theodore roethke · study 1 pelle nigard, psychologist and course supervisor at the school of...
TRANSCRIPT
1
2
The Waking (1953)
…
Great Nature has another thing to do
To you and me, so take the lively air,
And, lovely, learn by going where to go.
This shaking keeps me steady. I should know.
What falls away is always. And is near.
I wake to sleep, and take my waking slow.
I learn by going where I have to go.
Theodore Roethke
3
4
1
Purity and Anger af Pernille Struer, 2011
2
3
Table of Content
Acknowledgements ...................................................................................................................................... 7
English summary .......................................................................................................................... 9
Danish summary ......................................................................................................................... 13
Chapter 1 Introduction to anger and aggression ........................................................................................ 17
Definitions of anger .................................................................................................................... 18
Anger as a social construct ......................................................................................................... 21
Conceptualization of anger......................................................................................................... 21
Definitions of aggression ........................................................................................................... 22
Types of aggression .................................................................................................................... 23
The social information processing approach .............................................................................. 24
Emotion and cognitive processing ............................................................................................. 26
Instrumental versus hostile aggression ....................................................................................... 26
Anger dysregulation ................................................................................................................... 27
Chapter 2 Psychopathology, anger and aggression .................................................................................... 31
Psychosis and aggression ........................................................................................................... 31
Anger as the mediator between psychopathology and aggression ............................................. 32
Anger and psychopathology ....................................................................................................... 33
Anger and depression/anxiety .................................................................................................... 34
Anger and PTSD ........................................................................................................................ 36
Anger and psychosis................................................................................................................... 38
Chapter 3 Psychopathology and information processing ........................................................................... 39
The cognitive system .................................................................................................................. 39
Selective attention ...................................................................................................................... 41
Threat detection .......................................................................................................................... 42
Rumination ................................................................................................................................. 43
Rumination and worry ................................................................................................................ 47
Thought suppression .................................................................................................................. 47
Rumination and suppression ...................................................................................................... 51
Metacognition............................................................................................................................. 53
S-REF model .............................................................................................................................. 53
Chapter 4 Assessment ................................................................................................................................. 59
Assessment of metacognition ..................................................................................................... 59
Assessment of anger ................................................................................................................... 60
4
PART 2 Overview of methodology ........................................................................................................................... 69
Overview of the thesis studies ................................................................................................... 69
Chapter 1 Development of the MAQ in a non-clinical setting .................................................................... 71
Introduction ................................................................................................................................ 71
Participants ................................................................................................................................. 71
Measures……………………………………………………………………………………….71
Procedure ................................................................................................................................... 72
Results ........................................................................................................................................ 73
Discussion .................................................................................................................................. 75
Chapter 2 Prisoners, anger, and the MAQ .................................................................................................. 77
Introduction ................................................................................................................................ 77
Participants ................................................................................................................................. 77
Measures .................................................................................................................................... 77
Procedure ................................................................................................................................... 78
Results ........................................................................................................................................ 78
Discussion Study 1 and 2 ........................................................................................................... 85
Chapter 3 Clinical patients, anger, and the MAQ ........................................................................................ 91
Introduction ................................................................................................................................ 91
Participants ................................................................................................................................. 92
Measures .................................................................................................................................... 93
Procedure ................................................................................................................................... 94
Hypotheses ................................................................................................................................. 94
Results ........................................................................................................................................ 97
Discussion Study 3 ................................................................................................................... 106
Chapter 4 Forensic patients, anger, aggression and the MAP .................................................................. 115
Introduction .............................................................................................................................. 115
Setting ...................................................................................................................................... 116
Participants ............................................................................................................................... 116
Measures .................................................................................................................................. 118
Procedure ................................................................................................................................. 121
Hypotheses ............................................................................................................................... 121
Results ...................................................................................................................................... 125
Discussion ................................................................................................................................ 137
5
Chapter 5 General discussion .................................................................................................................... 147
Threat and anger ....................................................................................................................... 148
Bodily arousal .......................................................................................................................... 148
Negative beliefs about anger .................................................................................................... 149
Types of self-focus ................................................................................................................... 149
Anger inhibition ....................................................................................................................... 150
Positive beliefs about anger...................................................................................................... 151
Metacognitive patterns ............................................................................................................. 151
Dual anger experience .............................................................................................................. 153
General metacognition ............................................................................................................. 154
Transdiagnostic approach ......................................................................................................... 155
Appendix A: The MAQ-1
Appendix B: The MAQ-2
Appendix A: The MAQ-3
Appendix B: The MAP
Appendix E: The MAP –Danish
Appendix F: Norms study of the NAS-PI
Appendix G: Metacognitive Profiling
6
7
Acknowledgements
I would especially like to thank all of the people who participated in these 5 studies.
A total of 1128 people participated.
I also want to express my gratitude to the University of Copenhagen, Department of
Psychology, for granting me the doctoral scholarship that supported this thesis.
I would also like to acknowledge the support and encouragement of Chief
Consultant Helle Hougaard and Chief Psychologist Tine Wøbbe in the Forensic Department at
the Mental Health Centre Sct. Hans during the writing and proposal processes for this project. In
addition, I appreciate the continuous kindness and openness of the Mental Health Centre Sct.
Hans and for them allowing me to be a part of the department.
The following people made specific parts of this research process possible, and I am
very grateful to them:
Study 1 Pelle Nigard, psychologist and course supervisor at the School of Police Education
in Denmark, and the teachers at the School of Police Education in Denmark.
Study 2 Consultants Anne Okkels Birk and Kuno Herman Lund at the Prison and Probation
Service in Denmark and the quality coordinators and local contact persons at the various prisons.
Study 3 Eric Simonsen, secondary supervisor and Chief of Research at the Psychiatric
Research Unit, Region Zealand, Heads of the Psychiatric Department District Naestved, Region
Zealand, Lisbeth Lund Pedersen and Tove Kjærbo and the managers of the 6 participating teams
in Naestved and Vordingborg. Also, the clinical staff who recruited participants should be
acknowledged for their efforts.
8
Study 4 Thomas Schütze, the Chief Consultant in the Department of Forensic Psychiatry at
the Mental Health Centre Sct. Hans hospital, Lene Berring, Head of Care and Development, and
the ward managers and clinical staff at the 8 wards in the forensic department of Sct. Hans
hospital.
Study 5 Psychologist and Research Assistant Vivian Heinola from the Psychology
Department at the University of Copenhagen. From the two wards in the psychiatric hospital in
Frederikssund, Helle Hougaard, Chief consultant at the Psychiatric Hospital in Frederikssund,
Mette Lynge (quality coordinator and nurse), and Anne Mette Larsen (quality coordinator and
nurse).
English summary
9
English summary
Dysregulated anger exists in both the presence and absence of psychopathology and
across a range of conditions. As such, anger is a transdiagnostic symptom.
Recently, clinical psychology has focused on common features across
psychopathology and general aspects of cognitive processing, and transdiagnostic approaches
have become influential. One generic model representing an information processing approach is
the metacognitive model proposed by Wells and Mathews (Wells & Matthews, 1994; Wells,
2000). Because this model offers a generic clinical conceptualization of cognitive processes
involved in emotional distress, it has been applied to a range of conditions including general
anxiety disorder, obsessive-compulsive disorder, posttraumatic stress disorder, depression, and
psychosis. However, few studies have explored the metacognitive components of anger, and at
present, there is no coherent metacognitive framework on anger. The goal of this thesis was to
apply a metacognitive framework to anger by developing a new self-report anger scale. Through
a theoretical discussion of anger and aggression in relation to psychopathology and information
processing, the metacognitive approach to anger is formulated in the first part of the thesis.
The second part of the thesis presents the four empirical studies that were conducted
in the development of the new anger scale. In the pilot of Study 1, the utility of a metacognitive
framework on anger was used to explore whether individuals hold both positive and negative
beliefs about the functions and nature of anger and if these beliefs are connected to particular
strategies for processing negative stimuli; finally, I looked for indications that a self-perpetuating
cycle of processing negative stimuli can occur. All concepts were confirmed, leading to the
construction of the Metacognition and Anger Questionnaire (MAQ-1), which was then tested in a
non-clinical sample to explore factor structure and reliability. Four empirically distinct and
reliable factors emerged:
o positive beliefs (“anger helps me handle threats and danger”)
o negative beliefs (“anger could make me go mad”)
o rumination (“I cannot let go of angry thoughts”)
o cognitive consciousness (“I am constantly aware of my thinking”)
The first three subscales demonstrated the expected associations with the Provocation Inventory,
(PI;(Novaco, 2003).
English summary
10
The purpose of Study 2 was to further test the psychometric properties of the MAQ-
2 in a sample with higher anger levels. In addition, the inclusion of a general metacognitive
measure was used to address convergent validity. The factor structure was reproduced with the
expected subscale intercorrelations as well as the correlations with the PI Total. The results
supported the MAQ-2 as a metacognitive measure with specific relevance to anger. However,
because the items designed to measure cognitive consciousness showed inconsistent factor
loadings and the concurrent validity was again unsatisfactory, this subscale was omitted.
In Study 3, the MAQ's metacognitive anger framework was tested in a mixed
clinical sample to evaluate its advantage in relation to anger over a general metacognitive
framework. Using CFA, the factor structure was confirmed, and reliability was satisfactory. In
addition, convergent validity of the subscales was assessed and found to be adequate. Because the
correlations between the general metacognitive measure, the MetaCognitive Questionnaire
(MCQ-30; (Wells & Cartwright-Hatton, 2004) and the anger measures were generally weaker
than for the MAQ, the latter was confirmed to be a measure with specific
relevance for anger. Themes of uncontrollability, danger, and madness in the regulation and
control of mental phenomena materialized as particularly related to anger regulation, in
agreement with the general metacognitive measure. Finally, the results suggested that rumination
not only maintains emotional distress but also maintains elevated bodily arousal. Lastly, a
fundamental hypothesis concerning the unique benefits of the MAQ as a metacognitive measure
of anger was tested. The MAQ subscales entered into a hierarchical regression after the MCQ-30
eliminated the effect of the MCQ-30 Total, supporting this hypothesis.
The goal of Study 4 was to test the psychometric properties of the revised measure
and to evaluate the validity of the measure for anger and aggression, specifically. This was
achieved by choosing a forensic population characterized by psychopathology as well as anger
problems. Prior to Study 4, the revisions to the measure based on studies 1, 2 and 3 had altered
the composition of the MAQ substantially compared to the metacognitive framework of Wells
and Matthews (Wells, 2000; Wells & Matthews, 1994), and thus, it was renamed to indicate the
proper affiliation. The result was the Metacognitive beliefs and Anger Processing (MAP) scale.
In this scale, I aimed to incorporate a thought control strategy, which may be involved in anger
dysregulation. Hence, a suppression subscale, modeled on the White Bear Suppression inventory
(WBSI;(Wegner & Zanakos, 1994), was included. Results confirmed the expected factor
English summary
11
structure and reliability was satisfactory. However, regarding anger suppression, none of the
following theoretical assumptions could be confirmed: that negative beliefs about anger would
motivate the individual to withhold anger expression, that failure to suppress anger would
activate rumination, or that the suppression of anger would increase anger related thoughts about
anger and thus be related to anger.
Regarding the convergent validity of the rumination subscale, several tests were
conducted, substantiating its validity. Moreover, results again, as in Study 3, suggested that
rumination is associated with physiological arousal and anger. Furthermore, rumination was
found to be associated with violent fantasies, which supports the notion that these constructs are
comparable. Psychotic symptoms were found to be associated with anxiety and anger. PTSD
symptoms were found to be associated with anxiety and anger, and self-harm was associated with
anxiety.
I tested the depression model proposed by Papageorgiou and Wells (2003) as an
exploratory, preliminary exercise, and structural equation modeling supported their approach.
These results should be treated with caution due to a small sample size, though. In this model,
positive beliefs about anger appeared closely linked to rumination, and rumination was closely
linked to negative beliefs and possibly bodily arousal, the latter of which is associated with anger.
In sum, negative beliefs may function as a mediator of the relationship between rumination and
anger. Because negative beliefs were more strongly associated with aggression than the other
subscales that showed a non-significant association with aggression, it was inferred that this
pattern of interactions would transfer to aggression as well. That only the MAP Negative Beliefs
subscale of the MAP was significantly associated with aggression may indicate a more complex
relationship between rumination and aggression than initially assumed. These results indicate that
uncontrollability and danger are principal themes in a metacognitive conceptualization of
emotional distress including anger/aggression.
Finally, because anger was associated with aggression, and in particular the arousal
and cognitive domains of anger, the importance of bodily arousal in problematic anger was
substantiated as well as the view that anger is cognitively mediated.
The thesis represents a transdiagnostic approach to the understanding and treatment
of dysregulated anger.
Danish summary
12
Danish summary
13
Danish summary
Dysreguleret vrede eksisterer både i fravær af psykopatologi og sammen med
psykopatologi foruden på tværs af kliniske tilstande. Således præsenterer vrede sig som et
transdiagnostisk symptom.
I nyere tid har klinisk psykologi fokuseret interessen på fællestræk på tværs af
kliniske tilstande og generelle aspekter af kognitiv informationsbearbejdning blevet
indflydelsesrige. En model for emotionel forstyrrelse der repræsenterer en transdiagnostisk
tilgang, er den metakognitive model udviklet af Wells og Mathews (Wells & Matthews, 1994;
Wells, 2000). Da denne model tilbyder en generel klinisk model af kognitive processer involveret
i emotionelle forstyrrelser, er den blevet anvendt på en række kliniske tilstande herunder
generaliseret angst, tvangstanker- og handlinger, post traumatisk stress tilstand og psykoser, for at
nævne nogle.
Kun få studier har dog udforsket de metakognitive komponenter i vrede og
nuværende findes der ingen sammenhængende metakognitiv model. Denne afhandling forsøgte at
anvende en metakognitiv tilgang til vrede ved at udvikle en ny selvrapporterings skala. I den
første del af afhandlingen danner en teoretisk diskussion af psykopatologi og
informationsbearbejdning i relation til vrede/aggression grundlaget for formuleringen af den
metakognitive tilgang til vrede.
I anden del af afhandlingen præsenteres de fire studier der blev gennemført i
udviklingen af den nye vrede skala. I pilot studiet fra Studie 1, blev det udforsket hvorvidt
individer har både positive og negative overbevisninger vedrørende vrede og om disse
overbevisninger relaterer sig til bestemte strategier til at bearbejde information. Derudover ledte
jeg efter tegn på at onde cirkler af bearbejdning af negative stimuli kan forekomme. Alle disse
antagelser blev bekræftet og spørgeskemaet Metacognition and Anger Questionnaire (MAQ-1)
blev konstrueret. Spørgeskemaet blev først testet i en ikke-klinisk population for at udforske
faktorstruktur og pålidelighed. Fire empirisk adskillelige og pålidelige faktorer viste sig
o positive overbevisninger (“vrede hjælper mig til at håndtere trusler og farer”)
o negative overbevisninger (“vrede kunne gøre mig vanvittig”)
Danish summary
14
o rumination (“jeg kan ikke give slip på vrede tanker”)
o kognitiv bevidsthed (”jeg er konstant opmærksom på min tænkning”)
De første tre subskalaer demonstrerede de forventede associationer til vrede målt
med Provocation Inventory (PI;(Novaco, 2003).
Formålet med Studie 2 var at teste de psykometriske egenskaber ved spørgeskemaet
yderligere i en population med høj forekomst af vrede. Derudover blev konvergent validitet testet
ved at inkludere et spørgeskema, der måler generelle aspekter af metakognition. Faktorstrukturen
blev reproduceret og de forventede inter-korrelationer mellem subskalaer og med PI viste sig.
Resultaterne støttede MAQ-2 som et metakognitivt spørgeskema med særlig relevans for vrede.
Dog, fordi de spørgsmål der var konstrueret til at måle kognitiv bevidsthed viste inkonsistente
faktorladninger, foruden ikke viste de forventede korrelationer til vrede (PI), blev denne subskala
opgivet.
For at teste spørgeskemaets egenskaber i en klinisk population og i forhold til et
generelt metakognitivt spørgeskema, blev det i Studie 3 testet i en blandet klinisk gruppe sammen
med MetaCognitive Questionnaire (MCQ-30;(Wells & Cartwright-Hatton, 2004). Ved at anvende
konfirmatorisk faktoranalyse blev faktorstrukturen reproduceret og pålideligheden var
tilfredsstillende. Derudover blev validiteten af subskalerne bekræftet. Da korrelationerne mellem
det generelle metakognitive spørgeskema og vredes spørgeskemaerne var mindre end for det nye
spørgeskema, bekræftede resultaterne MAQ-3 som et metakognitivt spørgeskema med særlig
relevans for vrede. I overensstemmelse med den generelle metakognitive model, viste temaer som
kontrol, fare og vanvid sig som centrale for regulering af vrede. Videre indikerede resultaterne at
rumination ikke bare vedligeholder følelsesmæssigt ubehag, men også det kropslige arousal.
Slutteligt blev en fundamental hypotese bekræftet da MAQ, i en hierarkisk regressionsanalyse,
eliminerede effekten af det generelle metakognitive mål (MCQ-30). Dette resultat bekræfter den
unikke betydning af det nyudviklede metakognitive spørgeskema (MAQ) med særlig relevans for
vrede.
I Studie 4 var målet at teste de psykometriske egenskaber af det reviderede
spørgeskema, foruden at evaluere validiteten i relation til vrede og særligt i relation til aggression.
Til dette formål blev en retslig patientgruppe med psykopatologi såvel som høj vredesdisposition
valgt. Inden Studie 4 blev spørgeskemaet revideret baseret på Studie 1, 2, og 3. Disse ændringer
Danish summary
15
var så substantielt afvigende fra den oprindelige metakognitive model foreslået af Wells og
Matthews, at spørgeskemaet skiftede navn. Resultatet var Metacognitive beliefs and Anger
Processing (MAP) skalaen. I dette spørgeskema inkluderedes en kognitiv kontrolstrategi som kan
være involveret i dysreguleret vrede. Således blev en subskala til at måle suppression, konstrueret
ud fra White Bear Suppression Inventory (WBSI;(Wegner & Zanakos, 1994), inkluderet.
Resultaterne bekræftede den forventede faktorstruktur og pålideligheden var tilfredsstillende.
Vedrørende validiteten af den nye suppressions-subskala, blev ingen af de følgende teoretiske
antagelser bekræftet: at negative overbevisninger vedrørende vrede motiverer individet til at
tilbageholde at udtrykke vrede; at ikke succesfuld suppression af vrede aktiverer rumination; eller
at suppression af vrede øger vredesrelaterede tanker og således er relateret til vrede.
Vedrørende validiteten af ruminations subskalaen, blev flere validitetstests
bekræftet. Derudover understregede resultaterne igen, som i Studie 3, at rumination er associeret
med kropslig arousal og vrede. Derudover var rumination relateret til voldelige fantasier, hvilket
styrker antagelsen om at disse begreber er sammenlignelige. Psykotiske symptomer var associeret
med angst og vrede, og selv-skade var associeret med angst.
Som en foreløbig udforskende øvelse, testede jeg den metakognitive
depressionsmodel (Papageorgiou & Wells, 2003) ved anvendelse af structural equation modeling
(SEM). Resultaterne støttede tilgangen. Positive overbevisninger vedrørende vrede viste sig
således tæt forbundne med rumination og rumination var tæt forbundet med negative
overbevisninger og muligt med kropsligt arousal, som er associeret med vrede. Samlet synes
negative overbevisninger at fungere som mediator mellem rumination og vrede. Fordi negative
overbevisninger var stærkere associeret med aggression end de andre subskala´er som viste en
ikke-signifikant association med aggression, synes dette mønster af sammenhænge at kunne
overføres til aggression. At kun negative overbevisninger var signifikant associeret med
aggression, indikerer tilstedeværelsen af en mere kompliceret sammenhæng mellem rumination
og aggression, end først antaget. Resultaterne peger på at ukontrollerbarhed og fare er
overordnede temaer i en metakognitiv model af emotionel lidelse, herunder også
vrede/aggression.
Afslutningsvist fordi vrede var associeret med aggression, særligt arousal
komponenten og den kognitive komponent, pegede resultaterne på vigtigheden af kropsligt
arousal ved problematisk vrede, og støttede det synspunkt, at vrede er kognitivt medieret.
Danish summary
16
Afhandlingen repræsenterer en transdiagnostisk tilgang til forståelsen og
behandlingen af dysreguleret vrede.
Chapter 1 Introduction to anger and aggression
17
Chapter 1 Introduction to anger and aggression
Anger is best understood as a complex set of interacting cognitive processes and
structures. In this thesis, cognition is principally conceived of as central to the instigation and
mediation of anger. As such, I will largely adhere to the view that emotions are the products of
cognitive processes. From this position, it follows that cognitive belief structures and cognitive
processes in relation to anger are meaningfully connected, and therefore, it may be clinically
worthwhile to integrate these elements into a framework of anger. In this introduction, the
following objectives will be addressed:
1. What are the essential belief structures about anger?
2. What roles do these belief structures play in the experience of anger?
3. What are the connections between belief structures and processing routines in anger?
4. What are the associations between processing routines and anger regulation?
In the aggression literature, the social information processing (SIP) approach (Crick
& Dodge, 1994; Dodge & Crick, 1990) is an influential theory that combines cognitive belief
structures with cognitive processes. The measure developed in this thesis is based on the
information processing approach.
First, a few comments on the conceptual connection between anger and aggression
may help to clarify the discussion. The link between anger and aggression is widely accepted in
the literature (Monahan, Steadman, Silver, Appelbaum, Robbins, and Mulvey et al., 2001;
Novaco, 1994). However, it is important to bear in mind that even though anger may lead to
aggression, anger in itself is not aggression. Furthermore, while investigating aggression,
researchers have to some degree neglected to address conceptual considerations and dynamic
cognitive processes involved in aggressive behavior (Nagtegaal, Rassin, and Murris, 2006) .
Attention has largely been directed towards identifying risk factors for aggressive behavior, while
anger as the construct underlying aggression has been explored in surprisingly few studies
(DiGiuseppe & Tafrate, 2007; Taylor & Novaco, 2005). Due to the close association between
anger and aggression, further investigation of the concept of anger is warranted for the prevention
of aggression.
Chapter 1 Introduction to anger and aggression
18
Inconsistent definitions of concepts like anger and aggression have caused
confusion and ambiguity in the literature (Spielberger & Reheiser, 2009). In this thesis, anger will
largely be conceived of as a complex and dynamic social construct involved in aggression.
Because terms like anger, aggression, hostility and irritability have been used interchangeably, a
brief conceptual clarification is required as a starting point.
As previously mentioned, anger has received surprisingly little attention in the
theoretical literature, while aggression has garnered more attention; several models of aggression
have been offered, including the frustration-aggression theory (Dollard & Miller, 1998; Dollard,
Doob, Miller, Mowrer, and Sears, 1939), the neo-associative network theory (Berkowitz, 1993;
Berkowitz, 1990) and the social learning model on aggression (Bandura, 1973). However, some
researchers have focused more directly on conceptualizing anger, such as in the works of Averill
(1982), Novaco (1976; 1994), Deffenbacher (1999), and Kassinove & Sukhodolsky (1995).
Spielberger (Spielberger, Reheiser, and Sydeman, 1995; 2009) focused heavily on anger in
relation to health issues but has also offered a conceptualization of anger. First, anger will be
addressed, followed by several remarks on aggression.
Definitions of Anger
Anger is an emotion that occurs in everyday life. Based on subjective reports, anger
is experienced between several times per day and several times per week (Averill, 1982). Hence,
anger must be considered a common emotion involved in everyday living and only under certain
circumstances does it call for clinical intervention. As far as a formal definition of anger is
concerned, most academic writers take the position that anger is a multidimensional affective
experience connected to an inner state of unpleasant arousal. The intensity of anger varies from
mild feelings of irritation to anger, rage, and hate (Berkowitz, 2005). Anger is predominately
defined in terms of its subjective and phenomenological qualities (Eckhardt, Norlander, and
Deffenbacher, 2004). Anger is described as, “…an emotional state that consists of feelings that
vary in intensity, with associated activation or arousal of the autonomic nervous system”
(Spielberger & Reheiser, 2009) or as, “an internal, mental, subjective feeling-state with
associated cognitions and physiological arousal patterns” (DiGiuseppe & Tafrate, 2007).
Nonetheless, these definitions suffer the limitation of being fairly general. Hence, in
theory they could describe almost any negative emotion that shares characteristics that are not
unique to anger. In order to differentiate between anger and other negative emotions, the urge to
Chapter 1 Introduction to anger and aggression
19
do harm to the perceived source of the aversive event has been suggested as an exclusive
characteristic of anger. For example, anger is described as a subjectively experienced, negatively
toned emotion composed of a state of inner arousal directed towards the perceived source of an
aversive event (Novaco, 1994). To continue with this definition of anger, unlike other negatively
toned emotions anger seems largely associated with approach rather than avoidance behavior
(Berkowitz & Harmon-Jones, 2004; DiGiuseppe & Tafrate, 2007; Novaco, 2010a). In conclusion,
anger is viewed as a complex emotion derived from thoughts, actions, impulses and physiology
in addition to cognitive processes.
Below, causes of anger are discussed with a focus on the cognitive components of
anger arousal.
Several causes of anger have been proposed, ranging from general causes, such as
frustration when goals are thwarted (Berkowitz, 1993; Dollard & Miller, 1998), to the specific
role of cognition in anger (Beck, 1999; Deffenbacher, 1999b; DiGiuseppe & Tafrate, 2007;
Novaco, 1994). The physiological component of anger is considered important, but because anger
and anxiety are relatively similar in this regard, features other than bodily arousal are involved in
the experience of anger (Bandura, 1973). As such, inferring the meaning of the arousal is pivotal.
Anger has been closely linked to the threat-perception system and ultimately to
survival responses (Chemtob, Novaco, Hamada, Gross, and Smith, 1997; Renwick, Black,
Ramm, and Novaco, 1997; Novaco, 1998); hence, bodily arousal in association with threat is
considered an important aspect of experiencing anger (Novaco, 1976). Also, less direct and less
physical threats are thought to be involved in anger arousal. The notion that humans seem
predisposed to anger arousal as a response to situations of perceived threat is widespread in the
literature. In chapter 3, in which information processing is discussed, the significance of
overestimating threat in the perception of stimuli is expanded. Other attributions typically linked
to activation of anger involve threats to one’s possessions, to preferred norms or social rules, or
more specifically, to personally significant concepts or ideas. In addition, allocation of resources
in a social context, in particular those that relate to social status or self-image, may elicit anger
when threatened. The theme of justification in relation to anger has been emphasized by several
researchers (Averill, 1983; Deffenbacher, 1999; DiGiuseppe & Tafrate, 2007; Taylor & Novaco,
2005). When ill-will is perceived and when an unpleasant act is considered voluntary, unjustified
and potentially avoidable, anger is likely to arise. According to Novaco, one of the most common
Chapter 1 Introduction to anger and aggression
20
forms of appraisal that instigates anger is, “perceived malevolence” (Novaco, 2000) or in the
words of Averill (1983, p. 1150), “More than anything else, anger is an attribution of blame”.
