-Ewing ES, Finlon KJ. 2008. Accelerating the Methods Instrum Comput 36:717–731.
development of emotion competence in Head Start children: Prentice-Dunn S, Rogers RW. 1982. Effects of public and private
Effects on adaptive and maladaptive behavior. Dev Psychopathol self-awareness on deindividuation and aggression. J Pers Soc
20:369–397. Psychol 43:503–513.
Kinney TA, Smith BA, Donzella B. 2001. The influence of sex, Ramel W, Goldin PR, Carmona PE, McQuaid JR. 2004. The
gender, self-discrepancies, and self-awareness on anger and verbal effects of mindfulness meditation on cognitive processes and
aggressiveness among U.S. college students. J Soc Psychol affect in patients with past depression. Cogn Ther Res 28:
141:245–275. 433–455.
Lattin J, Carroll JD, Green PE. 2003. Analyzing Multivariate Data. Rude SS, Maestas KL, Neff K. 2007. Paying attention to distress:
Pacific Grove, CA: Brooks/Cole-Thomson Learning. What’s wrong with rumination? Cogn Emotion 21:843–864.

Aggr. Behav.
44 Borders et al.

Rusting CL, Nolen-Hoeksema S. 1998. Regulating responses to Sukhodolsky DG, Golub A, Cromwell EN. 2001. Development and
anger: Effects of rumination and distraction on angry mood. validation of the anger rumination scale. Pers Individ Differences
J Pers Soc Psychol 74:790–803. 31:689–700.
Ryan EL, Deci RM. 2000. Self-determination theory and the Teasdale JD, Segal Z, Williams JMG. 1995. How does cognitive
facilitation of intrinsic motivation, social development, and therapy prevent depressive relapse and why should attentional
well-being. Am Psychol 55:68–78. control (mindfulness) training help? Behav Res Ther 33:25–39.
Scheier MF, Fenigstein A, Buss AH. 1974. Self-awareness and Teasdale JD, Segal ZV, Williams MG. 2003. Mindfulness
physical aggression. J Exp Soc Psychol 10:264–273. training and problem formulation. Clin Psychol Sci Pract 10:
Segerstrom SC, Shortridge BE, Stanton AL, Alden LE. 2003. 157–160.
A multidimensional structure for repetitive thought: What’s on Thompson BL, Waltz JA. 2008. Mindfulness, self-esteem, and
your mind, and how, and how much? J Pers Soc Psychol unconditional self-acceptance. J Rat Emotive Cogn Behav Ther
85:909–921. 26:119–126.
Sexton KA, Dugas MJ. 2008. The Cognitive Avoidance Question- Trapnell PD, Campbell JD. 1999. Private self-consciousness and the
naire: Validation of the English translation. J Anxiety Disord five-factor model of personality: Distinguishing rumination from
22:355–370. reflection. J Pers Soc Psychol 76:284–304.
Shapiro SL, Carlson LE, Astin JA, Freedman B. 2006. Mechanisms Tyson P. 1998. Physiological arousal, reactive aggression, and the
of mindfulness. J Clin Psychol 62:373–386. induction of an incompatible relaxation response. Aggr Violent
Shapiro SL, Oman D, Thoresen CE, Plante TG, Flinders T. 2008. Behav 3:143–158.
Cultivating mindfulness: Effects on well-being. J Clin Psychol Van Dam NT, Earleywine M, Borders A. 2009. Item response
64:840–862. theory analysis of the Mindful Attention Awareness Scale; under
Shrout PE, Bolger N. 2002. Mediation in experimental and review.
nonexperimental studies: New procedures and recommendations. Vigil-Colet A, Lorenzo-Seva U, Codorniu-Raga MJ, Morales F.
Psychol Methods 7:422–445. 2005. Factor structure of the Buss-Perry aggression questionnaire
Singh NN, Lancioni GE, Singh Joy SD, Winton ASW, Sabaawi M, in different samples and languages. Aggr Behav 31:601–608.
Wahler RG, Singh J. 2007a. Adolescents with conduct disorder Wegner DM, Zanakos S. 1994. Chronic thought suppression. J Pers
can be mindful of their aggressive behavior. J Emotional Behav 62:615–640.
Disord 15:56–63. Wenzlaff RM, Wegner DM. 2000. Thought suppression. Annu Rev
Singh NN, Lancioni GE, Winton ASW, Adkins AD, Wahler RG, Psychol 51:59–91.
Sabaawi M, Singh J. 2007b. Individuals with mental illness can Werts CE, Linn RL, Joreskog KG. 1974. Intraclass reliability
control their aggressive behavior through mindfulness training. estimates: Testing structural assumptions. Educ Psychol Meas
Behav Mod 31:313–328. 34:25–33.
Sobel ME. 1982. Asymptotic intervals for indirect effects in Williams TY, Boyd JC, Cascardi MA, Poythress N. 1996. Factor
structural equations models. In: Leinhart S (ed.). Sociological structure and convergent validity of the Aggression Question-
Methodology. San Francisco: Jossey-Bass. pp 290–312. naire in an offender population. Psychol Assess 8:398–403.

Aggr. Behav.

You might also like