In work by Beck (1999) that focuses on cognitive precursors of anger arousal,
distorted cognitions overestimating threat are called to attention. Derived from underlying
distorted cognitive schemata, a variety of thoughts with themes of unfairness are triggered in a
range of different social situations.
In association with the tendency to perceive threat and attribute unfairness, blame
and malevolence also underlie the construct of hostility. In the literature, hostility is defined as a
set of negative attitudes involving unfavorable judgments and negative evaluations of other
people that motivate the wish to see the person harmed (Berkowitz, 1993; Eckhardt et al., 2004).
Hostility is understood as a trait characteristic that seems to increase the risk of perceiving threat
and malice from the environment (Epps & Kendall, 1995) and predisposes the individual to
respond aggressively in situations of perceived threat (Novaco, 1998). To simplify things, Epps
and Kendall (1995) conceptualized anger, aggression, and hostility as components of a joint
constellation within a social context. Anger is described as the affective component, hostility as
the cognitive component, and aggression as the behavioral component. In sum, hostility may be a
set of attitudes that underlie anger and motivate aggression, yet neither is necessary nor sufficient
to instigate anger or aggression. They are separate constructs, yet they are often highly correlated.
Like other emotions, anger serves as a critical psychological marker of a person’s
well-being. Generally, anger facilitates the impulse to assert boundaries and has been described
as, “… the emotional complement of the organismic preparation for attack…” (Novaco, 2000).
The functional value of anger is that it serves as an emotional cue of an unwanted state of affairs,
signaling that something needs to change and thereby energizing the organism to take action. To
elaborate on this point, animals use aggression to enforce their personal space and set boundaries
just like humans successfully use anger and aggression to guard self-esteem, secure personal
space and protect values (Novaco, 1998; Novaco, 2007). Emphasizing that anger, via its
activation of physical and psychological resources, assists the organism in enduring hardship and
pain and sharpens attentional focus aimed at overcoming the external threat (Taylor & Novaco,
2005), the evolutionary relevance of anger is unmistakable.
Overall, the literature on anger suggests that the cognitive themes involved in anger
arousal are the perception of threat, perception of unnecessary, unpleasant and unjust actions, and
Chapter 1 Introduction to anger and aggression
21
violations of personal values. Moreover, anger seems to assist the individual in securing
protection over various personally significant goals.
Anger as a social construct
Because the physiological responses of anxiety/fear and anger are relatively similar
(DiGiuseppe & Tafrate, 2007), conceptualizing anger merely as a biological syndrome is
insufficient. A person may behave aggressively out in fear as well as anger, which is why the
analysis of emotional behavior needs to take into consideration the social context in which it
occurs. Thus, according to the influential emotion theorist James Averill, to understand anger one
needs to focus attention on the functions that anger might serve within a broader social system.
Anger is argued to be involved in the maintenance of social order and in the hierarchical structure
of the social group (Averill, 1982; Novaco, 2007). Exemplifying this idea, based on a series of
studies in which Averill asked ordinary people about their everyday experiences with anger, he
argued that anger is positively reinforced (Averill, 1982; Averill, 1983). In Averill´s data, people
said that they came to realize their own faults because of the other person´s anger and that the
relationship with the angry person was strengthened rather than weakened. Furthermore, when
people were asked if they had viewed their angry episodes as beneficial or maladaptive, the ratio
of responses was 2.5 to 1, respectively. Consequently, people attributed both negative and
positive roles to anger, and overall, these reports signify a functional value for anger in the
regulation of social encounters. The outcome of anger, seen from the viewpoint of the angry
individual, is causing the desired effect in a social system, which is also recognized in the social
learning perspective (Bandura, 1973).
Conceptualization of anger
Novaco (1994) conceptualized anger in three domains: cognitive, arousal and
behavioral. The cognitive domain processes environmental stimuli, the arousal domain triggers
physiological activation, and the behavioral domain conveys the behavioral manifestations of
these connected domains. According to this clinical model, and in agreement with network theory
and the SIP model, these domains are reciprocally associated. Representing the cognitive domain
of the model, attentional focus, suspicion, rumination, and hostile attitudes are articulated. In the
arousal domain, bodily excitation is acknowledged as an essential feature of anger. Hence, the
transfer of residual arousal from one situation to another, as formulated in the excitation transfer
Chapter 1 Introduction to anger and aggression
22
theory by Zillmann (1979; Zillmann, 1988), is emphasized. Four subdivisions are identified:
intensity, duration, tension, and irritability. In the behavioral domain, the associated inclination to
act, impulsive reactions, verbal aggression, physical confrontation, and indirect expression of
anger are also dimensions of anger.
Because anger in this thesis was measured predominantly using the NAS, which is
the manifestation of this model of anger, the interplay of aggressive domains and their
subdimensions will later be explained in detail.
Overall, particular cognitive themes have been proposed as related to anger and the
survival and social functions of anger have been acknowledged. Anger may be conceptualized as
a complex construct involving cognition, bodily arousal and action impulses. Through the
incorporation of beliefs about anger into the metacognitive framework of anger presented in this
thesis, the cognitive aspects of anger are focused on to conceptualize how anger functions are
involved in shaping anger. When linking these beliefs about the functions of anger to anger
processing, anger regulation is addressed.
Definitions of Aggression
Inspired by Freud, who predominately viewed aggression as an instinct, Dollard and
Miller formulated their drive and frustration-aggression theories (Dollard & Miller, 1998; Dollard
et al., 1939), which were also supported by Berkowitz (1989). Later, social learning (Bandura,
1973) and the social information processing perspective (Crick & Dodge, 1994; Dodge & Crick,
1990) developed. Most writers agree that aggression is characterized as behavior with,“ a wish to
hurt,” in combination with the knowledge that the receiver is motivated to avoid the treatment
(Geen, 2001; Anderson & Bushman, 2002). Berkowitz (1993) defined aggressive behavior
broadly when he offered the following two descriptions of aggressive behavior: (a) a deliberate
attempt to achieve a goal, which may be the goal of injuring the other party either physically or
psychologically, or to assert dominance, regain control over a situation or teach the other party
not to be annoying, and so on; and (b) when the aggressor has the intention to do harm. Novaco
(1998) also emphasized malicious intentions when stating that aggression should be reserved for
acts with the intention to inflict harm or damage on a person or an object. Including the
motivational components, some acts that inflict harm do not qualify as aggressive, such as
Chapter 1 Introduction to anger and aggression
23
accidents or harm inflicted with good intentions (e.g., a dentist), or battle games and contact
sports (Anderson & Bushman, 2002; Geen, 2001).
As a result, the evaluation of aggressive acts must be considered according to the
resulting injury and the social judgments of the behavior. This means that whether or not a given
act is judged to be aggressive depends on several factors, and some of these factors relate to the
observer rather than the aggressor (Bandura, 1973; Feshbach, 1964).
Types of Aggression
The distinction between instrumental and hostile aggression was initially put
forward by Feshbach (1964). The hostile–instrumental aggression dichotomy profoundly
influenced aggression research and is discussed by most writers in aggression and social
psychology (Berkowitz, 1993; Barratt, 1994; DiGiuseppe & Tafrate, 2007; Novaco, 1998;
Novaco, 2007). A similar distinction is made between proactive and reactive aggression (Crick &
Dodge, 1996; Hanneke, Bram Orobio, Willem, Herman, and Welmoet, 2007; Raine, Dodge,
Loeber, Gatzke-Kopp, Lynam, Reynolds et al., 2006). In what has been labeled instrumental or
proactive aggression, the main goal is not to do harm but to serve other purposes. These purposes
may be demonstrating power, gaining money or other possessions or preserving social status.
Thus, the aggression is a means to obtain a desired goal. Instrumental aggression is considered to
be a, “cold”, planned, unprovoked behavior where the aggressor is not in a state of arousal and is
indifferent about their victim's injuries. In what has been labeled "hostile" or "emotional or
reactive" aggression, the notion of emotion plays a more prominent role. With reference to some
kind of frustrating inner arousal based on the perception of intentionally harmful acts committed
by the aggressor towards the object of the aggression, the aggressive act may serve the purpose of
retaliation. The aggressor either finds hurting the victim satisfying or reacts impulsively without
considering the consequences. Emotional, hostile or reactive aggression can be understood as a
“hot,” angry, impulsive, and defensive response to a provocation or frustration. Working to
conceptualize impulsiveness and its relationship with aggression, Barratt (1994) classified
instrumental aggression as learned aggression and emotional/hostile aggression as impulsive
aggression; it was argued that impulsive aggression is related to personality traits, such as
impulsiveness and anger-hostility. When hostile aggression results in the aggressor obtaining
satisfaction by hurting the victim or successfully regulating the arousal state, hostile/emotional
Chapter 1 Introduction to anger and aggression
24
aggression is reinforced by the same means as instrumental aggression (Novaco, 1976). As such,
hostile aggression can also be argued to operate as a learned social behavior.
The dichotomous view of aggression is differentiated mainly by (a) the primary
goal; (b) the presence of anger; and (c) the amount of cognitive processing involved. However,
aggression in most situations occurs as a mixed type motivated by complex goals. Differentiating
types of aggression based on the presence of anger may also be problematic because some forms
of hostile resentment result in well-planned acts of revenge with a potentially long delay between
the initial provocation and the execution of the revenge. Finally, instrumental aggression should,
in theory, involve a greater amount of planning and the activation of higher cognitive processes
than hostile aggression, the latter of which should proceed more automatically and impulsively as
a result of the increased physiological arousal of anger. However, due to repeated rehearsal, an
aggressive act may become automatic with only a small amount of conscious cognitive
processing. In this way, even complex sets of actions can become automatic, and the amount of
information processing (automatic versus controlled) does not necessarily parallel the type of
aggression (instrumental versus hostile). An instrumental aggressive act may be a routine
behavior that does not require higher cognitive functioning.
For these reasons, some researchers have argued that the dichotomy of instrumental
and hostile aggression has outlived its usefulness (Bushman & Anderson, 2001; Novaco, 1998).
The following discussion on the social information processing approach to understanding
aggression will specify the theoretical suggestions for how information processing occurs
differently in different types of aggression.
The social information processing approach
Clinical models of anger and aggression generally incorporate notions of cognitive
processes, as in the social information processing (SIP) (Crick & Dodge, 1994; Dodge & Crick,
1990) and network models (Bower, 1981; Berkowitz, 1990).
In these clinical models, cognitive scripts and schemata are thought to influence the
repertoire of emotional and behavioral responses to social stimuli. Beginning in early childhood,
people learn schemas and scripts that influence how they perceive, interpret, judge, and respond
to events in their lives. In cases of enhanced prior experiences with hostile stimuli, aggressive
scripts and schemata may be overly primed. Through the spreading of the related associations in
cognitive networks, hypersensitivity to hostile stimuli may develop. In this way, and in
Chapter 1 Introduction to anger and aggression
25
accordance with network theories, prior experiences with hostile information processing
influence and shape future social encounters by priming the activation of related responses. Due
to rehearsal, these hostile responses may result from minimal threat cues and occur so quickly
with limited cognitive awareness that the engagement of the appraisal system is sparse
(Berkowitz, 2008; Bushman & Anderson, 2001b). The repeated activation of a particular
response deepens that neural pathway, increasing the efficiency and speed of the response and
decreasing the likelihood of deviations from the response.
The social information processing approach by Crick and Dodge (1994; 1996;
Dodge & Crick, 1990) explains how cognitive structures stored in long-term memory influence
ongoing cognitive processing. According to this theory, information processing occurs in several
stages, and attentional processes have different effects depending on the stage of the processing
sequence. In the encoding step, which is primarily a sensory, perceptual process involving short-
term memory, encoding may be guided by selective attention based on prior learning experiences
(e.g., selective attention to threatening cues). Due to the presence of hostile schemata stored in
long-term memory, individuals with high trait anger may be more likely to selectively attend to
hostile situational and internal cues (attentional bias) (Crick & Dodge, 1994).
Regarding the second step of the information processing sequence, aggressive
individuals are proposed to be more prone to interpreting ambiguous stimuli in a hostile manner
than non-aggressive individuals. Social cues are interpreted under the influence of social
knowledge that is already stored in memory (social schemata), and if this information is
predominantly hostile, interpretation may be biased towards hostility.
In the third and fourth steps of the information processing sequence, a search in
long-term memory for a response relies on social knowledge from prior experiences stored in
long-term memory. Thus, the selected goal is based on social schemata that may lead to a
maladjusted response selection.
In the fifth step of the information processing sequence, cognitive beliefs are
involved because the selected response is evaluated in three domains: (a) an evaluation of the
attractiveness of the strategy; (b) an evaluation of the expectancies of outcome; and (c) an
evaluation of personal success in implementing the strategy. Regarding evaluation of the
attractiveness of the strategy, aggressive children evaluate aggressive behavior more favorably
than prosocial behavior compared to nonaggressive children (Dodge & Crick, 1990). These
individuals are hypothesized to hold the belief that aggression is a desirable behavior. Regarding
Chapter 1 Introduction to anger and aggression
26
the evaluation of expectancies of outcomes, aggressive children expect more favorable
interpersonal outcomes for aggressive behavior than nonaggressive peers (outcome efficacy)
(Crick & Dodge, 1996) and may hold the belief that aggression will serve them well in their
efforts to succeed in the world. Finally, regarding the evaluation of personal success, aggressive
children report more confidence that they can behave aggressively than nonaggressive peers (self-
efficacy).
The operation of on-line social information processing that is based on cognitive
structures stored in long-term memory makes processing more efficient and faster; however, due
to distorted cognitive schemata and beliefs, this type of processing may result in poor judgment
and lead to maladjusted behavior.
Emotion and cognitive processing
The SIP model outlines how emotional arousal influences social information
processing along several steps in the sequence. In a situation of increased emotional arousal, the
preferential encoding of certain stimuli over others is hypothesized to cue attention. When highly
aroused, attention is allocated to the emotionally arousing stimuli at the expense of other stimuli.
An angry individual will therefore be more likely to attend to hostile stimuli.
Current emotional state is also suggested to affect interpretation (Anderson, 1997).
When we are angry, the likelihood of interpreting a particular situation negatively is increased
because schemata that are consistent with anger are more accessible. In addition, under the
influence of high emotional arousal, further cognitive processing is blocked, and interpretation is
based entirely on the automatically activated hostile schemata.
In the third step of information processing, the clarification of a goal is influenced
by the naturally occurring motivation to regulate arousal. As such, a highly aroused individual
may be motivated to achieve an immediate reduction in arousal rather than forming a long-term
goal that is more adaptive.
Instrumental versus hostile aggression
Based on the information processing approach in relation to instrumental
aggression, the belief that aggression is an efficient means of obtaining a desired goal is
hypothesized to influence the likelihood of an aggressive act specifically in the goal clarification
and evaluation steps of the information processing sequence. Beliefs about aggression as a
Chapter 1 Introduction to anger and aggression
27
strategy to achieve a desired goal have been empirically associated with aggression (Huesmann,
1988; Huesmann & Guerra, 1997; Archer & Haigh, 1997a; Archer & Haigh, 1997b; Bellmore,
Witkow, Graham, and Juvonen, 2005; Bailey & Ostrov, 2008). According to the information
processing approach, hostile aggression should be strongly affected by emotion, which
predominately influences the first two steps in the processing sequence as discussed above.
Regarding the role of emotion in social information processing, an individual´s
experiences with their own emotions may be involved. Anger can be described as an eruptive,
unsettling and intensely emotional experience. Individuals with prior experiences of intense anger
may thus form schemata in which they view anger as uncontrollable. Believing that anger is
uncontrollable may result in experiencing decreased competence in regulating one's emotional
state. According to the SIP approach, these hypothesized schemata may influence goal evaluation
because if an individual does not believe that he or she is able to regulate emotional arousal in a
controlled manner, the goal will not be attempted.
To conclude, instrumental aggression seems to be aligned with positive beliefs
about the function of aggression or anger. Hostile aggression seems to be associated with the
experience of emotional responses as uncontrollable, which in turn forms or strengthens negative
beliefs about anger. With reference to the claim that most forms of aggression are mixed, most
individuals can simultaneously hold positive and negative beliefs about anger.
The anger scale developed in this thesis incorporates these different types of
aggressive responses. The developed scale implies that different types of beliefs about
anger/aggression are simultaneously in operation and probably affect how anger-related stimuli
are processed in different ways.
The social information processing model explains how aggressive scripts/schemata
influence stimulus processing along various steps in the information processing sequence, and the
metacognitive framework suggests a clinical relevance for conceptualizing how specific beliefs
about anger influence anger processing. As such, the metacognitive framework suggested in this
thesis is fundamentally consistent with the SIP model.
Anger dysregulation
Anger processing may result in an adaptive and controlled regulation of anger
arousal, or it may result in what would be categorized as dysregulated anger. In the following
Chapter 1 Introduction to anger and aggression
28
section, dysregulated anger is discussed within a social information processing model and with an
eye towards the potential role of cognitive structures in anger.
Anger is considered dysregulated and categorized as a clinical condition when it is
judged to be maladaptive. Typically, dysregulated anger is associated with too frequent, too
intense or too prolonged anger (Novaco, 2007) as well as when it triggers excessive aggression or
violence.
Anger can be experienced and outwardly appear as a turbulent, powerful, and
eruptive emotion. Classically, it has been seen as a mental disturbance, a madness or an insanity;
there has been a general view that anger is an uncontrollable, diseased state of mind (Novaco,
2010a; Potegal & Novaco, 2010). In examining the construct of anger, Novaco (1976; Novaco,
2007; Novaco, 2010a; Potegal & Novaco, 2010) used the Roman Janus sculpture, which depicts
two faces pointing in opposite directions, to illustrate the duality of anger. On one hand, anger is
associated with eruptive and destructive feelings linked to madness. On the other hand, anger is
associated with an energizing and empowering emotional experience linked to survival systems.
In these discussions regarding what anger signifies about the angry individual, it is
assumed, at least to some extent, that anger is the manifestation of belief structures within the
individual. As such, negative beliefs about anger as an eruptive, uncontrollable emotional
experience may play a role in the regulation of anger by influencing goal selection during the
information processing sequence. As outlined above, when it is perceived that anger will be
poorly controlled, forming a goal of controlled anger regulation is unlikely.
Conceptualizations in which anger is regarded as a powerful, energizing emotion
that assists the individual in overcoming threats and dangers may also be internalized and
manifested as positive beliefs about anger; in turn, these beliefs also influence the social
information processing sequence. Particularly in the early stages of attention allocation, it is
suggested that positive beliefs about anger may result in selective attention to hostile stimuli.
Regarding goal clarification, it is hypothesized that in the goal evaluation and selection steps of
information processing, positive beliefs about anger increase the likelihood of forming and
selecting responses that maintain anger arousal. If individuals believe that anger will help them
overcome adversity and protect against threat and danger, goals that down-regulate anger seem
less likely.
In conclusion, anger may be dysregulated due to positive beliefs that anger is a
means to succeed or dysregulated due to the experience of not being able to control the
Chapter 1 Introduction to anger and aggression
29
physiological arousal of anger. The new anger measure developed in this thesis is designed to
include both negative and positive beliefs about the function and nature of anger in
conceptualizing how anger is processed.
Chapter 1 Introduction to anger and aggression
30
Chapter 2 Psychopathology, anger and aggression
31
Chapter 2 Psychopathology, anger and aggression
In the following section, the relationship between psychopathology and aggression
is briefly discussed, and it will be argued that although a range of mechanisms are at play in this
association, anger is critical and worth focusing on in the prevention of aggression among
psychiatric patients. What follows is a discussion of the association between anger and
psychopathology, with a focus on how this relationship influences individual beliefs about the
function and nature of anger.
Psychosis and aggression
It is generally agreed upon that an association between psychotic disorders and
aggression/violence exists (Douglas, Guy, and Hart, 2009; Walsh, Buchanan, and Fahy, 2002;
Joyal, Dubreucq, Gendron, and Millaud, 2007; Fazel, Gulati, Linsell, Geddes, and Grann, 2009;
Hodgins, 2008; Link, Monahan, Stueve, and Cullen, 1999; Bo, Abu-Akel, Kongerslev, Haahr,
and Simonsen, 2011). For example, from an epidemiological 3-year birth cohort investigation in
Denmark that included 358 180 individuals, Brennan, Mednick, and Hodgins (2000) reported that
individuals hospitalized for a major mental disorder had an increased risk of violent offense
compared to individuals who had never been hospitalized for a major mental disorder. The effect
remained when controlling for socioeconomic status, personality disorders and substance abuse.
Understanding of the association between psychotic disorders and crime may benefit from
investigations of the process and interplay between various factors involved in this association
rather than the presence of a psychotic disorder per se (Sirotich, 2008).
In summary, Douglas et al. (2009) state that methodological differences, study
design, choice of sample type and a range of moderators and confounding variables contribute to
these diverse findings and a complex relationship between psychosis and aggression. For
instance, many studies compare mentally disordered offenders with the general population and
find an increased risk of aggression/violence; however, if compared to a known criminal
population, mentally disordered individuals have a decreased risk of aggression/violence
compared to offenders without mental disorders (Bonta, Law, and Hanson, 1998). Douglas et al.
(2009) summarized that the important question regarding the association between psychosis and
violence is: “What particular symptoms of psychosis, under which situational circumstances, and
Chapter 2 Psychopathology, anger and aggression
32
in combination with which personal or situational factors, are associated with increased or
decreased risk of various kinds of violence?”
In conclusion, it seems reasonable to argue that the association between
psychopathology and aggression/violence is complex and that the mere presence of a mental
disorder is not sufficient to explain elevated levels of aggression/violence. This implies that
largely, mental disorders are indirectly associated with aggression. Mentally disordered
individuals are thus expected to react aggressively for the same reasons as people without mental
disorders. As a result, aggressive responses are likely to take place in situations where the
individual perceives unfairness, provocation and personal threat and when regulatory or other
inhibitors of aggression are not effective.
Anger as the mediator between psychopathology and aggression
Some researchers have reported a more prominent association between anger and
aggression than between psychotic symptoms and aggression (Kay, Wolkenfeld, and Murrill,
1988; Appelbaum, Robbins, and Monahan, 2000; Soyka, Graz, Bottlender, Dirschedl, and
Schoech, 2007; McNiel, Eisner, and Binder, 2003). Supporting that anger-related concepts are
more predictive of future violence than psychotic symptoms, Syoka et al.(2007) found, in a large
longitudinal study among patients diagnosed with schizophrenia, that hostility at admission and
discharge significantly predicted violent offense in a 7-12 year follow-up period. Neither
paranoid–hallucinatory, psycho-organic, obsessive–compulsive, manic, apathetic,
catatonic/stuporous nor autonomic symptoms predicted future violence. Hostility was measured
by suspiciousness, dysphoria, irritability, aggressiveness, lack of feeling ill, lack of insight, and
uncooperativeness; these measures probably captured an emotional construct resembling anger
even more than an attitudinal disposition. In a psychiatric setting using a self-report retrospective
design, McNiel et al. (2003) found that the presence of a schizophrenic diagnosis in itself was not
correlated with violent behavior. However, anger, persecutory beliefs, threat/control-override
(TCO) symptoms and external hostile attributions predicted violent behavior. In four different
models exploring the associations between these different symptoms while controlling for age,
substance abuse, manic disorder, depressive disorder and schizophrenic disorder, the model that
used anger as a predictor was the most successful at predicting violent behavior. Moreover, in a
longitudinal study conducted in a psychiatric setting, violent variables were investigated week-to-
week, allowing for conclusions about causality; Skeem, Schubert, Odgers, Mulvey, Gardner, and
Chapter 2 Psychopathology, anger and aggression
33
Lidz (2006) found that hostility measured by the hostility subscale of the BSI (Derogatis &
Melisaratos, 1983) in one week uniquely predicted violence in the next week. Because this
subscale assesses emotional reactivity as opposed to an attitude construct, it is reasonable to state
that the instrument measures anger rather than hostility (Jarvis & Novaco, 2006). The effect
remained even when controlling for the effects of violence and hostility (anger) in the previous
week.
In summary, in a recent review of the correlates of violence among persons with
mental disorders, Sirotich (2008) advised that researchers should further explore the role of anger
as a risk factor for aggression/violence. The positive association between anger and aggression in
clinical samples has been substantiated by a large body of research within institutions (Novaco,
1994), within forensic institutions among the intellectually disabled (Novaco & Taylor, 2004),
within forensic inpatients (Doyle & Dolan, 2006), within the community (Monahan et al., 2001),
and among juvenile offenders (Cornell, Peterson, and Richards, 1999).
In conclusion, anger must be considered a robust correlate of aggression across a
range of clinical settings. Anger may even mediate the general relationship between mental
disorders and aggression. In the following section, the relationships between anger and different
selected psychopathologies are briefly discussed.
Anger and psychopathology
Anger symptoms are manifested across a range of psychopathologies; however, this
issue has not been systematically addressed in the diagnostic classification system (DiGiuseppe
& Tafrate, 2007). Anger is known to be experienced as part of a range of psychopathologies (e.g.,
intermittent explosive disorder, PTSD, psychosis, borderline personality disorder, paranoid and
narcissistic personality disorders and major depression) (Barazzone & Davey, 2009; Novaco,
2010a; Wilkowski & Robinson, 2008).
Dysregulated anger may occur in both the absence and presence of
psychopathology and as a transdiagnostic symptom across disorders. This, of course, raises the
question of what role anger plays in psychopathology and if the mechanisms are similar across
disorders.
In a study including 1300 outpatients in a general psychiatric setting, Posternak and
Zimmerman (2002) explored the prevalence of self-reported anger and found elevated anger in
51% of cases. Normative data from two of the most widely used anger self-report scales, the
Chapter 2 Psychopathology, anger and aggression
34
STAXI-2 (Spielberger, 1999) and the NAS (Novaco, 2003), support the high prevalence of anger
in psychopathology. In the STAXI-2 data, which included 1600 normal adults and 274
hospitalized psychiatric patients, the psychiatric patients had significantly higher self-reported
anger scores than normal adults. Regarding the NAS, Jones, Thomas-Peter and Trout (1999)
found significantly higher self-reported anger in clinical patients compared to non-clinical
individuals.
Because anger is a key symptom in a range of psychopathologies including
psychosis, PTSD, and depression/anxiety, these associations are briefly discussed below. Because
the anger measure developed in this thesis focuses on conceptualizing cognitive belief structures
and information processing, the potential role of cognitive belief structures is also highlighted
here. These beliefs structures include both positive and negative beliefs about the nature and
functions of anger.
Anger and depression/anxiety
It is commonly accepted that irritable mood can be associated with mood disorders
(American Psychiatric Association, 2000). Hence, to presume that there is at least a moderate
correlation between anger and depression/anxiety seems reasonable. In support of this idea,
Posternak & Zimmerman (2002) demonstrated that major depressive disorder was associated
with the level of self-reported anger in a large dataset from psychiatric inpatients.
The question of how anger and depression/anxiety are related remains. Some of the
relevant issues are whether anger is causing depression or if anger is a consequence of
depression/anxiety that indicates general distress. In classic psychoanalysis, depression is thought
to be anger turned inward towards the self. As a result, a causal relationship between depression
and anger has been suggested by several prior theories of depression (DiGiuseppe & Tafrate,
2007; Novaco, 2010a). These theories are similar in their ideas about anger inhibition. Anger
turned inward, or anger suppression, was hypothesized to result in depressive symptoms. The
underlying mechanism involved in this association is the need to express emotions to stay
psychologically sound. However, definitions of the concepts are usually vague. An exception is a
study by P. Gilbert, J. Gilbert, and Irons (2004) that explored unexpressed anger in relation to
depression. In this study, the researchers explored whether unexpressed anger preceded the
development of depression. They found that 56% of depressed individuals reported that they had
restrained their expressions of anger before the onset of their depression. However, because there
was no control group in the study, the authors could not control for unexpressed anger in non-
Chapter 2 Psychopathology, anger and aggression
35
depressed individuals. In the same study, people were asked about their reasons for not
expressing their anger. These reasons focused on negative evaluations and loss of relationships
with significant others as well as the fear of losing control.
Reasons such at these are included in the metacognitive framework of anger
formulated in this thesis and are conceived of as negative beliefs about anger. Thus, negative
beliefs about anger may contribute to anger inhibition, which may then produce depressive
symptoms. However, this hypothesis does not address how anger inhibition is associated with
depression.
Other research has suggested almost the opposite relationship; namely, that anger
may be a consequence or a specific characteristic of depression. Empirical research on “anger
attacks” in depression supports this idea. Anger attacks are defined as impulsive, abrupt episodes
of anger with aroused physiology (Fava & Rosenbaum, 1998) and have been associated with
depression at prevalence rates generally between 30% and 40% (Painuly, Sharan, and Mattoo,
2005; Fava & Rosenbaum, 1998). Sudden anger attacks may be connected to unexpressed anger
because failure to express anger can result in a "bottling-up" of emotions followed by a sudden
and abrupt “attack” of negative emotions when the physiological system can no longer be
contained.
Again, failure to express anger may be explained by negative beliefs about anger. It
seems intuitive that anger may not be expressed due to a fear of negative evaluations. Failure to
express anger may lead to depressive symptoms, and in people with depression, it may lead to
anger attacks. In summary, anger is not expressed due to negative beliefs about anger, which in
turn produces depressive symptoms and aids in the maintenance of increased physiological
arousal that can suddenly detonate.
Others have suggested that the association between anger and depression may be
accounted for when considering depression with anger as a type of disorder from the bipolar
spectrum (Benazzi, 2003; Perlis, Smoller, Fava, Rosenbaum, and Nierenberg, 2004).
Other research points to a third variable as the mediator of the relationship between
depression and anger (DiGiuseppe & Tafrate, 2007). DiGiuseppe and Tafrate argue that the
correlation between anger and depression may be overestimated due to a lack of controls for
anxiety. They reported, referring to own data that the correlation between anger and depression
disappears when controlling for anxiety. This means that anger and depression are correlated with
anxiety, and anxiety is the "true" variable accounting for the link between anger and depression.
Chapter 2 Psychopathology, anger and aggression
36
It is hypothesized that due to the automaticity of the anger response and aversive feelings of
threat, anger may become associated with negative experiences. These experiences may manifest
as negative beliefs about anger.
In contrast to the suggestion that anxiety mediates depression and anger, one study
found a closer association between depression and anger attacks than between anxiety and anger
attacks (Gould, Ball, Kaspi, Otto, Pollack, Shekhar, et al. 1996); the authors suggest that co-
morbidity of anxiety and depression may be responsible for the connection between anxiety and
anger attacks. In support of this, a recent study found no unique association between anger and
anxiety and concluded that the association between anger and anxiety is largely due to symptoms
of depression (Moscovitch, McCabe, Antony, Rocca, and Swinson, 2008). In their study
exploring anger in the context of anxiety, they compared anxious patients with a non-clinical
control group and found that anxious patients reported more anger than controls. However, this
effect vanished when controlling for depression.
To conclude, even though the nature of the association between depression/anxiety
and anger is not unambiguous, negative beliefs about anger may be involved. The involvement
may be both in guiding particular responses, such as when it was suggested that negative beliefs
result in anger inhibition in depression, and it may also occur as the result of anger that has been
automatically activated in situations of perceived threat.
Anger and PTSD
Anger is a possible, but not necessary, symptom of PTSD (American Psychiatric
Association, 2000). As such, it is thought that anger could be involved in PTSD. In a meta-
analysis, Orth and Wieland (2006) found large effect sizes between anger and PTSD.
Anger as a core component of PTSD and trauma reactions is exemplified by studies
on the after-effects of combat or studies on the long-term effects of trauma (Novaco, 2010a). In
combat veterans, anger was found to account for 40% of the variance in PTSD symptoms even
when anger items in the PTSD scale had been removed (Novaco & Chemtob, 2002). The
association between anger and PTSD may be causal with anger as the risk factor, it may be due to
symptom overlap, or it may be that the relationship is more complex (Meffert, Metzler, Henn-
Haase, McCaslin, Inslicht, Chemtob, et al., 2008; Orth, Cahill, Foa, and Maercker, 2008).
Supporting the notion of anger as a risk factor, Feeny, Zoellner, and Foa (2000) used a
prospective design to investigate the role of anger in PTSD. In a regression analysis, anger
Chapter 2 Psychopathology, anger and aggression
37
expression at 1-month post-assault was predictive of the severity of PTSD symptoms 3 months
after the assault. In addition, a study on a population of police officers found that trait anger
predicted the development of PTSD (Meffert et al., 2008). However, Meffert et al. (2008) found
that state anger was strongly associated with PTSD symptoms even after controlling for trait
anger. As such, anger may be both a consequence and a predictor of PTSD. Further evidence
from a longitudinal study on crime victims points towards a more complex relationship (Orth et
al., 2008). Controlling for prior PTSD symptoms, the authors did not find evidence that anger
predicted PTSD. However, they did find that PTSD symptoms in the months after the traumatic
event predicted anger. In addition, they suggested that rumination may be responsible for the
relationship between anger and PTSD, although they did not use a validated scale to measure
rumination. They also argued that future theories should seek to explain how PTSD symptoms
influence anger and not how anger influences PTSD symptoms.
Chemtob, Novaco, Hamada, Gross, and Smith (1997) proposed a model accounting
for the link between PTSD symptoms and anger. In combat, which is clearly a situation of high
threat, anger may have beneficial functions due to its association with aggression; aggression in
turn may energize attack and survival behaviors. However, this “survival mode” may be overly
primed due to extended exposure to threat stimuli and as a result trigger anger in inappropriate
contexts. When "survival mode" is triggered, arousal increases, cognitive processing decrease,
speed of behavioral reactions increases, and the processing of threat-related stimuli is given first
priority.
In essence, a "survival mode" that includes strong anger arousal and a decrease in
self-monitoring and anger regulation can be activated maladaptively in response to perceived
threat in people with PTSD. The pivotal themes involved in the association between PTSD and
anger are thus threat perception and coping with threat. Furthermore, Novaco and Chemtob
(2002) posited that anger triggered in PTSD may involve intensified and refractory physiological
arousal due to a prior situation. Beliefs that the nature and function of anger are protective may
have a role in this relationship such that they prime activation of the “survival mode”. In addition,
it has been speculated that the increased and refractory physiological arousal associated with
anger arousal in PTSD may be intense and is likely to be experienced as beyond one's willful
control. This experience may lead to negative beliefs regarding anger.
In conclusion, the relationship between anger and PTSD is undoubtedly complex.
Belief structures about the function and nature of anger appear to play an active part in this
Chapter 2 Psychopathology, anger and aggression
38
relationship. In particular, the positive belief that anger serves to protect against threat and danger
and the negative belief that anger is uncontrollable.
Anger and psychosis
In a recent chapter on anger and psychopathology, Novaco (2010a) noted that in
patients suffering from psychosis, the relationships between anger and violence have been
investigated, while the relationships between anger and psychosis have been overlooked. When
acknowledging that anger can be understood as fundamentally linked to threat perception, which
is a key feature of psychosis, this seems odd. For example, paranoid schizophrenia often involves
persecutory delusions that include, “anticipation of danger” with components of physical, social,
or psychological threats (Freeman & Garety, 2000). As noted earlier, perceptions of threat and
danger perceptions are involved in the instigation of anger as a survival reaction. Potegal and
Novaco (2010) provide an in-depth narrative of how anger has been historically, linguistically
and semantically related to the concept of psychosis. Because threat perception defines delusions
of persecution, it is suggested that the link between anger and psychosis may be particularly
strong in regard to these specific symptoms. As far as anger and delusions of persecution are
concerned, Freeman and Garety (2003) suggested that the link involves the theme of being
deliberately wronged, generating delusions such as, “people are doing things to annoy me”. The
theme of being deliberately wronged is, as already discussed, closely linked to anger. In addition,
experiences of persecution seem logically linked to the survival systems described earlier. In this
sense, the anger response may be understood as a protective response based on experiences of
danger.
In sum, anger appears semantically related to psychosis in that they share themes of
threat, danger and persecution and purportedly involve beliefs about anger as protective and
helpful. As argued earlier, the activation of a survival response may trigger negative experiences
of anger, thus cementing negative beliefs.
In the following section, information processing patterns involved in emotional
disorders including anger are discussed, and a metacognitive framework connecting cognitive
belief structures with processing strategies is presented.
Chapter 3 Psychopathology and information processing
39
Chapter 3 Psychopathology and information processing
The information-processing approach to emotional disorders has expanded
cognitive theory from predominately investigating the content of cognitions to also exploring
cognitive processes (Wells, 2008; Harvey, Watkins, Mansell, and Safran, 2004; Williams, Watts,
MacLeod, Mathews, 1996b). The notion that particular disturbances in the regulation of cognition
and emotion are important features of psychopathology is widely accepted (Aldao, Nolen-
Hoeksema, and Schweizer, 2010). As a result, an increasing amount of clinical literature has been
dedicated to specifying how cognitive processes are regulated. Recently, important advances
have focused on general aspects of cognitive processing in relation to psychopathology, resulting
in a transdiagnostic position that explores common features across diagnoses (Harvey et al.,
2004). Said differently, cognitive emotion regulation theory has been attempting to bridge
different types of disorders (Aldao & Nolen-Hoeksema, 2010; Aldao et al., 2010). While still
focusing on common processes across psychological disorders, Wells and colleagues have
developed the self-reflective executive functioning model (S-REF model) (Wells & Matthews,
1994; Wells, 2000). This clinical model is known as the metacognitive model and provides a
general clinical conceptualization of information processing in psychopathology.
This chapter consists of a general discussion of the link between information
processing and psychopathology in general and more specifically in anger. Finally, the
metacognitive framework serving as the underpinning theoretical model of the new anger
measure developed in this thesis is presented. The S-REF model, with its concept of a Cognitive
Attentional Syndrome (CAS), is discussed with an eye towards the social information processing
approach (SIP) introduced earlier as it relates to anger and aggression.
The cognitive system
Because capacity limitations of the cognitive system are inevitable, information
processing is influenced at different stages by operating cognitive processes. For instance,
selective attention operates at an early stage to prioritize the available stimuli for further
processing in the system (Williams et al., 1996b). When selective attention is considered
dysfunctional, it is labeled attentional bias. Other cognitive processes involved in continued
Chapter 3 Psychopathology and information processing
40
processing consist of reasoning, memory, and responses to internal outputs (thoughts and
emotions) (Harvey et al., 2004).
Cognitive processes are commonly viewed as involving automatic as well as
strategic processes. As far as information processing and psychopathology are concerned, some
researchers have emphasized automaticity (Williams et al., 1996b; Williams et al., 1996a) while
others have emphasized the strategic aspects of the processes (Wells & Matthews, 1994; Wells,
2000; Matthews & Wells, 2000a), and others have advocated for combinations of automatic and
strategic processes (Cisler & Koster, 2010e; Beck & Clark, 1997).
Because of rehearsal, in the SIP model (Crick & Dodge, 1994; Dodge & Crick,
1990) a pattern of hostile responses is suggested to occur predominately as an automatic process;
this process occurs with relatively limited cognitive awareness and only modest activation of the
appraisal system. Through automatic spread to the associated neural networks (Berkowitz, 1993;
1990; Bower, 1981), this approach emphasizes automaticity. In the recently developed integrative
model of trait anger and reactive aggression by Wilkowski and Robinson (2008; 2010), three
variables account for information processing as it relates to anger. The first variable, the hostile
interpretation of a situation, typically operates as an automatic process due to rehearsal as
described in the SIP model. The second variable concerns prolonged attention to hostile stimuli;
this process, the capturing of attention, is consistent with the term rumination and is
predominately viewed as an automatically occurring process that amplifies existing anger. Lastly,
effortful control is emphasized as a potential moderator in this otherwise relatively automatic
anger-processing sequence. According to the model, the available strategies for controlling anger
are reappraisal, distraction and suppression.
In conclusion, I will argue that if cognitive processing of anger is viewed as
containing at least some features of strategic processing, specifying the mechanisms involved in
the selection and execution of strategic processing routines become theoretically as well as
clinically important. The anger scale developed in this thesis attempts to conceptualize processing
routines that maintain anger.
The literature regarding information processing in emotional distress, as well as in
relation to anger, is discussed below. Attentional processes, rumination and suppression are
addressed. Deviant attention allocation has been closely tied to various psychopathologies, and
rumination and suppression as responses to internal outputs (thoughts and emotions) have been
identified as two particular dysfunctional responses to unwanted negative thoughts across these
Chapter 3 Psychopathology and information processing
41
disorders (Nolen-Hoeksema, 1998; Watkins & Moulds, 2009; Aldao & Nolen-Hoeksema, 2010;
Purdon, Rowa, and Antony, 2005; Harvey et al., 2004; Nolen-Hoeksema, Wisco, and
Lyubomirsky, 2008; Nolen-Hoeksema et al., 2008).
Selective attention
There is a large body of research documenting the link between deviant attentional
processes and psychopathology, including self-focused attention (Ingram, 1990; Morrison &
Haddock, 1997; Woodruff-Borden, Brothers, and Lister, 2001), attentional avoidance (Chen,
Ehlers, Clark, and Mansell, 2002), disengagement difficulties (Moritz & Laudan, 2007; Cisler &
Koster, 2010; Koster, De Raedt, Goeleven, Franck, and Crombez, 2005), and threat detection
(Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, and van Ijzendoorn, 2007; Cisler &
Koster, 2010). The investigation of attentional processes has been conducted primarily within
specific disorders including anxiety (Cisler & Koster, 2010; Bar-Haim et al., 2007), depression
(Koster et al., 2005), psychosis (Moritz & Laudan, 2007; Morrison, Gumley, Schwannauer,
Campbell, Gleeson, Griffin, et al., 2005) and PTSD (Bryant & Harvey, 1997).
However, the findings allow for conclusions across disorders (Harvey et al., 2004; Ingram, 1990;
Mathews, Ridgeway, and Williamson, 1996; Matthews & Wells, 2000; Wells & Matthews, 1996;
Woodruff-Borden et al., 2001).
The deviant attentional patterns in psychopathology are characterized by: (a)
selective attention favoring attention to disorder-specific stimuli that reflect important personal
concerns consistent with the current concern model (Klinger, 1996); (b) difficulties in
disengaging from personally significant stimuli, consistent with the strategic processes view
(Beck & Clark, 1997; Wells & Matthews, 1994) (rumination) and; (c) a component of strategic
attentional avoidance (Green, Williams, and Hemsley, 2000; Cisler & Koster, 2010; Baegels &
Mansell, 2004) (suppression).
Selective attention can be categorized in the following way: (a) attention to
concern-relevant internal stimuli (self-focused attention) and (b) attention to concern-related
external stimuli (threat detection). Self-focused attention refers to an awareness of self-referent,
internally generated information (Ingram, 1990). By endorsing awareness for stored negative
material and increasing access to negative self-conceptions, self-focused attention intensifies and
prolongs negative emotional states. In addition, chronically self-focused attention may initiate an
emotional disorder consistent with a stress-vulnerability model (Ingram, 1990). Because some
Chapter 3 Psychopathology and information processing
42
aspects of self-focus intuitively seem benign, some have argued for the relevance of
differentiating between “normal” self-focus and dysfunctional self-focus. Differentiating by
parameters such as degree, duration and flexibility (Ingram, 1990), the valence of self-focus
(Woodruff-Borden et al., 2001) or the balance between self-focus and external focus (Segerstrom,
Stanton, Alden, and Shortridge, 2003) has been suggested. The need to conceptualize and
understand differences between adaptive and maladaptive self-focus is acknowledged by several
authors (Watkins, 2008; Segerstrom et al., 2003; Smith & Alloy, 2009).
Selective attention to threatening stimuli, or threat detection, refers to a process of
excessive attention towards potentially threatening external stimuli during an early stage in the
information processing sequence (Williams et al., 1996b). Two related though distinct concepts
are threat interpretation bias, in which neutral or ambiguous stimuli are interpreted in a
threatening manner (Harvey et al., 2004) or hostile attributional bias, in which malicious intention
is perceived. According to the SIP model, threat detection, threat interpretation bias, and hostile
attributional bias appear to be the product of schemata stored in long-term memory. Next, the
tendency to allocate more attention to threatening stimuli is addressed.
Threat detection
Even though deviant attentional patterns in psychopathology seems to cut across
disorders, variation in the content of the stimuli involved in threat detection across types of
psychopathologies would be in accordance with the current concern model (Klinger, 1996) (i.e.,
threat detection in relation to anxiety may involve content-specific fears, whereas in anger, threat
detection may be more likely to involve stimuli representing threats to personal values or direct
threats to physical well-being). On the other hand, overlap in content across types of
psychopathologies is also expected (Wenzel & Lystad, 2005). Furthermore, the threat perception
process may also be different, manifesting as distinct responses to the perception of threat across
disorders. While threat perception in anxiety disorders is associated with pathological “flight”
responses, threat perception in relation to anger is associated with increased “fight” responses
(Novaco, 2007).
In a study demonstrating attentional bias in anger using a pictorial Stoop task, it was
reported that high-anger students demonstrate an attentional bias for angry faces even when
controlling for anxiety (Van Honk, Tuiten, de Haan, and van den Hout, 2001). Supporting the
link between aggression and threat detection, it has also been reported that offenders with more
assaults are more likely than offenders with fewer assaults to detect threatening words in a
Chapter 3 Psychopathology and information processing
43
dichotic listening task (Seager, 2005; James & Seager, 2006). Extending this line of research to a
forensic sample, Smith and Waterman (2003) found that violent offenders demonstrated
significantly higher vigilance for aggressive words in a dot probe test than a sample of
undergraduates. Furthermore, in a Stroop task, violent offenders showed significantly more
interference than undergraduates for aggressive words compared to negative-emotion words.
Interestingly, the study also revealed that undergraduates with high levels of self-reported anger
showed the same attentional processing biases as the violent offender group (Smith & Waterman,
2003).
According to the concept of current concern advocated by Klinger (1996), people
will selectively attend to stimuli that are related to their own current concerns, which may include
detecting and eliminating danger. In the SIP model, attention allocation in high-trait anger
individuals is automatically biased toward hostile stimuli because these schemata are overly
primed (Crick & Dodge, 1994). However, attentional biases may also be influenced through other
types of cognitive structures such as cognitive beliefs related to anger and aggression (e.g., a
belief that anger/aggression serves a protective purpose for the individual). In this way, it is
speculated that believing that anger is helpful will guide attention towards threatening stimuli and
increase the risk of experiencing anger. Therefore, positive beliefs about anger were incorporated
into the present metacognitive questionnaire based on the assumption that positive beliefs guide
attention towards threatening stimuli,.
Rumination
Although the majority of research on rumination focuses on the relationship
between rumination and depression, there are numerous suggestions that rumination is an
important process in several psychological disorders (Harvey et al., 2004; Smith & Alloy, 2009).
Rumination has been associated with various psychopathologies other than depression including
anger (Simpson & Papageorgiou, 2003; Rusting & Nolen-Hoeksema, 1998), anxiety (Segerstrom,
Tsao, Alden, & Craske,2000), and PTSD (Orth et al., 2008).
Rumination is defined as a mental control strategy in which a person repetitively
focuses attention on negative feelings or personal problems and dwells on causes and
consequences without constructive actions to relieve the symptoms (Nolen-Hoeksema, 1991).
Rumination is seen in both normal subjects and in clinical patients, yet in clinical patients it is
more prolonged and associated with more subjective distress (Nolen-Hoeksema et al., 2008).
Chapter 3 Psychopathology and information processing
44
Experimental studies have been used to investigate the association between
rumination and aggression including displaced aggression (Bushman, Bonacci, Pedersen,
Vasquez, and Miller, 2005), aggressive responses to insults (Collins & Bell, 1997), aggression in
combination with irritability (Caprara et al., 2007), and aggressive responses in interaction with
frustration as a third variable (Vasquez, Bartsch, Pedersen, and Miller, 2007). Regarding angry
rumination, Wilkowski, Robinson and Meier (2006) found that students low in agreeableness
showed a prolonged processing of hostile stimuli and proposed that this prolongation of attention
can be conceived as rumination. Regarding the disengagement of attention from relevant stimuli
(disorder relevant and mood-congruent), in an experimental study in a non-clinical sample using
a visual search task, high-trait anger individuals allocated more attention to anger-related stimuli
when insulted than low-trait anger individuals (Cohen, Eckhardt, and Schagat, 1998).
Furthermore, high-trait anger individuals reacted more strongly to an insulting situation than low-
trait anger individuals. These results were supported in a study using a word task (Eckhardt &
Cohen, 1997). In this study, when high-trait anger individuals were insulted, they demonstrated
longer response latencies to angry words than to neutral words when they had not been insulted.
This effect was not found for the low-trait anger individuals (Eckhardt & Cohen, 1997). In
another study using a Stroop task, men who had abused their wives were relatively slower in
responding to aggressive words than to neutral words when compared to normal controls (Chan,
Raine, and Lee, 2010). A possible interpretation is that high-trait anger individuals have more
difficulty disengaging from an insult. Difficulty in disengaging from stimuli is argued to
resemble the construct of rumination.
The next logical question becomes why people ruminate. Usually, rumination is
employed by distressed people as an attempt to gain self-insight or to solve problems
(Papageorgiou & Wells, 2001b). In a qualitative study exploring rumination in clinically angry
patients, Simpson and Papageorgiou (2003) reported that all the patients ruminated and held
positive metacognitive beliefs about angry rumination. The beliefs about the functions and
benefits of rumination may be involved in the selection of this coping response. If an individual
believes that rumination will help solve problems, the probability of selecting rumination as a
coping response is increased. No studies, however, confirm that rumination solves problems
(Nolen-Hoeksema & Morrow, 1993; Nolen-Hoeksema, 2004).
Chapter 3 Psychopathology and information processing
45
In the study by Simpson and Papageorgiou (2003), patients also identified negative
metacognitive beliefs about the impact of rumination on anger and social functioning. Negative
beliefs may result from intense feelings of lack of control over the emotional experience that
follows the ruminative proces.
Several studies have been conducted on the effects of angry rumination. Using
experimental designs, a number of researchers have found an association between angry
rumination and anger. Sukhodolsky et al. (2001) used the STAXI (State-Trait Anger Expression
Inventory (Spielberger, 1988; Spielberger, 1999) to validate their new anger rumination scale
(ARS). In the development of the present anger measure, the STAXI-2 as well as the ARS were
used to validate the scale and will be further discussed in Study 3.
Regarding long-term effects of rumination, in a large community sample Nolen-
Hoeksema (2000) found that rumination among non-depressed people at one timepoint predicted
depression at timepoint two. In relation to health, a group of researchers found that long-term
effects of rumination were related to increased intensity and duration of affect and bodily arousal
(Thomsen, Mehlsen, Olesen, Hokland, Viidik, Avlund, et al., 2004). Hence, by increasing bodily
arousal, rumination has been found to exert an effect on the arousal domain of anger (Gerin,
Davidson, Christenfeld, Goyal, and Schwartz, 2006c; Ray et al., 2008; Ottaviani, Shapiro, and
Fitzgerald, 2010).
Violent fantasies have also been explored because they share features similar to
rumination. Using data from the large MacArthur Violence Risk Assessment Study (Monahan et
al., 2001; Steadman et al., 1994), Grisso, Davis, Vesselinov, Appelbaum, and Monahan (2000)
reported that hospitalized patients with persistent violent fantasies engaged in more violence in
the period after discharge compared to hospitalized patients without violent or occasionally
violent fantasies. Furthermore, the severity of symptoms was associated with violent fantasies.
This may indicate that greater stress results in a limited capacity to access infrequently rehearsed
cognitive scripts, leaving frequently rehearsed aggressive scripts to guide responses to threatening
stimuli. These results may point to the role of bodily arousal as a mediator where more arousal
means less of an ability to alter the automaticity of information processing. As such, in situations
of frequently rehearsed aggressive scripts co-occurring with stress, the ongoing processing of
aggressive scripts is not interrupted. Nagtegaal and Rassin (2004) used the same procedures to
assess violent fantasies in a non-clinical sample as in the Grisso et al. study, and found a
correlation between violent fantasies, hostility, and self-reported aggression.
Chapter 3 Psychopathology and information processing
46
In Study 4 of this thesis, the associations between violent fantasies and the newly
developed anger scale, as well as future aggression, is explored using the same procedures as
Grisso et al. (2000) to assess violent fantasies.
Theoretical accounts of how rumination brings about the effects described above,
such as the response styles theory (Nolen-Hoeksema, 2004; Nolen-Hoeksema, 1991), have
suggested that rumination exacerbates and prolongs depression by enhancing the effect of
negative mood on thinking. Prolonged negative mood results in more easily activated negative
thoughts, an interruption of efficient problem solving and finally, interference with instrumental
behavior that could have relieved the individual from the stressful situation. Within a network
theory framework (Berkowitz, 1993; 1990; Bower, 1981), rumination is claimed to maintain
anger because it increases the probability of activating concepts, thoughts, memories, etc. related
to the current angry mood. Activation of these related concepts reactivates anger. The SIP
framework accentuates that angry rumination favors the processing of anger/aggression-related
stimuli over other information, leading to increased automaticity and stability of anger-related
responses by generating additional connections to other concepts in a person's memory
(Huesmann, 1988). Thus, rumination is involved in the central mechanisms by which aggressive
scripts are stored and structured in memory. Furthermore, the excitation transfer theory
(Zillmann, 1979) proposes that rumination is associated with anger because it maintain bodily
arousal, and the risk of transporting residual excitation across situations is increased causing
easier activation of an angry response in a new situation.
Overall, based on the presented empirical findings the consequence of angry
rumination seems to be increased intensity and duration of affect, including bodily arousal,
resulting in maintenance of angry mood and increased activation of related concepts in the neural
network.
Rumination is assumed to occur both voluntarily and involuntary. Based on the
empirical findings relating to rumination, it is hypothesized that positive beliefs may be involved
in the voluntary selection of rumination as a strategy for processing negative affect. However,
due to its contribution to negative feelings, the byproduct of rumination may be a negative
experience perceived as uncontrollable and involuntary, perhaps even manifesting as negative
beliefs about rumination.
Chapter 3 Psychopathology and information processing
47
In conclusion, it is hypothesized that rumination is associated with anger through
mediation by increased bodily tension. Moreover, it is hypothesized that positive beliefs about
anger increase rumination, which consequently leads to increased negative beliefs about anger.
Rumination and worry
Rumination and worry are considered to be related concepts by several researchers
(Segerstrom et al., 2000). Ingram (1990) suggested a general concept labeled repetitive thought,
while Harvey et al. (2004) suggested the label recurrent negative thinking. Nolen-Hoeksema
(2008) stated that rumination and worry are both processes of repetitive and self-focused thought.
She further argued that rumination and worry are statistically distinguishable and that differences
may be found in the degree of perceived uncontrollability. People will ruminate when they
perceive no control over events, and they will worry when they see events as potentially
controllable, she claimed. Others argued that rumination and worry share the same processes but
have different content (Watkins, Moulds, Mackintosh, 2005); worry focuses on future events and
rumination centers on past events (Papageorgiou & Wells, 1999). The close association between
rumination and worry is supported by significant correlations between scales measuring
rumination and worry (Segerstrom et al., 2000; Watkins et al., 2005; Fresco, Frankel, Mennin,
Turk, and Heimberg, 2002).
The fact that rumination and worry are closely associated may indicate the
involvement of anxiety in rumination.
Thought suppression
Thought suppression is defined as a mental control strategy referring to the act of
intentionally trying not to think about something (Wenzlaff & Wegner, 2000; Purdon, 1999).
Generally, it is assumed that thought suppression is used when thoughts create unpleasant
emotions (Wegner & Zanakos, 1994) . Suppression differs from repression because the latter is
an unconscious and unintentional process. Some writers neglect to specify a precise definition of
suppression, causing theoretical confusion and inconsistency in empirical findings. Sometimes
intrusive thoughts themselves are characterized as thought suppression, while at other times they
are not (Segerstrom et al., 2003; Smith & Alloy, 2009). Furthermore, definitions of the term
suppression in the literature can be found to include a range of related but distinct constructs (i.e.,
thought suppression, expressive suppression (inhibited expression of an emotional experience),
Chapter 3 Psychopathology and information processing
48
experiential avoidance (mental distraction) and behavioral avoidance). These terms, which
constitute different constructs, may have different associations with psychopathology and make
specificity crucial (Aldao et al., 2010). To avoid conceptual confusion, suppression as it is
investigated within the emotion regulation literature (Gross & John, 2003; Gross, Richards, and
John, 2006) is not included in the present discussion.
The suppression paradigm was conceptualized by the two subscales of the White
Bear Thought Suppression Inventory (WBSI;(Wegner & Zanakos, 1994), which evoke
involuntary intrusive thoughts and attempts to suppress thoughts. Wegner suggested that because
intrusive thoughts are frequent among non-clinical and clinical populations, the mechanism
associating intrusive thought with psychopathology is suppression. Occasionally, all people need
to suppress certain thoughts in order to succeed in self-control and to achieve certain goals that
require thought suppression. However, some people use suppression as a mental strategy more
than others and across different situations and thought topics (Wegner & Zanakos, 1994). Ideally,
thought suppression should rid the individual of the unwanted thought and leave no trace;
however, the process does not seem to be that simple.
Research on the effects of thought suppression has produced conflicting results.
Some studies report negative effects of suppression as a thought control strategy while others
report positive effects of thought suppression (Boden & Baumeister, 1997). This may partially be
due to variations in methodology, sample type and individual success in suppressing thoughts
(Abramowitz, Tolin, and Street, 2001; Purdon, 1999; Wenzlaff & Wegner, 2000). In general,
thought suppression does seem to increase the frequency of the same thoughts that one is
attempting to suppress although this outcome varies from study to study (Abramowitz et al.,
2001; Harvey et al., 2004).
The empirical literature on suppression as it relates to anger is sparse, and the
suppression paradigm as proposed by Wegner and colleagues has only been adopted
unambiguously by one researcher, to my knowledge (Nagtegaal & Rassin, 2004; Nagtegaal et al.,
2006; Nagtegaal, 2008). The State-Trait Anger Expression Inventory-2 (Spielberger, 1999) is
widely used in the literature on anger suppression. In this self-report assessment tool, the Anger
Expression In (AX-I) subscale is used as a measure of anger suppression. However, the content
of the items in the AX-I focuses on verbal and behavioral inhibition of anger. Therefore the AX-I
does not measure suppression as it is conceptualized in the suppression paradigm, and the
Chapter 3 Psychopathology and information processing
49
literature is somewhat unclear as to what precisely is suppressed (i.e., the internal experience of
anger or the drive to express anger overtly). Conversely, some have proposed that the STAXI-2
AX-I measures rumination and have found high correlations between STAXI-2 AX-I and
rumination (Sukhodolsky et al., 2001).
At any rate, this conceptualization of suppression differs from that proposed by
Wenzlaff and Wegner (2000). The majority of studies exploring the effects of anger suppression
neither refer to the suppression paradigm nor focus on the effects of anger levels. Instead, this
body of research focuses on the effects relating to health (Venable, Carlson, and Wilson, 2001),
pain (Quartana and Burns, 2007; Burns, Quartana, and Bruehl, 2008) or depression (DiGiuseppe
& Tafrate, 2007; Sperberg & Stabb, 1998).
However, some studies have focused on anger. In one study using an experimental
design in a student population, the effectiveness of different emotion regulation strategies on
anger was explored (Szasz, Szentagotai, and Hofmann, 2011). In this study, the suppression
strategy was not effective in relieving the angry experience because the participants who were
instructed to suppress anger remained angrier than those instructed to reappraise. In two other
studies that were part of her doctoral work, Nagtegaal (2008) focused on dysfunctional thought
processes in relation to aggression. In these studies using non-clinical samples and the original
WBSI (Wegner & Zanakos, 1994) and the general thought control questionnaire (TCQ;(Wells &
Davies, 1994), she investigated the impact of thought control strategies on self-reported
aggression. In both studies, suppression was positively correlated with self-reported aggression
(Nagtegaal & Rassin, 2004; Nagtegaal et al., 2006). Still, one should bear in mind that these
studies used non-clinical populations, which may compromise generalization. Nagtegaal also
used general measures of thought suppression, which means that the valence and content of the
thought suppression is unknown. Furthermore, self-report of aggressive behavior was used.
In a review of over-controlled anger (anger inhibition), Davey, Day, and Howells
(2005) suggested that inhibiting anger does decrease its behavioral expression but has the
unintended side effect of maintaining internal anger arousal. In support of this, (Richards &
Gross, 1999) argued that thought suppression seems to enhance physiological arousal. Finally,
Gilbert et al. (2004) found that anger inhibition is associated with depression because patients
reported having inhibited their anger before their current depressive episodes. An interesting
aspect of this study was that it investigated the reasons for inhibiting anger; the explanations were
Chapter 3 Psychopathology and information processing
50
all related to negative beliefs about the consequences of expressing anger (e.g., fear of rejection
by others, fear of losing control, and fear of harming others).
Regarding a theoretical understanding of the effects of anger suppression, the
following mechanisms have been identified: (1) The monitoring mechanism. When attempting to
suppress a thought, the process of monitoring the occurrence of the suppressed thought will
ironically lead increased accessibility of the unwanted thought (ironic process theory) (Wenzlaff
& Wegner, 2000; Purdon, 2004); this effect has been referred to as the "paradoxical effect" of
suppression. The “immediate paradoxical effect”, in which subjects report increased target
thoughts during the suppression task, shows inconsistent replication. Some authors have
suggested that the WBSI measure fails to induce suppression (Rassin, 2003). This is supported by
the paradoxical finding that suppression is more replicable when cognitive resources are
occupied, causing a large cognitive load that results in a higher risk of failure to suppress
thoughts (Harvey et al., 2004; Wenzlaff & Wegner, 2000; Wenzlaff & Bates, 2000; Wenzlaff,
2004). A “delayed paradoxical effect” (post-suppression rebound effect) was initially
demonstrated with “The White Bear Experiment” (Wegner et al., 1987; Egner, Schneider, Carter,
and White, 1987). This experiment showed that after a period of suppressing thoughts about a
white bear, people reported more thoughts about the bear than people who had not been
instructed to suppress thoughts about a white bear. (2) The vulnerability mechanism. When
thought suppression is applied persistently, a lack of habituation to the unwanted thought will
sustain the emotional impact of the thought, resulting in hypersensitivity to the specific thoughts
people are motivated to suppress (Wegner & Zanakos, 1994). This mechanism may even be
enhanced because an increase in negative emotion will lead to more attempts to suppress the
associated negative thoughts. (3) The self-distraction mechanism. When people are trying to
avoid certain thoughts, they often try to distract themselves by thinking about something else;
however, the choice of a distracter is likely to resemble the unwanted thought (Wenzlaff,
Wegner, and Klein, 1991);(the mood-state-dependent-rebound effect). Moreover, when using
distracters to rid oneself of unwanted thoughts, the person may unintentionally develop implicit
associations between the distracters and the unwanted thought. As such, the distracters may
become triggers of the unwanted thought (Wegner & Zanakos, 1994).
The results of the studies conducted by Nagtegaal (2008) were also viewed in light
of the thought suppression paradigm. She proposed that when the individual suppresses violent,
intrusive thoughts, they unintentionally become hyper-accessible according to the ironic process
Chapter 3 Psychopathology and information processing
51
theory. Thus, suppression of violent, intrusive thoughts increases the frequency of those thoughts,
which may ultimately increase the risk of compliance with the violent intrusion.
In conclusion, suppression of anger may be caused by negative beliefs about
experiencing and expressing anger. Furthermore, the mechanisms involved in suppression of
anger are still not clear. Further investigation of the effects of attempts to suppress anger and the
potential moderators that are involved may prove to be important. In pursuing this, it may be
useful to use self-report measures of thought control strategies with content relevant to specific
disorders.
Rumination and suppression
In nonclinical as well as clinical samples, associations between rumination and
suppression have been found. For example, to explore the link between suppression and
rumination, Wenzlaff and Luxton (2003) conducted a longitudinal study in a student population.
Individuals high in suppression who had experienced high stress within a period of ten weeks
also had higher levels of rumination and dysphoria than individuals with low levels of
suppression or no preceding stress; this effect persisted even when controlling for initial levels of
rumination and dysphoria. The authors suggested a mechanism in which suppression is
undermined by the cognitive load induced by the stress. Because the suppression process itself
has proven to be cognitively demanding, overload may easily occur. When suppression is
disrupted, the monitoring process involved in suppression will expose negative thoughts with
high intensity. A person may attempt to control these negative thoughts using rumination as a
strategy even though it will increase negative mood and emotional symptoms (Wenzlaff &
Luxton, 2003).
In the State-Trait Anger Expression Inventory (STAXI-2;(Spielberger, 1999), the
expression of anger is conceptualized in three modes: anger-out, anger-in and anger-control.
Anger-out refers to a tendency to express anger through either verbal or physical behaviors.
Anger-in, or suppressed anger, refers to the tendency to hold one's anger on the inside without
any outlet. Anger-control refers to the tendency to engage in behaviors intended to reduce overt
anger expression.
Initially, anger rumination appears to resemble the construct of suppressed anger
(anger-in mode of anger expression). However, anger rumination may constitute what happens
after anger has been suppressed. Thus, the suppression of anger might provide the material for
Chapter 3 Psychopathology and information processing
52
subsequent rumination. In addition, although it may be difficult to tease cognition and emotion
apart, anger-in can be viewed as an emotional activity and anger rumination can be considered a
cognitive activity.
In conclusion, when suppression is successful it may actually relieve the individual
of unwanted thoughts; however, many things can potentially lead to the failure of this strategy. In
particular, stress may increase the risk of failure, and in this situation, the individual may switch
to rumination. It is hypothesized that if an individual holds both positive beliefs about the
functions of anger (e.g., that it serves as protection against threats and danger or facilitates goal
achievement) as well as negative beliefs about anger (e.g., that it is uncontrollable or related to
negative outcomes), the strategy for controlling anger may vacillate between rumination and
suppression.
In summary, there seems to be convincing evidence that attention allocation biases
can create interference in individuals prone to anger and aggression. Furthermore, state anger
exceeds the tendency to allocate attention towards threatening stimuli. Rumination and
suppression are strategies employed to control the experience of negative emotions, including
anger. Neither is unambiguously effective, and both seem to lead to increased bodily arousal and
prolonged negative affect resulting in an increased risk for anger-related responses. These
findings are consistent with the SIP model and network theories.
Moreover, the influence of metacognitive beliefs related to anger is suggested to
assist in the understanding of how anger is processed. Considering dysregulated positive beliefs
about anger may guide attention towards threatening stimuli and be involved in the selection of
rumination as a strategy to process anger. Negative beliefs about anger may follow the increased
arousal in anger and drive the selection of suppression as a strategy to process anger.
In the following section, the metacognitive model developed by Wells and
colleagues is presented as a clinical model of information processing that accounts for cognitive
processing as they relate to emotional disorders.
Chapter 3 Psychopathology and information processing
53
Metacognition
Metacognition is a term referring to “knowledge or processes involved in the
appraisal, monitoring, or control of cognition” (Harvey et al., 2004). As such, metacognition is a
fundamental characteristic of human cognition (Fernandez-Duque, Baird, and Posner, 2006;
Nelson et al., 1999; Wells, 2000; Flavell, 1979; Fernandez-Duque et al., 2000; Lories, Dardenne,
and Yzerbyt, 1998). It was initially introduced by Flavell (1979) from a learning perspective,
however, recent clinical research has explored the role that metacognition may play in
psychopathology. Evidence confirming a link between characteristics of metacognition and
psychopathology is currently emerging (Wells, 2000; Teasdale, 1999).
In the original conceptualization proposed by Flavell (1979), he argued that
cognitive monitoring and regulation are important features of communication, comprehension,
reading, writing, language acquisition, attention, memory, problem solving, social cognition, and
various forms of self-control and self-instruction. In this way, metacognition is decisive because
it controls the on-going monitoring and regulation of cognitive processes. Flavell provided an
overview of the features involved in cognitive regulation. In this overview, he noted that the
current cognitive goals and the concurrent cognitive strategies that are being executed interact
with stored knowledge and experience with cognitive goals and strategies; together these
components form a regulatory process with the purpose of influencing on-going cognition to
achieve a desired outcome. Thus, within this cognitive architecture, cognitive components are
differentiated at a meta level, controlling and regulating other cognitive activities, that comprises
the object level of the cognitive architecture (Martinez, 2006; Nelson et al., 1999; Wells, 2000;
Flavell, 1979; Fernandez-Duque et al., 2000; Lories et al., 1998). Therefore, a basic assumption
in a metacognitive framework is the principle that cognitive processes function on more than one
level and that these levels interact (Nelson et al., 1999). A metacognitive framework should
specify how this interaction occurs.
Contemporary clinical research has adopted this basic metacognitive idea, and the
framework by Wells and Matthews (Wells & Matthews, 1994; Wells, 2000) even considers this
process of regulation in itself to be the true psychopathology.
S-REF model
The metacognitive framework proposed by Wells and Matthews offers a unique and
generic conceptualization of the link between metacognitive components and psychopathology.
Chapter 3 Psychopathology and information processing
54
Their essential point is that regulation of cognitive activity is conducted as a result of the
individual´s cognitive goals and by the application of the different mental strategies that are
implemented to reach that goal (Wells, 2008).
The framework contains two aspects of metacognition, namely metacognitive
knowledge and metacognitive regulation. Metacognitive knowledge has two components; (1)
relatively implicit structures of self-relevant information about one's own cognition and strategies
for influencing cognition (e.g., knowledge about one´s ability to remember numbers or ability to
control one's negative emotions) that operate mostly outside of awareness; and (2) more explicit
beliefs about one's own thinking processes (metacognitive beliefs such as, “worrying makes me
ill”). In attempting to achieve the desired mental state (i.e., the cognitive goal), the metacognitive
belief contains a plan for processing. By directing attention, initiation, continuation and
termination of various cognitive activities, these plans are responsible for on-going cognitive
functioning. In this way, metacognitive regulation is informed and guided by metacognitive
knowledge stored in long-term memory, which in the proposal by Wells and colleagues was
emphasized mainly as a strategic rather than automatic processes.
More precisely, the Self-Regulatory Executive Functioning Model (S-REF model)
addresses the mechanisms responsible for self-regulatory processing bias (attentional bias and
dysfunctional coping strategies for further processing1). These strategies are applied to achieve a
desired mental state, which is related to metacognitive beliefs because these are the structures that
determine what the desired mental state is and how it should be reached. An example of this may
be, “I need to worry in order to function well”, which is a metacognitive belief that constitutes the
underlying drive for the selection of specific coping strategies aimed at dealing with lower-level
intrusive or negative thoughts. If, for instance, the individual is troubled by an intrusive worrying
thought, this metacognitive belief would initiate a worry about the worry. Due to the
metacognitive belief about the need to worry to control negative affect, the individual would be
likely to select worry as a coping strategy. Consequently, the individual would be even more
troubled due to excessive worrying and the implemented strategy would not be efficient because
1 To avoid confusion, it is necessary to comment briefly on the term coping strategy: Some writers consider a
coping strategy to be any response with the attempt to cope, while other reserve the term for responses aimed at a positive outcome (Nolen-Hoeksema et al., 2008). Related concepts are those of emotion regulation (Gross et al., 2006), emotional processing (Stanton, Kirk, Cameron, and Noff-Burg, 2000) or executive functioning processes (Fernandez-Duque et al., 2000). In the present thesis, because the discussion is limited to metacognition, the term coping strategy is defined as a mental strategy involving a plan for processing aimed at achieving a desired mental state (Wells & Matthews, 1994; Wells, 2000).
Chapter 3 Psychopathology and information processing
55
this style of thinking limits restructuring and locks thinking processes in a manner that maintains,
rather than alleviates, an emotionally negative inner state. This line of processing does not reach
its goal although the processing attempt may intensify the risk of initiating Cognitive Attentional
Syndrome (CAS). This syndrome refers to a response to an internal trigger that “…consists of
persistent thinking in the form of worry and rumination, focusing attention on sources of threat,
and coping behaviours that back-fire…” (Fisher & Wells, 2009). A characteristic of the CAS is
that attention, which under normal circumstances can flexibly shift from an internal to external
focus and be directed towards an adaptive and attainable goal, is locked on self-referent, self-
conscious and predominately threat-related stimuli.
Underlying the CAS is a range of specific positive and negative metacognitive
beliefs that maintain the CAS by supporting unhelpful thinking styles. When the CAS is not
terminated, it interferes with the restructuring of cognition and shifts in attention that occur under
normal circumstances. With this continuation of ineffective coping strategies, stress adds up.
The S-REF model specifies threat detection, rumination and worry as mental
control strategies (Wells & Matthews, 1994; Wells, 2000), and in more recent work, suppression
has also been included as an unhelpful thinking style (Wells, 2008; Fisher & Wells, 2009). While
threat detection, worry and rumination are unhelpful because they increase awareness of negative
stimuli, suppression is unhelpful largely because inevitably it collapses at some point. As
discussed earlier, suppression failure may expose negative thoughts as well as initiate rumination.
Figure 1 displays this formulation at its simplest. An internal trigger activates the control of
cognitive processes (metacognition), which interact and guide metacognitive beliefs. As a result
of dysfunctional metacognitive beliefs influencing controlling level, the CAS may activate the
emotional consequences of maintaining and exaggerating negatively valences items.
Figure 1. A, (antecedent) M, (metacognitive beliefs), and C, (consequence) analysis. Source (Fisher & Wells, 2009)
M: Metacognitive beliefs
A: Trigger (internal) Metacognition and (CAS) C: Consequences (emotional)
Although the S-REF is a generic model, the specific content of the metacognitive
beliefs and use of specific maladaptive thinking styles may vary across different disorders.
Chapter 3 Psychopathology and information processing
56
To illustrate, a patient with generalized anxiety disorder (GAD) that holds both
positive and negative metacognitive beliefs about worry will use worry as the predominant
strategy for dealing with threat yet experience worry as uncontrollable and dangerous; in turn,
stress and negative affect will increase. This further primes the use of worry as a strategy for
controlling distressing emotions. Through this processing routine, emotional distress is
maintained, and the beliefs driving the dysfunctional processing are strengthened.
Wells (2005) argued that particular negative beliefs about worry play a central role
in etiology and maintenance of GAD. However, to demonstrate the variations in how
metacognition influences psychopathology, consider the following scenario: when a depressed
patient holds positive metacognitive beliefs about rumination, the likelihood of recurrent negative
thinking in a situation of lowered mood is enhanced. The ruminative process maintains the
depressed mood and strengthens negative metacognitive beliefs regarding the uncontrollability
and harmfulness of rumination. In this model, positive beliefs about rumination as a coping
strategy motivate people to ruminate in situations in which they feel stress and negative mood.
However, the negative byproducts of this processing style will activate negative beliefs about the
uncontrollability and consequences of rumination. An increase in stress will reactivate this
vicious cycle (Mathews & Wells, 2004).
The initial aim of developing the Metacognition and Anger Questionnaire (MAQ)
was to apply the metacognitive framework, as it is modeled in the S-REF conceptualization of
metacognition, on anger. Contrary to a focus on assessing metacognitive components of worry in
the S-REF framework, the focus in the MAQ is on anger. Therefore, to construct new items the
content needed to be modified to reflect anger instead of worry. In the S-REF model, worry is
viewed as a strategy for processing negative stimuli, whereas in the metacognitive framework of
anger, anger is viewed as a strategy for processing negative stimuli. Whereas the S-REF
framework attempts to illuminate metacognitive beliefs that are involved in the selection of worry
as a coping strategy, the MAQ attempts to illuminate metacognitive beliefs that are involved in
the selection of anger as a coping strategy.
The CAS, which is a maladaptive on-line processing syndrome that causes
maintenance of emotional distress, is a non-specific syndrome. As previously described, a
characteristic of this syndrome is that when the goal that triggered the processing is not reached
(for example, to feel safe in a situation of perceived threat), the processing attempt intensifies
rather than decreases. In this way, restructuring is limited and thinking processes are locked.
Chapter 3 Psychopathology and information processing
57
According to this conceptualization, in a situation of perceived threat an individual
may be overly primed to detect threat due to prior experiences, triggering an anger response. It is
hypothesized that positive beliefs about anger may lead to the selection of anger rumination as a
response to anger arousal, with goals that potentially involve feeling safe, solving problems or
achieving other goals that are likely to result from angry rumination. However, even though the
processing goal is not reached, the distorted attentional pattern is not extinguished. This process
constitutes the CAS.
Chapter 4 Assessment
58
Chapter 4 Assessment
59
Chapter 4 Assessment
Assessment of metacognition
In the following section, the tools for assessing metacognition that guided the
development of the measure in this thesis are presented. First, the MetaCognitive Questionnaire
(MCQ), which is the tool most consistently used to operationalize the S-REF model, is presented.
Next the White Bear Suppression Inventory (WBSI) is presented, which is used to model the
suppression subscale used in Study 4. Finally, the negative and positive beliefs about rumination
scales (NBRS and PBRS), which incorporate particular metacognitive beliefs and ruminations as
a strategy for processing information in depression, are presented. To begin, a few comments on
the challenging task of assessing metacognition are appropriate.
Assessing metacognition is a demanding task because it is a complex construct.
Metacognition consists of both belief structures (metacognitive beliefs) and processing routines
(mental control strategies/copings strategies). When meta-cognitive assessment is conducted
using self-report, it requires insight. Some individuals, especially seriously impaired psychiatric
patients, may not be capable of providing an accurate assessment. Moreover, as discussed in
chapter 3 regarding strategic versus automatized information processing, the extent to which
people actually use “strategies” in the sense of willed, goal-directed cognitive activities intended
to reach a desired mental state is debated. It is also important to take into account that different
thought control strategies may be applied more or less successfully, and certain strategies may
work well for some but not for others. In addition, it is reasonable to assume that some strategies
work well in some situations, but the same strategy may work badly in another situation.
Together, all of these ideas imply that 1) the type of control strategy, 2) frequency of its use, and
3) the quality of the application of the particular strategy may all prove to be important. In this
line of thinking, both too little as well as too much use of a specific control strategy may be
potentially dysfunctional.
The MetaCognitive Questionnaire (MCQ;(Cartwright-Hatton & Wells, 1997): The MCQ is a
measure used to assess general aspects of metacognition. The MCQ is presently the most
consistently used tool for operationalizing the S-REF model. Wells and colleagues initially
developed the MCQ, and it was later revised into a shortened version, the MetaCognitive
Questionnaire (MCQ-30; (Wells & Cartwright-Hatton, 2004). The questionnaire was developed
Chapter 4 Assessment
60
for anxiety, however, it is argued that it also measures general components of metacognition. The
general focus of the questionnaire is worry and more specifically experiences and beliefs related
to worry.
The original 57-item questionnaire uses a scale from 1 to 4 in which 1 = do not agree, 2 = agree
slightly, 3 = agree moderately, and 4 = agree very much. The questionnaire assesses general
metacognition on the following 5 subscales:
1) Positive beliefs about worry (e.g., "Worrying helps me to solve problems")
2) Beliefs about uncontrollability and danger related to worry (e.g., "My worrying could
make me go mad")
3) Experiences/evaluations of one's own cognitive function (e.g., "I do not trust my memory")
4) Negative beliefs about mental control, including themes about superstition, punishment
and responsibility (e.g.," I will be punished for not controlling certain thoughts")
5) Experiences/evaluations of one's own awareness of cognition (e.g., "I am constantly aware
of my thinking")
Some alterations were made from the MCQ to the MCQ-30, the latter of which consists of 30
items that the participant is asked to rate using the same scale as the MCQ.
In the MCQ-30 (Wells & Cartwright-Hatton, 2004), factor analyses reproduced the
original 5 subscales although they emerged in an alternate order as outlined below:
1) Experiences/evaluations of one's own cognitive function
2) Positive beliefs about worry
3) Experiences/evaluations of one's own awareness of cognition
4) Beliefs and experiences about danger and uncontrollability
5) Beliefs about the need to control one's own cognition
The psychometric properties of the MCQ-30 were addressed satisfactorily,
including validation with other measures of worry and anxiety (Wells & Cartwright-Hatton,
2004).
The MCQ and MCQ-30 are suited to measure metacognitive beliefs and the
tendency to monitor cognitive events. These scales do not reflect information about the use of
other cognitive processes, such as rumination, worry or thought suppression, but instead attempt
to specify the dysfunctional metacognitive beliefs underlying dysfunctional cognitive processes,
aside from the process of monitoring one's own cognition.
Chapter 4 Assessment
61
The Danish version of the scale was permitted for use by Danish translators who had conducted a
formal translation with permission from the original author.
Construction of the items for the questionnaire applying metacognition to anger was modeled on
the MCQ (Cartwright-Hatton & Wells, 1997).
The White Bear Suppression Inventory (WBSI;(Wegner & Zanakos, 1994): This inventory was
developed to assess the tendency to suppress thoughts. It consists of 15 items rated on a five-
point scale from “strongly disagree” to “strongly agree”. Psychometric properties have been
addressed satisfactorily and significant correlations with depression, obsession, and anxiety have
been reported (Wegner & Zanakos, 1994). Later, the factor structure was revisited; in a student
sample, Hoeping and de Jong-Meyer (2003) found correlations with depression, anxiety and
obsession similar to those reported by Wegner and Zanakos (1994). Regarding factor structure,
they found that a two-factor structure comprising `unwanted intrusive thoughts´ as the first factor
and `thought suppression´ as the second factor was indicated. They argued for the need to
differentiate between unwanted intrusive thoughts and thought suppression, and they also advised
leaving out unwanted intrusive thoughts when investigating the possible link between thought
suppression and psychopathology. Rassin (2003) argued that the WBSI measures failed
suppression instead of suppression per se, and he conducted three studies to confirm this point.
The first study, which was conducted in a non-clinical population, replicated the factor structure
and correlations found in the study by Hoeping and de Jong-Meyer (2003). The second study
repeated the factor structure in a clinical sample although some of the items were loaded on
different factors. In a third study, also in a non-clinical sample, they used a measure that
differentiated between successful and unsuccessful suppression. Rassin (2003) found that
successful suppression was negatively correlated with dysfunctional thought control strategies,
while suppression attempts were positively correlated with dysfunctional thought control
strategies. In a non-clinical sample, Luciano, Belloch, Algarabel, Tomas, Morillo, and Lucero
(2006) tested several of the previously proposed models on the WBSI and concluded that the
WBSI has an unclear factor structure.
In summary, the items on the WBSI scale reflect intrusive thoughts as well as
attempts to suppress unwanted thoughts. The suppression subscale developed for Study 4 was
modeled based on the suppression items of this scale (see measures).
Chapter 4 Assessment
62
The Negative Beliefs about depressive Rumination Scale (NBRS;(Papageorgiou, Wells, and
Meina, 2001) and Positive Beliefs about depressive Rumination Scale (PBRS;(Papageorgiou &
Wells, 2001a): These scales were developed to enhance knowledge about the mechanisms
involved in ruminative processes. The questionnaires focus on illuminating beliefs held by people
about the nature and function of rumination. The NBRS consists of 13 items reflecting two types
of contents; 1) themes concerning uncontrollability and harm (e.g., `ruminating means I´m out of
control´), and 2) interpersonal and social consequences (e.g., `people will reject me if I
ruminate´). The PBRS consists of 9 items that reflect themes of rumination as a coping strategy
(e.g., `ruminating about my problems helps me to focus on the most important things´).
Responses to each item were made on a four-point scale ranging from `do not agree´ to `agree
very much´. Both positive and negative beliefs were significantly correlated with rumination and
depression in non-clinical and clinical samples (Papageorgiou & Wells, 2001a; Papageorgiou &
Wells, 2004; Papageorgiou & Wells, 2001b). This questionnaire reflects a specified model for
rumination in depression. In Study 4, the applicability of this rumination model to anger is tested.
The Metacognitive and Anger Questionnaire (MAQ) and the Metacognitive beliefs and Anger
Processing (MAP) are categorized as anger measures and are described below.
Assessment of anger
How does one assess anger?
One perspective on this question, emphasizing the subjective experience of anger, is that “If we
want to know how people feel, what they experience and what they remember, what their
emotions and motivations are like, and their reasons for acting as they do – why not ask them?”
(Allport, 1942, p. 37).
Allport further argues that the use of personal information encourages the understanding of the
patient as a human being, which is appropriate because mental illness and its symptoms are a
personal issue.
However, due to social constructs surrounding anger, such as those discussed in
chapters 1 and 2, accurate self-report faces some serious challenges. These challenges are evident
in the following quotation:
Chapter 4 Assessment
63
“What does it mean to be `angry´? – it suggests being `mad´ and out of control. What does it
mean to be labelled an `angry person´? – it implies that you are a `bad´ person. What are the
consequences of an `angry person´ labelling? – you get extra constraint. What is the residual
significance of anger? – it is indicative of continued `psychopathology´” (Novaco, 2010b).
As indicated by the first quote, there are obvious reasons for using self-report to
assess anger, namely that anger is a subjectively experienced inner state of arousal. Furthermore,
self-report is an easy and quick way to collect information, and it facilitates self-monitoring and
self-awareness, which are clinically valuable.
However, a relevant objection to assessing anger solely on the basis of self-report is
that anger is a multidimensional construct requiring a comprehensive battery of assessments to be
captured completely and accurately.
In addition, the validity of using self-report measures can be compromised by a
variety of issues (Eckhardt et al., 2004). One source of compromised accuracy in relation to anger
is that anger arousal interferes with information processing, thus decreasing self-monitoring
(Novaco, 2000). Also, the presence of a mental disorder or intellectual disability may decrease
the ability to self-monitor. In principal, all circumstances that interfere with the quality of
executive functioning may cause a decrease in the ability to self-monitor.
Another source of bias is that people with long-lasting anger problems tend to be so
closely associated with and protective about their anger that it sometimes prevents them from
monitoring and reporting their anger. Difficulties in the ability to report anger may thus stem
from anger as an embedded part of one's self-identity, self-protective system and sense of self-
worth. When anger is entangled with other troublesome emotions, reporting anger may involve
the activation of other associated, distressing emotions and experiences (e.g., trauma and abuse),
which is demanding on the individual and may contribute to the complexity of emotional reports.
Furthermore, subjects may fail to report their “true” feelings due to a desire to
present themselves in a socially favorable way (Averill, 1982). Because anger is burdened with a
long tradition of negative evaluation and viewed as a form of temporary insanity or madness
(Potegal & Novaco, 2010), limited interest in revealing anger experiences may be a large obstacle
to self-report of anger. Thus, the anticipated consequences of revealing anger experiences
influence the validity of self-reports. In hospital settings, respondents may anticipate undesirable
consequences of revealing information (e.g., a loss of privileges) or negative views or evaluations
Chapter 4 Assessment
64
by staff members if they disclose anger experiences. Individuals under long-term care with long-
lasting anger problems tend to be distrustful and suspicious, and as a result, they may be inclined
to reveal as little as possible about themselves and their anger experiences. Regarding offenders,
concern with the validity of their self-reports has also been noted (Simourd & Mamuza, 2000;
Seager, 2005).
Alternatives to self-report of anger include physiological measures such as heart
rate or blood pressure. In terms of aggression, reports from observers and reports of criminal
conduct may be useful. For research conducted in clinical settings, including what is presented in
this thesis, anger is measured by self-report. However, in forensic studies aggression as a
behavioral response is not measured by self-report but observed and evaluated by clinical staff.
Below, anger and aggression measures are presented together with the other measures that were
used in the present studies.
The Provocation Inventory (PI;(Novaco, 2003): The PI is a 25-item self-report
instrument measuring anger intensity in specific types of provocative situations. The instrument
describes situations that could potentially elicit anger, and the respondent rates anger intensity on
a 4-point scale that covers the following content areas: disrespectful provocations, unfairness,
frustration, annoying traits of others and irritations. Higher scores indicate greater anger. In the
standardization of the NAS-PI (N = 1546), the PI Total alpha score was .95 (Novaco, 2003); in a
civil psychiatric sample (N = 1101), the PI Total .92 (Monahan et al., 2001) and among
developmentally disabled forensic patients (Taylor & Novaco, 2004) the PI Total .92. Stability
and validity has been investigated in a variety of different samples an using alternate anger
measures (Lindqvist, Dademan, Hellstrom, 2005; Baker, van Hasselt, and Sellers, 2008; Jones et
al., 1999; Novaco, 2003). Several translations of the tool have been successfully made, including
a translation into Swedish (Lindqvist, Dademan, and Hellstrom, 2003).
Novaco Anger Scale (NAS;(Novaco, 2003): The NAS is a 60-item scale constructed
to measure anger as guided by Novaco (Novaco, 1994). The scale measures anger in cognitive,
arousal and behavioral domains that together form the NAS Total score; there is also a separate
anger regulation subscale. The participant is asked to rate each item on a scale in which 1 = never
true; 2 = sometimes true; and 3 = always true. Because there is no direct relationship between
external events and anger arousal, anger arousal is a function of cognitive perception and
processing with the inclination towards a behavioral response as is implied by the definition of
Chapter 4 Assessment
65
anger. The NAS Cognitive subscale includes items reflecting the following categories:
justification, suspicion, rumination, and hostile attitude. Cognitive representations influence the
experience and expression of anger by guiding information processing as discussed in chapter 3
of the introduction. Central to anger is the physiological component of heightened bodily tension,
which intensifies the experience of anger. The NAS Arousal subscale reflects dimensions of
intensity, duration, somatic tension, and irritability. The inherent inclination to behave
aggressively when angry is captured in the NAS Behavioral component, consisting of items that
operationalize impulsive reactions, verbal aggression, physical confrontation, and indirect
expression. The NAS Total is the summed values of the Cognitive, Arousal and Behavioral
subscale. The NAS also includes a regulation scale targeting the capacity to regulate anger-
engendering thoughts, to self-calm, and to engage in constructive behavior when faced with
provocation. Alpha scores and test-retest reliability across various settings have shown excellent
reliability (Monahan et al., 2001; Taylor & Novaco, 2004). As shown by independent
investigators, the NAS is a strong anger assessment instrument with a clear theoretical
conceptualization and solid psychometric properties across various settings (Lindqvist et al.,
2005; Baker et al., 2008; Jones et al., 1999); in addition, the NAS predicts future violence
(McNiel et al., 2003; Monahan et al., 2001).
Stait Trait Anger eXpression Inventory (STAXI-2;(Spielberger, 1999): The STAXI-
2 is a 57-item scale constructed to measure a broad range of anger experiences and control. It has
been revised and adjusted over the last 10 years. Anger is assessed based on the state-trait
personality theory. As such, anger is conceived of as a joint combination of individual differences
in dispositional anger (trait anger) and the momentary experience of anger (state anger). The
scale consists of 6 subscales measuring anger trait, anger state and the anger components of
expression and control. The anger expression subscale measures tendencies from both ends of the
spectrum, from outward expression of anger (AX-O) to suppression of anger (AX-I). The anger-
control subscale measures attempts to control anger. Anger control –in (AC-I) measures the
tendency to invest energy in calming down and securing inner control, overriding the experience.
Anger control-out (AC-O) measures the tendency to invest energy in monitoring and preventing
the outward expression of anger. The STAXI-2 is generally considered a strong anger assessment
instrument with a clear theoretical conceptualization and solid psychometric properties in varied
settings. In a STAXI-2 (Spielberger, 1999) study that included data from 1600 normal adults and
274 hospitalized psychiatric patients, the reliability scores were as follows: T-Ang ranged from
Chapter 4 Assessment
66
.84 to .87; S-Ang ranged from .92 to .94; AX-O ranged from .74 to .80: AX-I ranged from .74 to
.82; AC-O ranged from .84 to .87; and AC-I ranged from .91 to .93. Regarding validity, the
STAXI differed between healthy and clinical participants.
The NAS and the STAXI scales are intended to measure similar constructs.
Previous studies have found meaningful and strong correlations between the STAXI and NAS
subscales in different settings (Novaco, 1994; Novaco & Renwick, 2002; Taylor & Novaco,
2004; Lindqvist et al., 2005; Lindqvist et al., 2003), validating both measures.
The NAS-PI and the STAXI-2 were translated into Danish with written permission
from the original author. The questionnaires were translated by the author of this thesis and then
back-translated by a bilingual translator. The rewording of a few items was conducted during this
process. The back-translation of the NAS-PI was reviewed by the original author. See appendix
F, norm study of the NAS-PI for details.
The Anger Rumination Scale (ARS;(Sukhodolsky et al., 2001): This scale was
constructed to measure the tendency to think about anger. It contains 19 items on four factors,
including Angry-Afterthoughts (6 items); Thoughts of Revenge (4 items); Angry Memories (5
items); and Understanding the Causes (4 items). Participants are asked to rate each item on a
scale from 1 = almost never, to 4 = almost always. The scale was tested in a student sample, and
factor structure, reliability and validity issues were addressed satisfactorily. The questionnaire
was translated by the author of this thesis with permission from the original author and back-
translated by a bilingual translator.
The Schedule of Imagined Violence (SIV;(Grisso et al., 2000): This scale was used
to guide how to measure violent thoughts. The SIV consists of 8 questions; the first question
assesses the presence of violent thoughts either at present or previously, and the following 7
questions are only given to participants who answered the first question affirmatively. The
content of these successive questions relates to recency, frequency, chronicity, type of harm,
target focus, seriousness of harm, and proximity to target. For the present study, two questions
were used; the first question of the SIV was about whether the participants had ever experienced
violent thoughts/fantasies, and if confirmed, the second question about recency and frequency
(`when this has happened and how often it happens´) was asked. In the original measure, people
were assigned either SIV+ or SIV-. Participants were deemed to be SIV+ if they confirmed ever
Chapter 4 Assessment
67
having violent thoughts within the past 2 months. The same categorical criteria were used in the
present study. A formal translation procedure was not used.
The Metacognitive and Anger Questionnaire (MAQ) version 1 (MAQ-1, see
appendix A): This questionnaire consists of 57 items measuring metacognition in relation to
anger. The respondent is asked to indicate for each statement whether it is (1) never true; (2)
sometimes true; (3) often true; or (4) always true. The questionnaire consists of 4 domains:
Positive beliefs about anger (e.g., `anger helps me cope with things´), negative beliefs about
anger (e.g., `my anger harms myself´), angry rumination (e.g., `I cannot let go of angry thoughts´)
and cognitive consciousness (e.g., `I am aware of my thoughts´). After pilot tests with 192 police
students, the instrument was reduced to 45 items (MAQ-2, see appendix B) and pilot-tested with
167 prisoners, leading to another process of item selection that resulted in the MAQ-3 with 34
items (see appendix C). The questionnaire was developed in English and in Danish. Details of its
construction are available in Study 1.
The Metacognitive beliefs and Anger Processing (MAP, see appendix D): This
scale includes the same items as the MAQ-3, except that the cognitive consciousness subscale has
been omitted and a subscale designed to measure suppression is developed. The subscale was
modeled on the framework of the White Bear Suppression Inventory (Wegner & Zanakos,
1994m). The questionnaire was developed in English and in Danish. Details of the construction
are available in Study 4.
In part two of this thesis, the empirical development of this scale through 4
different studies is described.
PART 2 Overview of methodology
68
PART 2 Overview of methodology
69
PART 2 EMPIRICAL STUDIES
Overview of Methodology
In the study of psychopathology, characteristics of different samples tend to mediate
the relationship between certain variables (Aldao et al., 2010).Therefore, a multi-sample
approach to scale development was chosen. Indeed, sample type was found to moderate
relationships between the variables of interest. Because scale development requires item
adjustment and retesting, the studies for developing the MAQ were conducted in sequence. Study
5, however, which did not include the MAQ, was conducted concurrent with the other studies.
With the exception of Study 4, the design of the studies was cross-sectional and the measures
used were predominately self-report questionnaires. Study four focused on prediction, adopting a
longitudinal design and using observational data as well. The particular measures used in each
study are noted in the description of the individual study.
Overview of the thesis studies
With the purpose of developing a metacognitive measure related to anger, the
following studies were conducted:
Study 1: The pilot conducted before Study 1, was conducted at the Mental Health
Centre Sct. Hans and included 12 volunteer forensic inpatients. The purpose of the study was to
explore the clinical utility of anger in a forensic setting using the metacognitive framework
proposed by Wells and colleagues. The pilot resulted in the construction of the MAQ-1, which
was then tested in a sample of 192 police students. During this process, a number of the items
were deleted because on reflection, they were judged to be ambiguous or inaccurate in capturing
the intended concept. New items were added and some items were reworded in an attempt to
increase the clarity of the items; this resulted in the MAQ-2.
Study 2: The MAQ-2 was tested in 5 different prisons in Denmark. A total of 167
male prisoners participated. During the testing process, items were refined and adjusted resulting
in the MAQ-3.
Study 3: The MAQ-3 was tested in a mixed clinical setting, involving both
inpatients and outpatients, and with patients representing various diagnoses. Participants
completed an assessment package with a total of 221 questions assessing anger and
metacognition. A total of 88 patients were included. This test led to scale refinements and
PART 2 Overview of methodology
70
renaming the MAQ. One subscale was omitted and a new one was included. The resulting scale
was named the Metacognitive beliefs and Anger Processing (MAP) scale.
Study 4: The MAP was tested using a longitudinal design to further test its validity
and to evaluate the predictive power of the MAP. The sample consisted of male forensic
inpatients from the Mental Health Centre Sct. Hans. A total of 54 patients were recruited.
Study 5: Several datasets, which were gathered on different occasions, generated the
Novaco Anger Scale and Provocation Inventory (NAS-PI;(Novaco, 2003) of normative data. A
total of 454 non-clinical individuals and 87 clinical patients completed both the NAS and the PI.
In addition, 192 police students and 167 prisoners completed only the PI. Lastly, 77 clinical
patients and 64 forensic patients completed only the NAS. A total of 1064 individuals
participated; this study is presented briefly in appendix F.
Chapter 1 Development of the MAQ in a non-clinical setting
71
Chapter 1 Development of the MAQ in a non-clinical setting
Introduction
The purpose of the present pilot study was to explore the utility of a metacognitive
framework on anger. The following three research questions were formulated:
Do individuals hold both positive and negative beliefs about the functions and nature of
anger?
Are particular beliefs about anger connected to the selection of particular strategies to
process negative stimuli?
Are there indications that individuals get stuck in a self-perpetuating circle of processing
negative stimuli in a manner similar to that proposed in the CAS?
Based on a qualitative pilot study in which semi-structured interviews were
conducted with 12 forensic patients (Wells & Matthews, 1994; Wells, 2000), initially forensic
patients seemed to hold both positive and negative beliefs about anger and they reported
experiences of getting stuck in ruminative processes (Appendix G displays the semi-structured
interview guide). Therefore, the MAQ-1 was modeled on the MCQ and pilot tested in a non-
clinical sample to explore factor structure and reliability.
Participants
A convenience sample of 192 police students was recruited during a teaching
lesson that was part of their law enforcement education. The sample was a non-clinical sample
given that the students were part of the general population, but they were also a sample of
specific relevance for the construct of interest. The participants were assured that participation in
the study was voluntary and anonymous. All available participants volunteered. The average age
was 28 (range 19-35, SD = 2.6); 44 (23 %) of the participants were male and 148 (77 %) were
female. There were no significant gender differences in PI (t (190) = 1.73, p = .09) or for the
MAQ (t (190) = 1.86, p = .07). Thus, the data for both genders were pooled.
Measures
The Provocation Inventory (PI) (Novaco, 2003) is a 25-item, self-report instrument
measuring anger intensity in specific types of provocative situations (see anger assessment).
Chapter 1 Development of the MAQ in a non-clinical setting
72
The Metacognition and Anger Questionnaire (MAQ). In performing interviews with
the forensic patients, features concerning cognitive ability were presumed to be less relevant for a
metacognitive framework aimed at understanding cognitive processing as it relates to anger than
for other cognitive frameworks. Regarding worry, it makes sense that people may worry more if
they do not believe in their abilities to remember things. However, in the interviews with the
forensic patients they expressed confusion about the relevance of confidence to their own
cognition. Thus, in modeling the MAQ on the MCQ (Cartwright-Hatton & Wells, 1997), the
cognitive confidence subscale was not reproduced. The cognitive self-conscious subscale, the
need to control thoughts subscale, the uncontrollability and danger subscale, and the general
negative beliefs subscale were all considered potentially relevant to anger. Items reflecting these
subscales with specific anger-related content were formulated. The following content was
deemed important in relation to anger:
o Uncontrollability of the experience of anger and anger-related thoughts
o Negative conceptions related to danger, particularly those focused on harm and
madness in association with anger
o General negative beliefs about the consequences of anger
o General positive beliefs about the functions of anger
o General evaluations of one's own cognitive awareness and abilities
The MAQ-1 was designed solely to assess an anger construct, and therefore I attempted to avoid
interference from the aggression construct using careful wording for the individual items.
Procedure
The PI and the MAQ-1 were administered in classroom groups of 14-18
individuals. A brief introduction about the background and purpose of the study was offered, after
which the participants individually completed the questionnaires. Respondents were then asked
for their comments about the questionnaires. To evaluate test-retest reliability, three groups (39
participants) were retested after 3-weeks.
Chapter 1 Development of the MAQ in a non-clinical setting
73
Results
Provocation Inventory (PI)
The mean score of the PI Total was 51.1 (SD = 8.6). Compared to Swedish male
undergraduates with a mean of 55.4 (Lindqvist et al., 2003), the mean PI Total score for this
sample was significantly lower (t (191) = 6.87, p < .000). Compared to non-clinical participants
in Study 5, N = 477, M = 53.5, SD = 10.3) (see appendix F), the present sample of police students
also had a significantly lower mean PI (t (191) = 3.82, p < .000). Skewness and kurtosis were
examined and found absent. The alpha for the 25-item PI was .87. The test-retest (Pearson)
correlation was .75. The PI score was not significantly correlated with age.
Metacognition and Anger Questionnaire (MAQ-1)
The MAQ-1 data met assumptions of normality, permitting a factor analysis of the
scale. The MAQ-1 was not significantly correlated with age. The primary goal of the initial factor
analyses was to reduce a large number of variables into a smaller number of components,
therefore a principal components analyses (PCA) was conducted (Tabachnick & Fidell, 2007).
An oblique rotation using the Promax technique was chosen because the underlying factors were
believed to be correlated. In the Promax technique, orthogonal factors are rotated to oblique
positions to allow correlations among factors (Tabachnick & Fidell, 2007).
A PCA with a Promax rotation was thus conducted on the initial 57 MAQ-1 items.
Based on theoretical relevance, a four-factor solution was considered optimal for this dataset. The
solution accounted for 33.3 % of the variance with 34 items loading above .48 and only on one
factor. Item 50 loaded at .48, but it was omitted due to redundancy with item 6.
In summary, out of the first pool of 57 items, 34 items on four factors remained.
The first factor was Positive Beliefs about anger (9 items, alpha = .85), the second factor was
Negative Beliefs about anger (14 items, alpha = .84), the third factor was Rumination (7 items
alpha = .79) and the fourth factor was Cognitive Consciousness (4 items alpha = .61). Table 1
displays the results of the factor analyses.
Chapter 1 Development of the MAQ in a non-clinical setting
74
Table 1. Factor loadings from PCA with Promax rotation for police students, N = 192
Police students (N = 192) MAQ-1
1.PB 2.NB 3.Rum 4.CC
3. When I am angry I keep thinking about it -.13 -.03 .53 .21
4. I cannot distance myself from angry thoughts -.16 -.02 .65 .22
6. I am constantly aware of my thinking -.03 -.02 .24 .65
7. I must be aware of unjust actions against me .20 -.06 .15 .54
8. I cannot let go of angry thoughts -.04 .00 .71 .18
9. Anger is hard to control; it controls you .04 .17 .58 -.09
11. My anger harms me .04 .50 .17 .04
12.Anger helps me see things the way the really are .58 -.12 -.03 .00
14. It is bad to have angry thoughts -.26 .61 -.07 .09
15. When I start getting angry I cannot stop .04 .07 .68 -.04
16. Anger is bad for me -.36 .52 .12 .22
17. I can easily understand other people´s emotional reactions -.08 -.21 .10 .51
19. Anger helps me solve problems .70 -.14 .02 -.05
21. I must control my thoughts .27 .21 -.06 .51
22. Anger helps me handle things .78 -.08 .03 -.03
23. Anger could make me go mad .18 .48 .27 -.12
25. I cannot ignore my anger .27 .06 .49 -.00
26. Anger keeps me safe .68 .12 .00 -.03
27. Anger will make other people reject you -.04 .57 -.12 -.01
30. My anger can harm other people .18 .66 -.09 -.17
31. I do not think clearly when I am angry .02 .49 .22 -.10
32. Being angry will make me lose control and go mad .06 .58 .07 -.10
33. Anger is good for me .71 -.32 -.08 .04
35. My anger is dangerous for me -.06 .52 .12 .06
37. I cannot distract myself from anger .04 -.03 .68 -.06
41. Anger means loss of control -.16 .51 .07 .13
42. When I am angry I lose sight of different points of view .06 .52 .16 .02
43. Anger protects me from being exploited by others .53 .19 .02 .02
45. Anger makes me a strong and capable person .68 -.14 .11 .02
48. Anger makes me a bad person -.10 .64 -.09 -.24
49. Others will be judgmental of you for getting angry .16 .58 -.18 -.12
54. Anger is necessary to get by in the world .65 .11 -.17 -.12
55. Anger makes me insensitive to other people .05 .49 .11 -.06
57. Anger keeps me alert .57 .09 -.03 .05
Note. Bold typing highlights the highest loading on the subscale for the item.
Internal reliability analysis yielded an alpha of .85 for the remaining 34 items of the MAQ-1 after
the first item selection. Test-retest reliability (Pearson) was .78, indicating very good stability for
the new measure.
Chapter 1 Development of the MAQ in a non-clinical setting
75
Subscale correlations
Examination of the MAQ-1 showed significant positive intercorrelations between
the subscales, aside from a zero-order correlation between Negative Beliefs and Positive Beliefs.
Rumination and Negative Beliefs were strongly correlated, r = .35. All MAQ-1 subscales were
highly correlated with the MAQ-1 Total.
Intercorrelations revealed a correlation between the MAQ-1 and the PI Total of r =
.60. Three of the four subscales comprising the MAQ-1 showed significant correlations with
anger level (PI). The Rumination subscale was most strongly correlated with anger (r = .37),
followed by Positive Beliefs (r = .27) and Negative Beliefs (r = .22), whereas the correlation for
the Cognitive Consciousness subscale was non-significant (r = .14). These results are presented
in Table 2.
Table 2. Correlations (Pearson) between MAQ-1 subscales and anger level (PI Total).
MAQ-1
Positive Beliefs Negative Beliefs Rumination Cognitive
Consciousness
PI_Total .60* .27
* .22
* .37
* ns
Positive beliefs .53* 1 ns .24
* .21
*
Negative beliefs .71* 1 .35
* .23
*
Rumination .62* 1 .27
*
Cognitive Consciousness .55* 1
Note. N = 192, * p < .01.
Lastly, a hierarchical regression was performed with PI Total as the criterion
variable and MAQ-1 subscales, excluding the cognitive consciousness subscale that showed a
non-significant correlation with PI, as the first and only block. The overall model accounted for a
significant amount of the variance in the criterion variable (PI) (R² = .180, F (3,191), = 13.78 (p <
.000). Positive Beliefs and Rumination were significantly associated with the PI Total at the
p<.01 level although the Negative Beliefs were non-significant (p = .10).
Discussion
The non-significant correlation between age and PI Total was not consistent with
the literature, which generally reports an association between younger age and anger level.
Chapter 1 Development of the MAQ in a non-clinical setting
76
However, because the age difference in the sample was relatively small, this was not surprising.
The finding that the mean total PI total score was significantly lower in this sample than in
Swedish male undergraduates or Danish non-clinical individuals indicates that the sample
reported here was unexpectedly low in anger. This may be explained by the social desirability for
police officers to report low levels of anger, a vision that police students may possess. Another
explanation is that police students may actually represent a sample type characterized by lower
anger disposition (PI Total score) that the average person. Likely, a combination of these two
factors is occurring.
Because the factor analyses demonstrated that the scale measured four distinct and
reliable categories of beliefs and processes in relation to anger, the MAQ-1 was deemed
promising as a clinical anger scale. Because three subscales of the MAQ-1 showed significant
correlations with the anger measure, the scale shows potential value for understanding the
cognitive mechanisms involved in individuals who present with anger-related problems. The
psychometric properties should be tested in greater detail, particularly focusing on convergent
validity with other populations reporting high anger levels.
In addition, the Rumination subscale and the Cognitive Consciousness subscale
consisted of too few items compared with the other subscales. Thus, new items should be
developed.
Chapter 2 Prisoners, anger, and the MAQ
77
Chapter 2 Prisoners, anger, and the MAQ
Introduction
The purpose of the present study was to further test the psychometric properties of the
MAQ-2 in a relevant population by choosing a sample with a higher level of anger. In addition,
general metacognitive measure was included to address convergent validity. Prior to study 2, the
MAQ-1 was revised based on study 1. The result of revising the MAQ-2 was the inclusion of 11
new items to form a scale that contained a total of 45 items (see MAQ-2 in appendix B). The new
items were constructed to load on the Rumination subscale (5 items) and the Cognitive
Consciousness subscale (6 items). In addition, minor changes in the wording of some of the items
were conducted prior to study 2.
Because they measure a metacognitive construct, the subscales of the MAQ-2 were
expected to show moderately positive correlations with the general metacognitive questionnaire
(MCQ-30). Consistent with the metacognitive approach to emotional disorders, the inter-subscale
correlations of the MAQ-2 were expected to be moderately positive, and all subscales of the MAQ-
2 were expected to be positively correlated with anger.
Participants
A sample of 167 male prisoners was recruited from 5 different prisons in Denmark (3
closed and 2 open). Participants gave their consent to participate. The study was approved by the
Danish Prison and Probation Service. The average age of the prisoners was 30.8 (range: 18-62, SD
= 9.7); the average length of scholarly education was 9.2 years (SD = 2.2). Sixty-one percent of the
participants had no education other than compulsory schooling and sixty-one percent of the
participants were serving a sentence for a violent crime. The average length of the sentence was 3.1
years (range 1-13 years, SD = 3.1). The sample was recruited from different types of institutions
representing the variability of Denmark’s prisons. Eight people (5%), did not speak Danish and
received the questionnaires in English. The remaining prisoners were tested using questionnaires in
Danish.
Measures
The MAQ-2, which was the revised version of the MAQ-1, included 11 new items,
resulting in a 45-item scale.
Chapter 2 Prisoners, anger, and the MAQ
78
The Provocation Inventory (PI) (Novaco, 2003) is a 25-item self-report instrument
measuring anger intensity in specific types of provocative situations (see anger assessment).
The MetaCognitive Questionnaire (MCQ-30) (Wells & Cartwright-Hatton, 2004)
measures general aspects of metacognition on 5 subscales: (1) Experiences/evaluations of one's own
cognitive function; (2) Positive beliefs about worry; (3) Experiences/evaluations of one's own
awareness of cognition; (4) Beliefs and experiences about danger and uncontrollability; (5) Beliefs
about the need to control one's own cognition (see metacognitive assessment).
Less than 5% of responses were missing, and no respondent was missing more than 3
items. The values for the missing items were replaced with the series mean for the item.
Violent offense was characterized as any offense including actual physical contact or
threats of violence. School length was coded as years of scholarly education, and sentence length
was coded as the total number of years of the prisoner's current sentence.
Procedure
The instruments were administered in random order. The questionnaires were read
aloud in small groups of 2-6 for some of the participants, and others individually filled out the
questionnaires. On a few occasions in the open prisons, the administration was conducted in groups
of 16-18. Participants received written information and were orally assured that they would remain
anonymous and that the study was independent of their involvement with the prison system. Groups
of participants were approached during educational or work activities by prison staff and the
researcher. Participants were served coffee and cake while completing questionnaires but did not
receive any additional rewards or benefits for participating. To evaluate test-retest reliability, 17
participants were retested after 1-3 weeks.
Results
Factor analyses
The coefficients of skewness and kurtosis for the PI, MCQ-30 and MAQ-2 were less
than 2 when divided by their standard errors, indicating the absence of skewness and kurtosis.
Examinations of the unstandardized residuals plotted against the unstandardized predicted residuals
were satisfactory. The data met assumptions of normality, permitting factor analysis of the scales.
An oblique rotation using the Promax technique was chosen because the underlying
factors were believed to be correlated. In the Promax technique, orthogonal factors are rotated to
oblique positions to allow correlations among factors (Tabachnick & Fidell, 2007).
Chapter 2 Prisoners, anger, and the MAQ
79
The 45 items comprising the MAQ-2 were thus entered in a PCA with Promax rotation, and four
factors were fixed. The solution accounted for 46.2 % of the variance after rotation. In this solution,
the Cognitive Consciousness subscale showed items loading on more than one factor; however, it
did not load on any factor greater than .4. When including items loading above .4 on the Cognitive
Consciousness subscale, 8 items with a reliability score of .82 were identified. However, the
external validity was unsatisfactory because its correlation with the PI Total was .08.
Due to the unstable factor loadings and unsatisfying external validity of the Cognitive
Consciousness subscale (MAQ CC), a principal components factor analysis using Promax rotation
without the 10 MAQ CC items was conducted. Three factors were fixed, resulting in a solution
accounting for 44.2 % of the variance. Items loading on the expected factor and greater than .43
were kept for the remaining analyses. Three items loaded on more than one factor, but were
included in additional analyses of the factor to which the item was initially assumed to load because
they seemed theoretically valuable. These items were item 30, `My anger can harm other people´;
item 42, `When I am angry, I lose sight of different points of view´; and item r2, `If I just let go of
my anger, people will not understand that they went too far´.
The first factor of the analysis was Rumination (10 items, alpha = .86), the second
factor was Negative Beliefs (11 items, alpha = .84), and the third factor was Positive Beliefs (8
items, alpha = .82). Twenty-nine items were thus included in the subsequent analysis. The results of
the factor analysis are displayed in Table 1.
Chapter 2 Prisoners, anger, and the MAQ
80
Table 1. Factor loadings of the PCA with Promax rotation for a sample of prisoners, N = 167.
Prisoners (N = 167) MAQ-2
1.Rum 2. NB 3. PB
3. When I am angry I keep thinking about it .56 .02 .07
4. I cannot distance myself from angry thoughts .56 -.04 -.02
8. I cannot let go of angry thoughts .85 -.20 -.08
9. Anger is hard to control; it controls you .56 .10 .19
11. My anger harms me -.06 .66 -.06
12.Anger helps me see things the way the really are .06 .09 .56
14. It is bad to have angry thoughts .09 .53 -.19
15. When I start getting angry I cannot stop .78 -.07 .09
16. Anger is bad for me .13 .62 -.19
19. Anger helps me solve problems -.09 -.11 .67
22. Anger helps me handle things -.01 .01 .55
23. Anger could make me go mad (.57) .26 -.05
25. I cannot ignore my anger .72 -.21 .11
26. Anger keeps me safe .07 .01 .71
27. Anger will make other people reject you -.22 .67 .09
30. My anger can harm other people (.28) .38 (.23)
32. Being angry will make me lose control and go mad .66 .13 .02
33. Anger is good for me -.31 -.03 .74
35. My anger is dangerous for me .29 .50 -.07
37. I cannot distract myself from anger .82 -.14 -.01
41. Anger means loss of control .20 .57 -.14
42. When I am angry I lose sight of different points of view (.44) .40 -.14
45. Anger makes me a strong and capable person .13 -.08 .68
48. Anger makes me a bad person -.27 .79 .02
49. Others will be judgmental of you for getting angry .05 .72 .07
54. Anger is necessary to get by in the world .20 .01 .61
55. Anger makes me insensitive to other people .17 (.33) .06
57. Anger keeps me alert .21 -.07 .54
R2. If I just let go of my anger, people will not understand that they
went too far
(.44) .06 (.41)
R3. It is impossible not to think about anger .48 .11 .09
R4. When I am angry, I can only think about that .77 -.06 -.03
R5. Thinking about anger will produce solutions -.30 .57 .38
Note. Brackets indicate deviations from the expected loadings.
In the revision and item selection for the MAQ-2, there were several deviations from
strict adherence to factor loadings as the inclusion criterion. Of the 32 items displayed in Table 1,
items 14, 16, and 48 were integrated to form one item (item 27 in the MAQ-3), "Anger makes me a
bad person"; Item 23, "Anger could make me go mad," was kept due to its theoretical relevance;
Items 27 and 49 were integrated to one item (item 31 in MAQ-3), "Anger will make other people
Chapter 2 Prisoners, anger, and the MAQ
81
think badly about me"; Item 30, "My anger can harm other people," was kept due to its theoretical
relevance; Item 32 was omitted due to redundancy with item 23; Item 33, "Anger is good for me,"
was omitted due to redundancy with item 26, "Anger keeps me safe," which was preferred from a
theoretical standpoint; Item 42, "When I am angry, I lose sight of different points of view," was
kept due to theoretical relevance even though it loaded on two factors; Item 55, "Anger makes me
insensitive to other people," was kept was kept due to its theoretical relevance; r2, "If I just let go of
my anger, people will not understand that they went too far," was reworded into item 24 in MAQ-3,
"My anger will make people realize that they went too far," because it loaded on two factors: r3, "It
is impossible not to think about anger," and r5, "Thinking about anger will produce solutions," were
combined into one item (item 33 in MAQ-3; "Anger stays with me for a long time").
In summary, the revision of MAQ-2 resulted in a scale containing 26 items and did
not include items from the unstable subscale, MAQ CC2.
Background variables
The relationships between background variables were analyzed. For violent versus
nonviolent offense, using ANOVA there were no significant differences in anger level (PI), MAQ-2
or MCQ-30 scores. The correlations between age and PI (r = -.21, p = .008) and age and MAQ (r =
-.19, p = .013) indicated that the younger the prisoner, the higher the score on the PI and MAQ-2.
The correlation between age and MCQ-30 was not significant (r = .08). The relationships between
anger level and age were consistent with the literature (Taylor & Novaco, 2005). For duration of
schooling, no significant correlations were present. Correlations between sentence length and PI (r
= -.20, p = .01) and verdict length and MAQ-2 (r = -.19, p = .02) were significant, indicating that
those who had longer sentences were less angry and had lower scores on the MAQ-2. The
correlation between verdict length and MCQ-30 (r = -.09) was not significant.
Provocation Inventory (PI)
The mean PI Total score was 65.4, SD = 14.0. Comparing the PI Total from the
present study to that found by (Lindqvist et al., 2005) for Swedish violent prison inmates (M = 62.5,
SD = 15.1), the present PI Total mean was significantly higher (t (166) = 2.64, p = .009).
2 The items comprising the cognitive consciousness subscale were tested in the clinical sample. In the clinical sample,
the 9 items intended to load on the cognitive consciousness subscale showed an internal reliability score of alpha= .71, and again, the external validity was unsatisfactory because its correlations with the anger measures NAS Total and Trait anger were nonsignificant (r = .19 and r = .20). Therefore, the items were not included in the EFA or any additional analyses.
Chapter 2 Prisoners, anger, and the MAQ
82
The alpha for the 25-item PI Total was .92, and the test-retest (Pearson) correlation was .86,
indicating very god stability.
MetaCognitive Questionnaire (MCQ-30)
Comparing the MCQ-30 mean for the present study with that of the general
population in the UK (N = 1304; (Spada, Mohiyeddini, and Wells, 2008), the difference was
significant. In the present study, Cognitive Confidence (M = 12.22, SD = 4.37) was higher than in
the UK sample (M(UK) = 10.4, SD = 4.1; t (166) = 5.37, p < .001), and the Positive Beliefs (M =
12.96, SD = 4.44) score was also higher than that of the UK (M(UK) = 10.8, SD = 4.1; t (166) = 6.28,
p < .001). In the present study, Cognitive Self-Consciousness (M = 16.10, SD = 4.42) was higher
than that of the UK (M(UK)= 14.5, SD = 4.6; t (166) = 4.68, p < .001), and it was also higher on
Negative Beliefs about Uncontrollability and Danger (M = 13.35, SD = 4.62) than that of the UK
(M(UK) = 11.4, SD = 4.7; t (166) = 5.46, p < .001). Finally, in the present study Negative Beliefs
about the Need to Control thoughts (M = 14.26, SD = 4.07) was higher than in the UK sample
(M(UK) = 10.3, SD = 3.8; t (166) = 12.59, p < .001).
Comparing the mean of the MCQ-30 in the present study with that of psychotic
Danish individuals (N =101; (Austin, 2011) produced significant results on two of the MCQ-30
subscales. In the present study, the mean score for Negative Beliefs about Uncontrollability and
Danger (M = 13.35, SD = 4.62) was lower than that of the Danish psychotic patients (M = 14.2, SD
= 4.4; t (166) = 2.50, p = .013). Cognitive Confidence was also lower (M = 12.22, SD = 4.37) than
in Danish psychotic patients (M = 13.32, SD = 4.5; t (166) = 3.3, p = .001). These results indicate
that the present study sample had a higher mean MCQ-30 score than that of the general population
in the UK, lower levels of Uncontrollability and Danger beliefs and a better confidence in their own
cognition compared with Danish psychotic patients.
The alpha for the 30-item MCQ-30 was .89, and the test-retest (Pearson) correlation was .64,
indicating good stability.
Metacognition and Anger questionnaire (MAQ-2)
Internal reliability analyses for the MAQ-2 yielded an alpha coefficient of .86 for the
29 items. The test-retest (Pearson) correlation was .67, indicating good stability for the MAQ-2.
Chapter 2 Prisoners, anger, and the MAQ
83
MAQ-2 inter-subscale correlations
The Pearson intercorrelations among the subscales of the MAQ-2 are available in
Table 1. All subscales of the MAQ-2 were highly correlated with the MAQ total.
With the exception of the correlation between Positive and Negative Beliefs, which was non-
significant in study 1 as well as study 2, the MAQ-2 subscale intercorrelations had increased over
those of study 1. For Positive Beliefs and Rumination, the correlation increased from r = .24 to r =
.52 (z = 3.1, p < .001) and for Negative Beliefs and Rumination the correlation increased from r =
.35 to r = .47 (z = 1.4, p = .09).
Convergent validity
To examine the theoretically expected relationship between the general metacognitive
measure and this new measure of metacognition that specifically targets anger, the intercorrelations
between the two metacognitive measures were computed. Results showed moderate correlations
between several of the subscales on the MCQ-30 and the MAQ-2, supporting the MAQ-2 as a
metacognitive measure. The MCQ-30 subscale regarding Uncontrollability/Danger in relation to
worry was highly correlated with both MAQ-2 Rumination (r = .47) and MAQ-2 Negative Beliefs
about anger (r = .42). In addition, the MCQ-30 Negative Beliefs about the Need to Control thoughts
was highly correlated with MAQ-2 Negative Beliefs about anger (r = .46). These results indicate
similarity between these constructs.
Concurrent validity
To examine the correlations between the previously validated measure and anger level
(PI), correlations for the MAQ-2 and MCQ-30 with PI Total were computed. Overall, the MAQ-2
was more strongly correlated with anger level. The Uncontrollability/Danger subscale of the MCQ-
30 was the most highly correlated with PI (r = .26) of any of the subscales.
As in Study 1, the subscales of the MAQ-2 were positively correlated with PI Total and the same
pattern of correlations emerged as in Study 1, only stronger. The correlation between PI Total and
the MAQ-2 subscales increased for Rumination from r = .37 to r = .65 (z = 3.6, p < .01), for
Positive Beliefs from r = .27 to r = .45 (z = 2.0, p = .03), and for Negative Beliefs from r = .22 to r
= .26 (z = 0.4, p = .35). The increase in correlations between the MAQ-2 subscales and the PI Total
support the scale revisions.
In summary, the subscale intercorrelations and the correlations with PI Total support
the MAQ-2 as a metacognitive measure that has specific relevance to anger. Results of these
correlations are available in Table 2.
Chapter 2 Prisoners, anger, and the MAQ
84
Table 2. Correlations (Pearson) between the MAQ-2 subscales and measures of metacognition and anger level.
MAQ-2 MCQ-30
Rum NB PB 4 5 6 7 8 Total PI Total
PI Total .07 .26* .20
* .22
* ns .22
* 1
M
AQ-2 Total
90* .73
* .64
* ns .51
* .32
* .44
* ns .46
* .60
*
Rum
1
47*
.52*
ns
.47*
.31*
.33*
ns
.37*
.65*
NB 1 ns .22* .42
* .24
* .46
* .38
* .48
* .26
*
PB 1 ns .24* ns ns ns ns .45
*
Note. N = 167. * p < .01. MAQ-2: Rum = Rumination, NB = Negative Beliefs, PB = Positive Beliefs. MCQ-30: 4 = Positive Beliefs, 5 =
Uncontrollability/Danger, 6 = Cognitive Confidence, 7 = Need to Control thoughts, 8 = Cognitive Self-Consciousness, 9 = MCQ-30 Total.
To further examine the relationship between the MAQ-2 subscales and anger level
(PI), a hierarchical regression with forced entry was conducted with PI Total as the criterion
variable and age, verdict length, MCQ-30 Total and MAQ-2 subscales as the predictors. On the first
step, age and verdict length were entered as background covariates. To explore whether the MCQ-
30 is related to anger level, the MCQ-30 Total was entered on the second step. Our intention was
both to test its contribution to and control for its effects when testing the MAQ-2 subscales entered
on step 3.
For the first step, age and verdict length entered alone were significantly associated
with anger level (PI), adjusted R² = .054 (p < .01). When the MCQ-30 Total was added to this
equation on the second step, an additional 6% of the variance in the criterion variable was
explained, producing a significant increase (ΔR² = .058; p = .002). Entering the MAQ-2 subscales
on the third step explained an additional 35% of the variance in the criterion variable (ΔR² of .347,
p < .000). The final model was significant, adjusted R² = .45, F (6,157) = 22.45, p < .000. In the
final model, Rumination and Positive Beliefs from the MAQ-2 subscales were significantly
associated with anger level measured by the PI Total, whereas the MCQ-30 Total was no longer
significant. Results of the regression are presented in Table 3.
Chapter 2 Prisoners, anger, and the MAQ
85
Table 3. Hierarchical regression of anger level (PI), background variables and metacognitive measures MCQ-30 and
MAQ-2
Model 1 Model 2 Model
3
Variable B â t p B â t P B â t p
Step 1
Age -.24 -.17 -2.08 .039 -.28 -.20 -2.50 .014 .11 .08 1.15 .254
Verdict length -.74 -.16 -2.00 .047 -.61 -.13 -1.70 .093 -.37 -.08 -1.29 .201
Step 2
MCQ-30 .22 .24 3.20 .002 -.03 -.03 -.40 .689
Step 3
Rumination 1.26 .60 6.94 .000
Positive beliefs .51 .18 2.47 .015
Negative beliefs -.04 -.02 -.27 .785
Note. Prison sample N=167. ∆R² values for each step are .07 for step 1 (p = .005); .06 for step 2 (p = .002), .35 for step 3 (p < .000). For the final
model, adjusted R² = .45, F (6,157) = 22.45 (p < .000).
The data suggest that the MAQ-2 subscales were more closely associated with anger than the MCQ-
30.
Discussion of Studies 1 and 2
Based on the results described above, the MAQ was deemed a promising clinical scale
for measuring metacognition in relation to anger. The scale showed potential value for
understanding the cognitive mechanisms involved when individuals present with anger-related
problems. The pattern of correlations supported the idea that the subscales are meaningfully related
and represent dimensions of metacognition as it relates to anger.
The factor analyses of the MAQ-1 and MAQ-2 resulted in a four-dimensional
structure of metacognition: Positive Beliefs about anger, Negative Beliefs about anger, Rumination
and Cognitive Consciousness. However, the findings of the studies presented here do not support
the relevance of a maladaptive cognitive self-focus to a metacognitive framework on anger. The
items designed to measure Cognitive Consciousness showed inconsistent factor loadings, and their
concurrent validity was unsatisfactory. In both the police student and prisoner samples,
respectively, Cognitive Consciousness showed non-significant correlations with the anger measure
(PI Total), r =.14 and r =.08. When the meanings of the items on the subscale were studied in
greater detail, inconsistencies in the operational use of the intended construct became apparent. In
the first pool of items, the Cognitive Consciousness subscale consisted of items assessing: (a)
constant awareness of own thinking; (b) attentional focus on potential bad behavior from other
people; (c) difficulty understanding others' emotions; (d) beliefs about the need to control thoughts;
(e) threat detection; (f) difficulty in self-monitoring one's emotions; and (g) beliefs about
Chapter 2 Prisoners, anger, and the MAQ
86
punishment for not controlling thoughts. In the second pool of items on the re-test, the new items
were written to assess: (a) emotional preoccupation and (b) preoccupation with the thought
processes. On reflection, the items may have contained conflicting content that intertwined themes
of cognitive awareness, attempts at thought control, regulation skills and threat detection. Logically,
if the items reflected features of different constructs with different relationships to anger, empirical
findings would be inconsistent.
As addressed earlier, the need to differentiate the type of inner focus is essential for
understanding cognitive processing in maintenance of clinical conditions (Watkins, 2008). This
concept is supported by Salovey, Mayer, Goldman, Turvey, and Palfai (1995), who reported a non-
significant correlation between the Angry Rumination Scale (ARS) and the tendency to attend to
one´s emotional states, as measured by the trait Meta-Mood Scale. The authors speculated that two
different concepts were being assessed. Hence, distinguishing between rumination with an
unproductive self-focus and rumination with a reflective self-focus is necessary. The task of
disentangling a functional inner focus from a dysfunctional inner focus that is driving angry
ruminative processes and other problematic processing strategies related to anger is complex and
difficult. In the metacognitive framework, no distinction between benign and malignant self-focus is
specified. The rationale for considering a general heightened tendency to monitor and focus on
inner experiences as malignant is the risk to of developing the Cognitive Attentional Syndrome
(CAS). Conversely, because anger is an emotional response that may arise quickly and relatively
automatically with limited cognitive processing, an elevated tendency to monitor and focus on inner
experiences may, in fact, enable cognitive modification of the anger response (Wilkowski &
Robinson, 2010). The well-established finding that anger arousal tends to compromise information
processing and decrease self-monitoring skills (Taylor & Novaco, 2005) further supports the notion
that self-focused attention in relation to anger may actually be helpful in some situations. In
conclusion, the findings from these two studies involving the MAQ Cognitive Consciousness
subscale support the idea that self-focused attention may not be malignant in relation to anger per
se, but the type and quality of the self-focus is critical. Therefore, distinguishing a dysfunctional,
ruminative inner self-focus from a more productive self-focus with reflection seems essential in
understanding anger dysregulation.
The other subscales of the MAQ showed more consistent loadings, had higher internal
reliabilities and showed the expected positive associations with anger level; this was consistent with
the metacognitive framework. Examination of the content of the items in the subscales supported
Chapter 2 Prisoners, anger, and the MAQ
87
the emphasis on the symbolic and semantic meanings associated with anger that were reviewed in
the introduction. The Positive Beliefs subscale consisted of items assessing beliefs about anger: (a)
as a survival strategy; (b) as protection against threats and danger; (c) as helpful in asserting
personal borders; and (d) as necessary to deal with everyday life. The Negative Beliefs subscale
consisted of items assessing beliefs about anger: (a) as harmful; (b) as related to madness; (c) as
signifying a loss of control; (d) as dangerous; (e) as compromising information processing; (f) as
related to negative social evaluation; and (g) as compromising to directing adequate attention
towards others. The Rumination subscale consisted of items assessing anger processing: (a) as an
involuntary passion that takes control; (b) as uncontrollable; (c) as attention-demanding; and (d) as
prolonged in duration. The factor analyses confirmed the assumptions consistent with the
metacognitive framework, namely that Positive and Negative Beliefs about anger and Rumination
may be relevant to understanding dysregulated anger.
Regarding the subscale correlations between the MAQ and the general metacognitive
measure (MCQ-30), the results showed moderate correlations between several of the subscales and
the MCQ-30 and the MAQ-2; this supported use of the MAQ-2 as a metacognitive measure. In
particular, the MCQ-30 subscale Uncontrollability and Danger related to worry and the MCQ-30
subscale Need to Control thoughts were positively correlated with the MAQ Rumination and MAQ
Negative Beliefs subscale. This was interesting because the content of these subscales on the MCQ-
30 resembled the MAQ Negative Beliefs and the MAQ Rumination subscales. Furthermore, these
MCQ-30 subscales showed the highest correlations with anger (PI) for the MCQ-30 Need to
Control thoughts subscale (r = .22) and the MCQ-30 Uncontrollability and Danger subscale (r =
.26). The results point to uncontrollability and danger as principal, essential themes in a
metacognitive conceptualization of emotional distress.
The MCQ-30 Cognitive Confidence subscale also showed a positive correlation with
anger as measured by the PI. However, based on the initial pilot testing during which forensic
inpatients were interviewed with an eye towards the metacognitive framework, items reflecting this
domain were not included in the item constructions for the MAQ. This was because the patients in
the pilot interviews predominantly appeared confused about the relevance of confidence to one's
own mental capacity to discussing anger experiences and anger regulation. Due to the positive
correlation with the PI, the relationship between the Cognitive Confidence subscale (MCQ-30) and
anger should be further explored.
Chapter 2 Prisoners, anger, and the MAQ
88
The general idea to take a metacognitive perspective on anger, highlighting the link
between metacognitive beliefs and specific strategies for processing information (Wells, 2000;
Wells & Matthews, 1994), were supported by the high intercorrelations between the MAQ
subscales. Thus, the present data indicate that positive as well as negative beliefs are involved in the
tendency to ruminate about angry emotions. Furthermore, this new tool showed potential clinical
relevance in regard to anger related problems as its subscales were significantly correlated with
anger level (PI). The Rumination subscale, which was most strongly correlated with the anger
measure in both studies (r₁ = .37, r₂ = .65), measures the tendency to “get stuck” on angry thoughts
and emotions even when it is not intended. Others have developed tools to measure rumination as it
relates to anger (e.g. Angry Rumination Scale ARS; (Sukhodolsky et al., 2001) and the Dissipation-
Rumination Scale DRS; (Caprara, 1986), however, these scales have not attempted to integrate the
tendency to ruminate with the associated belief structures that drive the selection of this processing
strategy as was done in the MAQ. Rumination as a processing strategy for anger is interesting
because previous studies have found a robust link between rumination and anger level (Rusting &
Nolen-Hoeksema, 1998; Caprara et al., 2007; Caprara, 1986; Novaco, 1994).
The Positive Beliefs subscale, which was correlated at r₁ = .27, r₂ = .45 with the anger
measure, measures beliefs about the positive functions of anger. The construct bears a resemblance
to the belief that physical force is the best means of protection (Archer & Haigh, 1997b; 1997a).
However, because anger and aggression are distinct constructs, beliefs related to the functions of
anger and aggression are also distinct. Further exploring the construct validity of the positive beliefs
subscale further would be valuable.
The Negative Beliefs subscale, which correlated at r₁ = .22, r₂ = .26 with the anger
measure, measures beliefs concerning the negative labels related to anger that were discussed in the
introduction. The fact that these negative evaluations are significantly correlated with anger (PI)
may facilitate the understanding of anger-related problems. Because the study used correlations, it
is not possible to draw any causal conclusions; however, the present results may indicate that
negative beliefs about anger are involved in anger dysregulation. On the other hand, Negative
Beliefs were not significant in the regression analysis, and thus, the relationship between Negative
Beliefs and anger requires further investigation.
All correlations increased from study 1 to study 2, supporting the scale revisions and
indicating improved relevance of the scale for a clinical sample of prisoners with high anger levels
over healthy police students with relatively low anger levels. A theoretical view of anger that
Chapter 2 Prisoners, anger, and the MAQ
89
emphasizes the dual nature of the construct is consistent with the finding that both Positive and
Negative beliefs about anger were positively correlated with anger level as measured by the PI.
In conclusion, the aim of these studies was to investigate the relevance of a
metacognitive framework for anger. The studies evaluated the psychometric properties of a new
instrument, the MAQ, which was designed to assess metacognitive beliefs and processes in relation
to anger. Whereas the MCQ-30 questionnaire measures metacognition as it relates to worry, the
MAQ was developed to measure metacognition specifically as it relates to anger. The significantly
stronger correlations between the MAQ and the PI compared to those between the MCQ-30 and the
PI support that the MAQ is an instrument that specifically measures metacognition in relation to
anger. Moreover, correlations between the MCQ-30 and the MAQ provide evidence for convergent
validity. As expected, both positive and negative beliefs were correlated with rumination.
The two studies presented have several limitations. First, a substantial weakness
regarding the factor analyses conducted in these studies was sample size. In the first factor analysis
the ratio was 3.4 participants per item, and in the second, it was 3.7 participants per item. Usually, a
ratio of 5 participants per item is recommended (Gorsuch, 1983) to ensure stability of the factor
structure. Secondly, the studies are cross-sectional and rely on correlational statistics, meaning that
a causal relationship cannot be inferred. Finally, the assessment battery used in both studies was
very limited and anger was measured with only one instrument. Future studies that wish to refine
the scale will benefit from expanding the test battery, allowing for more precise investigation of the
psychometric properties of the MAQ and testing of more specific hypotheses. Using clinical
samples would also be valuable in future validations of the scale.
Chapter 2 Prisoners, anger, and the MAQ
90
Chapter 3 Clinical patients, anger, and the MAQ
91
Chapter 3 Clinical patients, anger, and the MAQ
Introduction
The metacognitive framework of Wells and Matthews (Wells & Matthews, 1994;
Wells, 2000) offers a conceptualization of emotional disorder that influenced the development of
the MAQ. In the two previously presented studies, the factor structure of the MAQ as an adapted
metacognitive framework for anger was explored. The results of those studies were promising, but
the clinical application of the MAQ still needs to be evaluated. Because the metacognitive
framework of Wells and Matthews (Wells & Matthews, 1994; Wells, 2000) is primarily used to
facilitate the understanding and treatment of clinical conditions and because anger has been
demonstrated to frequently be involved in psychopathology, evaluating the psychometric properties
of the MAQ in a clinical setting is crucial. Therefore, the present study tests the MAQ's
metacognitive anger framework in a mixed clinical sample to evaluate its advantages over a general
metacognitive framework, specifically in relation to anger.
The empirical evaluation of the MAQ in the present study concerns its convergent
validity with both metacognitive framework and anger criteria.
Prior to Study 3, the MAQ-2 was revised on the basis of Study 2 findings. Thus, Study
3 concerned the MAQ-3. In some cases, the revisions meant a deviation from strict adherence to
factor loading from the MAQ-2 as the criterion for item selection. Due to overlapping content, some
items were combined into one item, one item was omitted due to redundancy with another item, and
one item was retained because of its theoretical relevance even though it loaded on two factors.
While the results of the previous studies did not support the value of the Cognitive Consciousness
subscale, in case clinical status has a bearing on the outcome, the subscale was tested again in the
clinical sample.
Setting
The study was conducted in the psychiatric facilities located in two rural towns in
Denmark, Vordingborg and Naestved. In Vordingborg, two closed wards, one open ward and one
outreach team participated. In Naestved, two outreach teams participated. The participating wards
and included a variety of clinical conditions, ranging from outpatients with emotional disorders to
Chapter 3 Clinical patients, anger, and the MAQ
92
psychotic inpatients in closed wards. Thus, the settings represented the natural variability of
Denmark’s psychiatric facilities.
Participants
The participants were adult psychiatric patients, including forensic patients not placed
at a specific facility (N = 88). The mean age of patients was 38.1 (SD = 14.2, range 16-74).
Participants were approached by the researcher in both individual and group settings and invited to
participate. Recruitment was also conducted by the staff. Sample characteristics are presented in
Table 1.
Table 1. Socio-demographic and psychiatric characteristics of the 88 clinical patients.
Characteristic %
Gender (male)* 46.6
Education Compulsory school only 14.8
Finished graduate school only 33.0
Some form of education 44.3
Living arrangement With a partner 34.1
Employment Regular job 10.2
Job on special terms 13.6
Temporary social benefit 36.4
Retired due to mental problems 31.8
Diagnosis Affective 48.9
Psychotic 26.8
Other₁ 24.3
Drug/alcohol problems 12.5
Treatment type Outpatient 66.0
Forensic 12.5
Violent offence 81.8
Note. *4 missing. ₁. Alcohol abuse, attention deficit hyperkinetic disorder and personality disorder.
Analyses of the background variables presented in Table 1 found only two significant
differences in the MAQ-2 indices, both of which occurred in conjunction with the Positive Beliefs
scale. Males held more Positive Beliefs about anger than females, t (82) = 2.69, p = .022, and
individuals living with a partner held fewer Positive Beliefs about anger than did those living alone,
t (86) = -2.18, p = .032. In view of the largely non-significant differences in demographics and
psychiatric group characteristics, the participants were studied as one sample.
Chapter 3 Clinical patients, anger, and the MAQ
93
Measures
A total of seven self-report measures with established psychometric properties were
administered to examine the relationship between metacognition and anger in a clinical sample. The
assessment package contained a total of 221 items and included the following measures:
Novaco Anger Scale (NAS; Novaco, 2003): The NAS is a 60-item scale constructed to
measure anger. It measures anger in a Cognitive domain, an Arousal domain, and a Behavioral
domain that together form the NAS Total score, with a separate anger regulation subscale.
Stait Trait Anger eXpression Inventory (STAXI-2;(Spielberger, 1999): The STAXI-2
is a 57-item scale constructed to measure a broad range of anger experiences and controls. It
consists of 6 subscales measuring State Anger, Trait Anger, and components of Anger Expression
(Anger In, Anger Out, and Anger Control).
The MetaCognitive Questionnaire (MCQ-30;(Wells & Cartwright-Hatton, 2004): The
MCQ-30 measures general aspects of metacognition. The focus of the questionnaire concerns worry
and experiences and beliefs related to worry. It consists of 30 items that the participant rates on a
scale from 1 = "do not agree"; 2 = "agree slightly"; 3 = "agree moderately"; and 4 = "agree very
much".
The Metacognition and Anger Questionnaire (MAQ-3): This 26 item MAQ-3 is the
revised MAQ-2. The Cognitive Consciousness subscale was eliminated because in studies 1 and
Study 2, it was not significantly correlated with either the NAS or STAXI anger measures.
The Anger Rumination Scale (ARS;(Sukhodolsky et al., 2001): The ARS measures the
tendency to think about anger. It contains 19 items on four factors: Angry-Afterthoughts (6 items);
Thoughts of Revenge (4 items); Angry Memories (5 items); and Understanding the Causes (4
items). In a student sample, internal reliability coefficients satisfactorily ranged from .72 to .83 for
the subscales and .93 for the ARS Total. In addition, the one-month test-retest reliability was
adequate. Convergent validity tests were conducted with the STAXI-2 Trait, reporting positive
correlations ranging from .41 to .57 for the subscales. The questionnaire was translated by the
author of this thesis and back-translated by a bilingual translator, with permission from the author of
the original questionnaire.
The Hospital Anxiety and Depression Scale (HADS;(Zigmond & Snaith, 1983): This
instrument is a 14-item,self-report questionnaire measuring anxiety and depression. The respondent
Chapter 3 Clinical patients, anger, and the MAQ
94
provides ratings that reflect their most recent week. Seven items measure anxiety, and 7 items
measure depression. Higher scores indicate higher levels of anxiety and depression. Its reliability
and validity have been established in clinical settings as well as in the general population (Bjelland,
Dahl, Haug, and Neckelmann, 2002). The questionnaire was available in the public domain and
translated and back-translated by a bilingual translator.
Across participants and measures, less than 5% of responses were missing, and no
respondent was missing more than a total of 3 items. The values for the missing items were replaced
with the series mean for the item.
Procedure
The study was approved by the Danish Data Control system. The participants were
approached by ward nurses or by the primary researcher, who provided oral as well as written
information. The primary researcher participated in patient/staff meetings to inform them about the
study. All participants were assured that participation in the study was voluntary, and signed
consent to participation was obtained. It was emphasized that participants would not be identified in
subsequent reports and that their personal data would not be used in the psychiatric system.
All patients who spoke Danish and did not meet criteria for dementia or other organic
problems were invited to participate. One patient was not included due to acute psychotic symptoms
at the time of assessment (i.e., the patient hear voices that required him to select reply number two
for all questions). Some participants filled out the questionnaires on their own, some were
individually administered the questionnaire, and some participated in a group format (4-8
participants). It was not possible to analyze characteristics of the patients not willing to participate
although comparisons were made between the background variables of the participant sample and
the general hospital population.
Hypotheses
Hypothesis 1: The MAQ will be significantly correlated with HADS Anxiety and HADS
Depression.
As discussed in Chapter 2 of the Introduction, anger is associated with affective
symptoms of psychopathology, which is why it is expected that anxiety and depression will be
positively correlated with the MAQ as well as the anger measures (NAS and STAXI-2).
Chapter 3 Clinical patients, anger, and the MAQ
95
Based on the idea that threat is a common theme in anger and anxiety, it is
hypothesized that the MAQ Negative Beliefs subscale, which reflects experiences of danger and
madness in relation to anger, will show a significant, positive correlation with HADS Anxiety.
Furthermore, because uncontrollable worry is a core feature of anxiety, the HADS
Anxiety measures uncontrollable worry. Because rumination and worry have been conceived as
related concepts and the MAQ Rumination subscale reflects uncontrollable rumination, this
subscale is also expected to be positively correlated with the HADS Anxiety measure.
Hypothesis 2: The MAQ will be significantly correlated with the MCQ-30.
Given that the MAQ is intended to represent a metacognitive framework, it is
expected that the MAQ subscales will be positively correlated with the MCQ-30 Total. In
particular, due to the theoretical similarities in subscale content, the MAQ Rumination and MAQ
Negative Beliefs subscales should be positively correlated with the MCQ-30 Uncontrollability and
Danger subscale as well as the MCQ-30 Need to Control thoughts subscale.
In addition, given that the MAQ is a metacognitive measure specific to anger, it is
hypothesized that the MAQ subscales will be more strongly correlated with the anger scales (NAS
and STAXI-2) than the MCQ-30 subscales.
Hypothesis 3: Variables representing uncontrollability will be associated with anger criteria (NAS
Total and STAXI-2).
In support of the idea that the general themes concerning uncontrollability, danger,
and madness are centrally involved in the monitoring, regulation and control of mental phenomena,
the 4 subscales that address these issues, namely MAQ Rumination, MAQ Negative Beliefs, MCQ-
30 Uncontrollability and Danger, and MCQ-30 Need to Control thoughts, are hypothesized to show
significant, positive correlations with anger level (NAS-PI and STAXI-2).
Hypothesis 4: The MAQ Positive Beliefs will be significantly correlated with NAS
justification/hostility and suspiciousness items.
Cognition that justifies anger and externalizes blame for negative and dangerous
events is associated with anger. The Positive Beliefs subscale of the MAQ address beliefs about the
need for and benefits of anger, which are associated with cognition related to justification, hostile
Chapter 3 Clinical patients, anger, and the MAQ
96
attitude and suspiciousness. Based on the idea that these beliefs function as a higher-order cognitive
network facilitating an anger response in situations of perceived unjustified, malicious or
threatening events, the NAS Cognitive justification items (i.e., hostile attitude and suspiciousness
items) are expected to be positively correlated with the MAQ Positive Beliefs.
Hypothesis 5: MAQ Rumination will be significantly correlated with the ARS, the NAS
Rumination items, and with STAXI Anger-In.
Given that the ARS is a validated measure of anger rumination that is supported in
multiple studies, it is a central measure on the MAQ Rumination scale. The NAS is a validated
anger measure that includes 4 rumination items in its Cognitive Domain subscale. As another
convergent validity test, the MAQ Rumination scale is also expected to be correlated with the NAS.
It is also expected that MAQ Rumination will be associated with STAXI Anger In, which measures
the withholding of anger expression, because failure to express anger engenders rumination about
provoking experiences. In Sukhodolsky et al. (2001), the ARS was more strongly correlated with
the STAXI Anger In than with the other STAXI Anger Expression subscales (Anger Out and Anger
Control).
Hypothesis 6: The MAQ Negative Beliefs and the MAQ Rumination subscales will be significantly
correlated with anger arousal criteria (NAS Arousal).
Negative evaluations of anger are incorporated into the MAQ Negative Beliefs
subscale with items such as, "my anger could make me go mad"; "anger means loss of control";
"my anger is dangerous for me"; and, "anger makes me insensitive to others". The MAQ Negative
Beliefs items are hypothesized to reflect the experience of anger as uncontrollable or, "taking over,"
as reflected in the NAS Arousal subscale, which comprises items related to anger intensity, somatic
tension, and irritability.
In chapter 1, bodily arousal was asserted to be an important facet of the anger
experience. The intensity and duration of the physiological arousal that are experienced are
important for anger regulation because heightened and prolonged arousal can impair executive
functioning (Chemtob et al., 1997) and increase the likelihood of excitation transfer effects
(Zillmann, 1979; 1988). In chapters 3 and 4, rumination was discussed for its association with
physiological arousal and anger (Gerin et al., 2006). Given that the NAS Rumination correlates
strongly with the duration items of the NAS Arousal subscale (Novaco, 1994), this is substantiated.
Chapter 3 Clinical patients, anger, and the MAQ
97
Therefore, as a convergent validity test, the MAQ Rumination scale is hypothesized to show a
significant, positive correlation with the NAS Arousal. More specifically, the MAQ Rumination is
expected to show a strong positive correlation with the duration item set of the NAS Arousal.
Hypothesis 7: The MAQ will be significantly correlated with the NAS Regulation and the STAXI
Anger Control.
Because the MAQ measures dysfunctional beliefs and processes in relation to anger
regulation, the MAQ Positive Beliefs, MAQ Negative Beliefs, and MAQ Rumination are expected
to show significant negative correlations with anger regulation criteria (the NAS Regulation and the
STAXI Anger Control).
Hypothesis 8: In a hierarchical regression analysis, the MAQ will be a better predictor of anger (the
STAXI-2 and the NAS Total) than the MCQ-30.
Lastly, to test the MAQ as a metacognitive measure specific to anger, the MAQ
subscales are expected, in a hierarchical regression, to account for a significant portion of the
variation in anger level, as measured by the STAXI-2 and the NAS. Variation accounted for by the
MAQ subscales should be higher than the amount of variation accounted for by the MCQ-30.
Results
Factor analysis
Data for the MAQ met assumptions of normality, permitting factor analysis of the
scale. The Cognitive Consciousness subscale was not included in the factor analysis because (as in
Studies 1 and 2) it was not correlated with the anger measures. The remaining 26 items comprising
the MAQ (MAQ-3) were entered in a Principle Components Analysis (PCA) with Promax rotation
and three fixed factors. The solution accounted for 58.0 % of the total variance, with item loadings
ranging from .36 to 1.0 on the expected factors. Item loadings are displayed in Table 2.
Chapter 3 Clinical patients, anger, and the MAQ
98
Table 2. Factor loadings of the PCA with Promax rotation for clinical patients, N = 88.
Item Number Clinical patients (N = 88) MAQ-3
1. Rumination 2. Positive Beliefs 3. Negative Beliefs
1.When I am angry, I keep thinking about it .54 .14 .15
3. My anger harms me .06 .04 .76
4. Anger helps me see things the way they really are -.12 .74 .17
5. I cannot step back from my angry thoughts .77 -.04 .06
7. Anger could make me go mad .27 .17 .57
8. Anger helps me to solve problems -.07 .77 -.19
9. I cannot let go of angry thoughts .83 -.02 -.16
11. My anger could hurt others .05 .26 .56
12. Anger helps me handle threats and dangers -.24 .70 .34
13. Anger is hard to control; it controls you .63 -.13 .25
15. Anger means loss of control .13 -.40 .54
16. Anger protects me .45 .59 -.29
17. When I start to get angry, I cannot stop .87 .03 -.18
19. My anger is dangerous for me -.29 .02 1.0
20. Anger makes me a strong and competent person .15 .61 -.04
21. I cannot ignore my anger .66 .10 .12
23. When I am angry, I lose sight of different points of view .62 .04 .21
24. My anger will make people realize that they went too far .23 .57 -.08
25. When I am angry, I cannot distract myself .89 -.13 -.12
27. Anger makes me a bad person .28 -.41 .61
28. Anger is necessary to get by in the world -.10 .83 .07
29. When I am angry, I can only think about that .48 .18 .35
31. Anger will make other people think badly about me .34 .06 .36
32. Anger keeps me alert .01 .80 -.03
33. Anger stays with me for a long time .74 -.01 .13
35. Anger makes me insensitive to others .26 .16 .47
Note. Sample 3: PCA Promax rotated, 3 fixed factors (58.0 % of the variance explained).
Item 16 loaded on 2 factors, item 23 loaded on an unexpected factor and 31 loaded on 2 factors.
Confirmatory factor analyses
To confirm the factor structure suggested by the explorative factor analysis, the data
was fitted using a Confirmatory Factor Analysis that was performed with the M-plus statistical
software, version 6 (L.K. Munthén & B.O Munthén, 2010).
The test provides overall information about the degree to which a specified
structural model explains the data in a particular sample by comparing the expected covariance with
the observed covariance. In addition, other fit indices can be used to test the fit of a latent variable
model. The Tucker-Lewis Index (TLI), the Comparative Fit Index (CFI), the Root-Mean-Square
Error of Approximation (RMSEA) and the Weighted Root-Mean-square Residual (WRMR) tests
Chapter 3 Clinical patients, anger, and the MAQ
99
can be used. For a good-fit model, the test should be non-significant or the ratio of the
divided by the degrees of freedom should be less than 2; it is also recommended that CFI values
be larger than 0.9 -.95, the RMSEA value be below 0.05-.06, the WRMR value be smaller than .95 -
1.0, and the TLI value be larger than .95 (Ching-Yun, 2002; Ullman, 2007).
First, the three-factor model with 26 items derived from the EFA was tested. The first
CFA was significant ( (296, N = 88) = 470.24, p < .000), however, the value of divided by the
degrees of freedom was less than 2, and the other indices yielded an acceptable fit: CFI = .93; TLI =
.93; RMSEA = .08 and WRMR = 1.09. The residual variance not accounted for by the model led to
testing a model without items 1, 20, 27, and 32. This model comprising 22 items yielded improved
goodness-of-fit indices; (206, N = 88) = 307.11, p < .000; CFI = .96; TLI = .95; RMSEA = .08;
WRMR = .94. By allowing local dependency within items in the same subscale, the goodness-of-fit
indices increased to (189, N = 88) = 237.05, p = .01; CFI = .98; TLI = .97; RMSEA = .05;
WRMR = .73. This model suggests good fit on all indices. The first factor was Rumination (8 items,
alpha = .91), the second factor was Positive Beliefs (6 items, alpha = .84), and the third factor was
Negative Beliefs (8 items, alpha = .86). Overall, the internal consistency coefficients of the MAQ
were satisfactory to excellent. The results are displayed in Table 3.
Table 3. Fit indices for confirmatory factor analytic models of the MAQ
Sample Model 2 df 2/ df p CFI TLI RMSEA WRMR
Clinical (N = 88)
Three-factor (26 items) 470.24 296 1.59 <.000 .93 .08 1.1
Three-factor (22 items) 307.11 206 1.49 <.000 .96 .95 .08 .94
Three-factor (22 items)₁
237.05 189 1.25 <.01 .98 .97 .05 .73
Note: ₁ local dependence of two items within the same subscale was allowed in 17 incidents.
Chapter 3 Clinical patients, anger, and the MAQ
100
Descriptive Statistics for Study Measures
Table 4 presents the descriptive statistics for the NAS, STAXI, ARS, HADS, MCQ-30
and MAQ-3 for the clinical patient sample.
Table 4. Descriptive statistics for the mixed clinical patients (N = 88).
No. Of
items
M SD Skewness Kurtosis Alpha
NAS Cognitive 16 31.37 5.41 -.13 -.60 .78
NAS Arousal 16 33.62 6.80 -.19 -.83 .88
NAS Behavioral 16 30.14 7.35 .08 -1.03 .89
NAS Regulation 12 24.59 4.31 .16 -.24 .78
NAS Total 48 95.13 17.97 -.10 -.97 .94
STAXI-2 Trait 10 23.02 8.32 .21 -1.13 .91
STAXI-2 State 15 22.43 10.98 1.71 2.10 .92
STAXI-2 Anger Expression Out 8 16.38 5.25 .32 -.71 .79
STAXI-2 Anger Expression In 8 19.66 4.58 .13 -.19 .68
STAXI-2 Anger Control Out 8 20.99 6.15 .15 -1.13 .90
STAXI-2 Anger Control In 8 19.61 6.09 .23 -.67 .89
ARS Total 19 43.23 12.82 .25 -.25 .93
HADS Anxiety 7 10.49 4.62 .10 -.85 .80
HADS Depression 7 8.53 4.88 .13 -.65 .82
MCQ-30 Cognitive Confidence 6 14,45 5.46 3.26 -1.13 .87
MCQ-30 Positive Beliefs 6 12.57 5.16 .50 -.76 .88
MCQ-30 Cognitive Self-Consciousness 6 15.97 4.82 -.21 -.91 .84
MCQ-30 Uncontrollability and Danger 6 16.02 4.72 -.16 -.83 .77
MCQ-30 Need to Control 6 15.08 4.79 -.07 -1.03 .78
MCQ-30 Total 30 74.1 16.4 -.16 -.55 .89
MAQ Rumination 8 19.18 6.43 .09 -.87 .91
MAQ Positive Beliefs 6 12.48 4.54 .60 -.33 .84
MAQ Negative Beliefs 8 19.53 6.33 .04 -.84 .86
MAQ-3 Total 22 51.19 14.16 .04 -.79 .92
To provide a frame of reference for the sample means in the present study, t-test
comparisons for the NAS and STAXI-2 were conducted against the means reported by (Lindqvist et
al., 2005) for violent male Swedish inmates. There were significant sample group differences for
the NAS- Arousal (M(Lindqvist) = 30.09, SD = 6.3), t (87) = 4.87, p < .000, for the STAXI Trait Anger
(M(Lindqvist) = 18.92, SD = 5.5), t (87) = 4.62, p < .000, and for the STAXI Anger Control In
(M(Lindqvist) = 21.14, SD = 4.9), t (87) = -2.36, p = .02. These results indicate higher anger scores and
lower anger control scores in the present sample compared to the male Swedish inmates.
Chapter 3 Clinical patients, anger, and the MAQ
101
Comparing the mean of the MCQ-30 Total in the present study with the mean reported
by Spada et al. (2008a) for the general population in the UK (M = 57.4, SD = not reported) revealed
that the present mean was significantly higher (M = 74.1, SD = 16.4), t (87) = 9.60, p < .000. This
clinical sample mean for the MCQ-30 Total was also significantly higher than that of the male
prisoners in Study 2 (M = 68.9, SD = 15.6), t (87) = 2.98, p = .004. Testing for differences
regarding the present study's MCQ-30 subscale means and those found by Austin (2011) for 101
psychotic participants in Denmark, there were significant results on two of the MCQ-30 subscales.
The mean in the present study for Negative Beliefs about Uncontrollability and Danger (M = 16.02,
SD = 4.7) was higher than that of the Danish psychotic patients (M = 14.2, SD = 4.4); t (87) = 3.52,
p = .001) as was the Need to Control thoughts (M = 15.08, SD = 2.66 vs. M = 13.32, SD = 4.5; t
(166) = 2.66, p = .009).
Regarding the HADS Anxiety and Depression, there were no significant differences
between the present study means and those found by Spinhoven, Ormel, Sloekers, Kempen,
Speckens, and van Hemert (1997) for Dutch psychiatric outpatients.
Background variables and affective symptoms
The MAQ-3 subscales, NAS Total, and STAXI Trait Anger were examined for
correlations with age, education length, and symptoms of depression and anxiety as measured by
the HADS. Alpha was set at p < .01. Results are given in Table 5. Younger age and higher anxiety
were significantly associated with higher anger scores. As expected, anxiety was also positively
correlated with the MAQ Rumination and Negative Beliefs scales. The hypothesis that anxiety
(HADS Anxiety) would be positively correlated with anger measures (the NAS Total and the
STAXI-2 Trait anger) was confirmed, whereas the expected positive correlation between depression
(HADS Depression) and anger was not.
Table 5. Correlations between the NAS Total, STAXI-2 Trait and MAQ subscales with background variables and
HADS
Note. * p < .01.
Age Education HADS - anxiety HADS - depression
NAS Total -.38* - .29* .53* ns
STAXI-2 Trait anger -.41* ns .40* ns
MAQ
Rumination
Positive beliefs
Negative beliefs
ns ns .54* ns
ns ns ns ns
ns ns .36* ns
Chapter 3 Clinical patients, anger, and the MAQ
102
MAQ-3 inter-subscale correlations
The intercorrelations of the MAQ subscales ranged from .33 to .75. All of the
subscales were positively correlated with the MAQ Total at the .01 significance level. The pattern
of intercorrelations was different from the findings in Study 2. The MAQ Positive and Negative
Beliefs were, as expected, significantly correlated (r = .34). However, in Study 2, the correlation
between Positive Beliefs and Rumination was r = .52, whereas in the present study it was r = .33;
however, this difference is not significant. In addition, the correlation between Rumination and
Negative Beliefs had increased compared to Study 2 (r(study2) = .47 and r(present study) = .75; z = -3.5, p
< .000).
Convergent validity
To examine the MAQ as a metacognitive measure, correlations with the MCQ-30
were computed. The results are presented in Table 6. As hypothesized, the MAQ subscales were
positively correlated with the MCQ-30 Total. The correlations between the MAQ subscales and the
MCQ-30 subscales were generally in the same direction as in Study 2.
Confirming the hypothesis and indicating close affiliation between the subscales,
particularly strong correlations within the MAQ subscales were found for the MCQ-30 Negative
Beliefs about Uncontrollability and Danger and the Negative Beliefs about the Need to Control
Thoughts.
Table 6. Intercorrelations of MAQ-3, MCQ-30.
MAQ MCQ-30
Total 1 2 3 4 5 6 7 8 Total
MAQ
1. Rumination .89* 1 ns .43* .42* ns ns .24
2. Positive Beliefs .62* .33* 1 ns ns ns .38* ns .37*
3. Negative Beliefs .89* .75* .34* 1 ns .45* .37* ns ns .28*
Note. * p < .01. MCQ-30: 4 = positive beliefs about worry; 5 = negative beliefs about uncontrollability and danger; 6, = cognitive confidence; 7 =
negative beliefs about the need to control thoughts; 8 = cognitive self-consciousness.
Concurrent validity
Supporting the relationship between the MAQ and anger, all correlations between the
MAQ subscales and the NAS and STAXI subscales were significant. The results are presented in
Table 7.
Confirming the hypothesis that the MAQ is a measure with specific relevance for
anger, the correlations between the general metacognitive measure MCQ-30 and the anger measures
were generally weaker than for the MAQ.
Chapter 3 Clinical patients, anger, and the MAQ
103
In addition, in support of the general metacognitive idea arguing for the importance of
themes of uncontrollability, danger and madness in the regulation and control of mental
phenomena, the 4 subscales addressing this concept (i.e., the MAQ Rumination, MAQ Negative
Beliefs, MCQ-30 Uncontrollability and Danger, and MCQ-30 Need to Control thoughts) showed
the expected significant positive correlations with anger level (NAS-PI and STAXI-2).
Interestingly, the MCQ-30 Cognitive Confidence subscale showed a significant
correlation with the NAS and the STAXI-2 Trait anger.
In summary, the results provide support for the MAQ as a metacognitive construct
measure specific to anger.
Table 7. Intercorrelations with the NAS and STAXI-2 subscales for mixed clinical patients (N = 88).
Note. * p < .01. MCQ-30: 1 = positive beliefs about worry; 2 = negative beliefs about uncontrollability and danger; 3 =
cognitive confidence; 4 = negative beliefs about the need to control thoughts, and 5 = cognitive self- consciousness.
It was hypothesized that the MAQ Positive Beliefs would, in particular, be associated
with cognitions justifying anger and externalizing blame. The MAQ Positive Beliefs showed the
expected positive correlation with the NAS Cognitive subscale (r = .61), including the expected
positive correlations with the content categories of the NAS Cognitive subscale of justification, r =
.63; hostile attitude, r = .49, and suspiciousness, r = .52.
The MAQ Rumination was predicted to show a significant positive correlations with
the NAS Cognitive rumination, which it did (r = .67), and with the ARS, which it did (r = .72). In
addition, the MAQ rumination was positively correlated with the STAXI-2 Anger In (r = .52).
The hypothesis that negative evaluations as reflected in the MAQ Negative Beliefs
would be associated with the experience of anger as uncontrollable as reflected in the NAS Arousal
MAQ MCQ-30
Rumination Positive beliefs Negative beliefs 1 2 3 4 5
NAS
Cognitive
Arousal
Behavioral
NAS-Total
.60* .61* .63* ns .40* .31* .31* ns
.79* .39* .71* ns .56* .42* .30* ns
.69* .43* .68* ns .33* .38* ns ns
.75* .50* .76* ns .47* .41* .28* ns
STAXI
Trait Anger
State Anger
.64* .42* .63* ns .36* .31* Ns ns
.47* .42* .43* ns ns ns .28* ns
Chapter 3 Clinical patients, anger, and the MAQ
104
subscale was confirmed; these subscales showed the expected positive correlation (r = .71). In
greater detail, the MAQ Negative Beliefs showed the expected significant positive correlations with
the NAS Arousal and items relating to intensity (r = .70), somatic tension (r = .47) and irritability (r
= .50).
The hypothesis concerning rumination and physiological arousal was particularly
important. It was predicted that the MAQ Rumination subscale would show a significant and
positive correlation with the NAS Arousal, which it did (r = .79), and with the duration items of the
NAS Arousal, (r = .72). These results confirmed that rumination intensifies and prolongs
physiological arousal. The results are presented in Table 8.
Table 8. Correlation of MAQ-3 subscales and anger control/regulation subscales (STAXI-2- Expression and Control
and NAS Regulation) and categories of anger subscales (NAS).
MAQ-3
Rumination Positive beliefs Negative beliefs
STAXI-2 Anger Expression Out (AX-out) .61* .37* .69*
Anger Expression In (AX-in) .52* ns .53*
Anger Control Out (AC-out) -.45* ns - .38*
Anger Control In (AC-in) - .42* ns - .33*
NAS
Regulation - .35* ns ns
Cognitive
Justification .49* .63* .48*
Rumination .67* .34* .57*
Hostile attitude .48* .49* .48*
Suspiciousness .37* .52* .49*
Behavioral
Impulsive reaction .66* .21* .65*
Verbal aggression .46* .39* .52*
Physical confrontation .51* .39* .56*
Indirect expression .54* .29* .62*
Arousal
Intensity .65* .51* .70*
Duration .72* .28* .62*
Somatic tension .61* .29* .47*
Irritability .60* ns .50*
ARS .72* .42* .63* Note. * p < .01.
Chapter 3 Clinical patients, anger, and the MAQ
105
Metacognition and anger regulation
Reflecting dysfunctional beliefs and processes in relation to anger regulation, the
MAQ subscales were expected to show negative correlations with measures of anger regulation and
control. Regarding the predictions concerning the STAXI-2 measure, nine of 12 correlations with
the STAXI Anger Expression subscales were significant. The hypothesis that MAQ would show
significant, positive correlations with STAXI Anger Control was partially confirmed because the
MAQ Positive Beliefs did not show the expected correlation. Regarding the predictions concerning
the NAS regulation, these were confirmed only for the Rumination subscale.
No specific hypotheses were formed regarding the relationship between the MCQ-30
subscales and anger regulation. However, it is worth noting that the results showed the following
two significant correlations with the NAS Regulation; Positive Beliefs about worry, r = .29 (p <
.01) and Cognitive Self-Consciousness, r = .42 (p < .01). Interestingly, the MCQ-30 Cognitive
Confidence and the NAS Regulation showed a negative correlation, r = -.28 (p < .01), indicating
that the less confidence in one's own cognitive processes, the lower the anger regulation. This
finding is in accordance with the S-REF model.
Regression analyses of the MAQ subscales
Finally, to further examine the MAQ as a tool with specific relevance for assessing
metacognitive processes in relation to anger, hierarchical regressions with forced entry were
performed with the NAS Total and STAXI Trait Anger as the criterion variables. Age, length of
education, anxiety, and MCQ-30 Total were entered as covariates and the MAQ subscales were
tested as predictors. Age and length of education were entered as background covariates on the first
step. The HADS Anxiety was entered on the second step. Because the HADS Depression had
shown a zero-order correlation with the anger measures and with the MAQ, it was not entered. On
the third step, the MCQ-30 was entered to test its use as a metacognitive measure and to raise the
bar for the test of the MAQ subscales on the final step. When age and education were entered on
Step 1, their association with the NAS anger level was significant (R² = .21, p < .01). Entering the
HADS anxiety on Step 2 accounted for an additional 21% of the variance. On the third step in
which the MCQ-30 Total score was entered, the change in variance explained was not significant
(ΔR² = .03; p = .066). When the MAQ subscales were entered on Step 4, they were highly
significant (ΔR² = .34; p < .000), with Positive Beliefs and Negative Beliefs as significant predictors
of the NAS Total and Rumination approaching significance (p = .067). Also, age, education length,
Chapter 3 Clinical patients, anger, and the MAQ
106
and anxiety were significant in that model. In the final model, 77% of the variance in the criterion
variable was accounted for (R² = .77; F (7,79) = 38.75, p < .000). The results are displayed in Table
9.
Table 9. Hierarchical regression of anger level (NAS) as associated with background variables and metacognitive
measures MCQ-30 and MAQ-3.
Model 1 Model 2 Model 3 Model 4
Step B â t p B â t p B â t p B â t p
Variable
1
Age
-.47
-.37
-3.61
.001
-.38
-.30
-3.41
.001
-.40
-.32
-3.65
.000
-.28
-.22
-3.64
.001
Education -2.40 -.28 -2.80 .006 -2.40 -.28 -3.26 .002 -2.15 -.25 -2.92 .005 -1.20 -.14 -2.47 .016
2
Anxiety
1.92 .47 5.37 .000 1.65 .40 4.32 .000 .90 .22 3.09 .003
3
MCQ-30 Total
.20 .18 1.87 .066 .00 .00 .05 .957
4
Negative
Beliefs Positive Beliefs
Rumination
1.04
.36
4.06
.000
1.11 .28 4.60 .000 52 .18 1.86 .067
Note. Clinical sample N = 88. Criterion variable = NAS Total. ∆R² values for each step are .21 for step 1 (p < .000), .22 for step 2 (p < .000), .03 for
step 3 (p = .066), .34 for step 4 (p < .000). For the final model, adjusted R² = .77, F(7,79) = 38.75 (p < .000). With Trait anger (STAXI-2) as criterion
variable the R² change values for each step were ∆R² = 0.17 (p = .001), ∆R² = 0.11 (p = .001), ∆R² = .02 (p = .116), and ∆R² = .26 (p < .000).
When the STAXI-2 trait Anger was substituted as the anger self-report criterion, the
results were similar. In the final model, the adjusted R² = 0.52, F(7.79) = 13.2 (p < .001) and the
change in R² associated with the MCQ-30 Total was not significant (p = .852). MAQ Negative
Beliefs (p = .022) and the Positive Beliefs (p = .002) were significant, as well as age (p = .001).
Discussion study 3
The MAQ was designed to assess metacognitive beliefs and processes in relation to
anger. In the present study, the psychometric properties of the revised MAQ were evaluated. The
MAQ, along with a battery of questionnaires assessing a broad range of dimensions of anger,
depression and anxiety, were given to a clinical sample of mixed inpatients and outpatients.
The exploratory factor analysis suggested the same three-dimensional structure as in
Studies 1 and 2. After the initial latent structure was examined using the EFA, factor analyses were
conducted to confirm the factor structure from the EFA. The results indicated that the first model
with 26 items had an acceptable fit to the clinical data when the model had a three-factor structure.
However, after eliminating 4 items with high residual variance not accounted for by the model, a
Chapter 3 Clinical patients, anger, and the MAQ
107
22-item model better fit the data and had excellent internal reliability (alpha coefficients ranging
from .84 to .91).
Regarding background variables, the present study replicated the established findings
that age and education are negatively correlated with anger. Intuitively, it makes sense that as we
grow older, we gain more knowledge and our skills for regulating anger improve.
Regarding symptoms of emotional distress, the present sample had similar anxiety and
depression scores as a Dutch clinical sample that had been assessed using the HADS. This signified
a clinical level of distressing emotional symptoms in the present sample. The clinical level of
distress felt by these acutely ill patients may explain why this sample had a higher score on the
arousal subscale of the NAS than the scores reported for Swedish prisoners by Lindqvist et al.
(2005). As such, the higher arousal levels on the NAS in this clinical sample may not only reflect
specific anger arousal but also general arousal related to additional symptoms.
Regarding the mean score on the MCQ-30, in support of the general metacognitive
model proposed by Wells et al. advocating that clinical samples have higher scores on the MCQ-30,
the present study found significantly higher means on the MCQ-30 for the clinical sample than the
general population in the UK and for the male prisoners from Study 2.
The first hypothesis stating that anxiety, depression and anger would be positively
correlated was only partially confirmed. This study did not replicate the reported relationship
between depression and anger because the correlations between the HADS Depression and the NAS
Total and Trait anger (STAXI-2) were nonsignificant. The correlation between the HADS
Depression and the MAQ-3 subscales was also nonsignificant, which further signified that
depression was not involved in anger processing. However, the HADS Anxiety showed the
expected moderate to strong correlation with the anger measures and with the MAQ Rumination
and MAQ Negative Beliefs subscales, signifying confirmation of the hypothesis that anxiety and
anger share threat as a common theme. The positive correlation between the HADS Anxiety and the
MAQ Rumination subscales supported the notion that worry and rumination share some important
features and that these features may be related to uncontrollable repetitive thinking that maintains
emotional distress (Smith & Alloy, 2009). According to the metacognitive model, experiences of
anxiety may activate a ruminative process and thus trap the individual in the dysfunctional pattern
of processing labeled the Cognitive Attentional Syndrome (Fisher & Wells, 2009). The fact that
angry rumination was associated with anxious worry overall supported the basic transdiagnostic
Chapter 3 Clinical patients, anger, and the MAQ
108
idea represented by the metacognitive framework. The positive correlation between the HADS
Anxiety and the MAQ Negative Beliefs may indicate that anger and anxiety are functions of threat
perception. Threat constitutes a theme of relevance for both emotions, and it may prove fruitful to
investigate how the MAQ Negative Beliefs may be involved in this. The metacognitive
conceptualization proposes that perception of threat may activate anxiety; in turn, rumination may
be used as an attempt to control anxiety, which unfortunately has the unintended side-effects of
strengthening negative beliefs about rumination and maintaining emotional distress (Papageorgiou
& Wells, 2003; Papageorgiou & Wells, 2001b; Papageorgiou et al., 2001). In terms of anger that is
activated by an unspecified perception of threat, the presence of the metacognitive belief that anger
is protective may cause the activation of rumination, which maintains arousal and strengthens
negative beliefs about the uncontrollable and dangerous nature of anger.
The second hypothesis concerned the MAQ as a metacognitive measure. The
significant subscale intercorrelations signified the expected relationship between the subscales of
the MAQ. In particularly, the MAQ Rumination and MAQ Negative Beliefs subscales were highly
correlated, r = .75, indicating that rumination and negative beliefs are closely associated. The
concurrent validity of the MAQ as representing a metacognitive construct was supported by the
moderately positive correlations ranging from non-significant to r = .45 with the general
metacognitive measure the MCQ-30. The correlations were generally in the same direction as in
Study 2, substantiating the relationship between the MAQ and the MCQ-30. The fact that the MAQ
Rumination and the MAQ Negative Beliefs were positively correlated with the MCQ-30
Uncontrollability and Danger subscale supports the similarity of their subscale content.
However, the fact that the MCQ-30 Need to Control thoughts subscale showed
nonsignificant correlations with the MAQ Rumination and the MAQ Negative Beliefs subscales
may indicate that general themes about thought control, superstition and punishment, as reflected in
the MCQ-30 Need to Control thoughts, is less relevant to anger. This assumption was supported by
a less significant relationship between the anger measures on the MCQ-30 Need to Control thoughts
than on the MCQ-30 Uncontrollability and Danger.
The high correlations between subscales measuring themes of uncontrollability,
danger and madness as they relate to the thinking process, on both of the metacognitive measures
indicates the importance of these themes in a metacognitive framework. While the MCQ-30
measures these themes in relation to worry, the MAQ measures these themes in relation to anger.
Chapter 3 Clinical patients, anger, and the MAQ
109
The results emphasize the principal influence of these themes in cognitive control and processing in
psychopathology as evidenced by the contemporary research on the metacognitive model that was
discussed in the introduction.
The positive correlation between the MCQ-30 Cognitive Confidence and the MAQ
Negative Beliefs and Rumination subscale was puzzling. To speculate, this may also reflect the
importance of uncontrollability as a general theme. If an individual experiences a generally low
confidence in the ability to control his or her own memory, the experience of uncontrollability in
relation to the thought processes may easily result. When an individual experiences uncontrollable
emotional processing, a decrease in confidence about one's own cognitive functioning seems
understandable. As such, the key theme involved in the associations across Negative Beliefs,
Rumination and Cognitive Confidence is uncontrollability.
Regarding anger, it was stated that general metacognitive themes concerning
experiences and beliefs about uncontrollability, danger, and madness are centrally involved in the
monitoring, regulating and control of mental phenomena in emotional disorders, including anger
processing. This hypothesis addressed the transdiagnostic features of the metacognitive view. This
was supported by the significant, positive correlations for the MAQ Rumination, MAQ Negative
Beliefs, MCQ-30 Uncontrollability and Danger, and MCQ-30 Need to Control thoughts with anger
measures (NAS and STAXI-2); one exception was the correlation between the MCQ-30 Need to
Control thoughts and the STAXI-2 Trait, which was nonsignificant. Thus, themes of
uncontrollability, danger and madness were instrumental in positioning dysregulated anger within a
metacognitive framework.
The finding that the MCQ-30 Cognitive Confidence was significantly and positively
correlated with the NAS Total, r = .41, and with the STAXI-2 Trait, r = .31, indicated that low
confidence in one's own cognitive functioning influences the ability to regulate anger. This may,
according to the SIP theory, be particularly relevant to the later stages of cognitive processing in
which goal selection is influenced by the evaluation of personal success in implementing certain
strategies. However, in the metacognitive view, this association is explained by the beliefs people
hold about their own memory functions as they relate to selecting worry as a cognitive control
strategy. Because worry is connected to vulnerability to feeling uncertain about future events, it
seems intuitive that low trust in metacognitive efficiency, such as low trust in one's own memory,
would be related to the tendency to worry to prepare for future problems. For example, if a person is
to organize a social event but is uncertain about their ability to remember all the necessary things,
Chapter 3 Clinical patients, anger, and the MAQ
110
this uncertainty may give rise to repetitive worrying about the event. At first glance, this proposed
mechanism accounts for the link between trusting one´s own memory and the tendency to engage in
repetitive thinking, but it does not appear to relate to anger in the same manner. However, if beliefs
about the quality of one´s own memory reflect the actual quality of metacognitive efficiency or
general executive functioning, this may explain the relationship between deficits in executive
functioning and anger problems (Fernandez-Duque et al., 2000). Individuals with superior executive
function are more successful at regulating anger, and thus, they present lower anger scores than
individuals with poor executive functioning. In this way, the proposed relationship between beliefs
about cognitive efficiency and choice of processing strategy, as proposed by the S-REF model, may
not be the mechanism responsible for the relationship between the MCQ-30 Cognitive Confidence
subscale and anger. In support of this notion, Papageorgiou and Wells (2003h) have suggested that
Cognitive Confidence may be conceived of as a by-product of depression, however, in suggesting
this, they also claimed that metacognitive efficiency contributes to the unhelpful metacognitive
beliefs that drive maladaptive processing strategies. Another possibility is that some features
captured in the MCQ-30 Cognitive Confidence subscale have implications for problematic anger in
their own right. As mentioned earlier, this finding may be of clinical significance if explored in
more detail.
The fourth hypothesis concerned justification and hostility as cognitive aspects of
anger. It was hypothesized that the MAQ Positive Beliefs about anger would serve as the higher-
order cognitive structure that facilitates the on-line tendency to interpret social events as hostile and
to justify angry responses, traits that are well-known in patients with clinical anger. This hypothesis
was confirmed by the positive correlation between the MAQ Positive Beliefs and the NAS
Cognitive, r = .63, and by correlations among the specified domains of the NAS Cognitive subscale
regarding justification for and the necessity of becoming angry; items reflecting hostile attitude
were also correlated, including statements about confrontation, as were items reflecting the need to
"watch out" to avoid being hurt by other people. Looking at the content of the NAS Cognitive in
more detail, the positive correlations between the MAQ Positive Beliefs and the NAS Cognitive
may underline how viewing anger as a problem-solving strategy for dealing with perceived
unpleasantness, adversity, danger and ill-will form a cognitive network that increases the risk of an
anger-related responses. This is consistent with the SIP models on anger and aggression in which
cognitive schemata and aggressive scripts are formed. Positive Beliefs about anger resemble both
aggressive scripts and instrumental beliefs about aggression as a helpful means to pursue a goal,
Chapter 3 Clinical patients, anger, and the MAQ
111
views which been associated with self-reported aggression (Archer & Haigh, 1997a; 1997b). Also,
`beliefs about venting anger´ reflects the assumption that outward expression of anger would
regulate mood, which Bushman, Baumeister, Roy and Phillips (2001) found to be associated with
aggression, and shares some common features with the MAQ Positive Beliefs about anger.
The fifth hypothesis concerned the relationship between anger inhibition and the
MAQ Rumination. First, the convergent validity of the MAQ Rumination subscale was
substantiated by the significant and positive correlations with the Anger Rumination Scale (ARS)
and the rumination items on the NAS Cognitive. Moreover, the idea that when anger is not
expressed there is increased risk of being caught in a ruminative process was confirmed by finding
that the Anger Expression Inward (STAXI-2-AX-in) was significantly and positively correlated
with the MAQ Rumination (r = .52).
Moreover, due to a significant positive correlation between the MAQ Negative Beliefs
and STAXI-2-AX-in, the results indicate a link between negative beliefs about anger and the
tendency to withhold the expression of anger. This result is related to the findings by Gilbert et al.
(2004) that show people restrain their anger due to negative beliefs about the consequences of
expressing anger (e.g., fear of rejection by others, fear of losing control, fear of harming others).
The sixth hypothesis concerned the physiological arousal of anger and its association
with the MAQ. Because bodily arousal is a core characteristic of anger and the primary target for
anger control, this is a potentially substantial clinical contribution of the metacognitive framework.
First, it was hypothesized that negative evaluations of anger, possibly based on prior experiences
with dysregulated anger that resulted in angry outbursts followed by negative consequences, were
related to increased arousal and impulsive reactions. This was confirmed by the significant positive
correlations between the MAQ Negative Beliefs and NAS Arousal subscale. These results support
the idea that prior experiences with anger lead to the development of particular beliefs regarding
anger that may influence future cognitive processing. Regarding negative beliefs and bodily arousal,
an associated concern is believing that controlling the anger experience is not possible. In this
manner, prior experiences with anger guide future processing of anger arousal, representing a core
feature of the metacognitive position in which cognitive structures relating to mental phenomena
and experiences guide both cognitive processing and the selection of processing strategies.
Regarding the physiological aspects of anger in relation to the MAQ, it was
hypothesized that rumination not only maintains emotional distress but also elevated bodily arousal.
This hypothesis is supported by the positive correlations between the MAQ Rumination and the
Chapter 3 Clinical patients, anger, and the MAQ
112
NAS Arousal and, more specifically, the duration items of the NAS Arousal. This finding is in
accordance with earlier studies of rumination and blood pressure (Gerin et al., 2006; Hogan &
Linden, 2004). Furthermore, this finding supports the residual transfer theory, which argues that
there is a risk of transferring residual arousal to new situations and thus lowering the threshold for
anger-related reactions.
The seventh hypothesis, which stated that the MAQ is a construct involved in
dysregulated anger, would be supported by significant negative correlations of the MAQ with anger
regulation measures. This hypothesis was partially supported because the NAS Regulation showed
the expected correlation with the MAQ Rumination, however, nonsignificant correlations between
the MAQ Positive Beliefs and Negative Beliefs also emerged. Regarding the STAXI-2 Control
subscale, the MAQ Rumination and the MAQ Negative Beliefs showed the expected correlations
whereas the correlations with the MAQ Positive Beliefs were nonsignificant. One cause of the
nonsignificant results for the STAXI-2 Control may be the low anger control scores in the present
sample compared to the Swedish inmates. The unexpected nonsignificant results involving the NAS
Regulation may question the validity of self-report of anger regulation, as discussed in chapter 4. In
Doyle and Dolan (2006), no significant differences were found for the NAS Regulation when
comparing violent and non-violent inpatients, possibly indicating validity problems with the self-
report of anger regulation skills among high-anger individuals. Additional evidence for a
relationship between the MAQ subscales and dysregulated anger came from the STAXI-2
Expression Out subscale, which measures the tendency to act on angry emotions. Overall, the
present study shows substantial evidence that all of the MAQ subscales are associated with
dysregulated anger.
In relation to general metacognition and anger regulation, the relationship between the
MCQ-30 and the NAS Regulation demonstrates the limitations of using the unmodified S-REF
framework for anger. In principal, the S-REF framework should predict negative correlations
between the MCQ-30 subscales and measures of anger regulation. However, the finding that the
MCQ-30 Positive Beliefs subscale and the Cognitive Self-Consciousness subscale were positively
correlated with anger regulation (NAS Regulation) indicates that the more positive beliefs about
worry, the higher the anger regulation and the more self-focused attention towards higher anger
regulation. This is an interesting finding because the results are inconsistent with the S-REF model.
Chapter 3 Clinical patients, anger, and the MAQ
113
Because these results did not support the unmodified application of the MCQ-30 framework for
anger, they support the usefulness of developing a metacognitive framework specifically for anger.
This finding also suggests that self-focus may be helpful for anger related problems. This would
certainly be an area worth investigating in future studies to gain more knowledge about the specific
types of self-focus that may be helpful for anger problems. These results are consistent with other
evidence suggesting that not all persistent internal focus is unhelpful (Watkins, 2008) and support
the present interest in the research literature on separating functional internal focus from
dysfunctional inner focus.
A fundamental hypothesis of this research was concerned with the unique benefits of
the MAQ as a metacognitive measure specific to anger. Thus, it was predicted that in a hierarchical
regression, the MAQ subscales would account for a significantly greater amount of variance in the
criterion variables, the STAXI-2 and the NAS, than the MCQ-30 Total. Given that entering the
MAQ subscales after the MCQ-30 Total in the regressions eliminated the modest effect of the
MCQ-30 Total anger, this hypothesis was confirmed.
To conclude, both positive and negative beliefs support the general metacognitive
ideas proposed by Wells (2000; Wells & Matthews, 1994) in which positive and negative beliefs are
involved in dysfunctional processing in emotional disorders. In the present study focused
specifically on anger, both positive and negative beliefs appear to give rise to the dysfunctional
rumination process in anger. It may be that these different types of beliefs are involved in the
rumination process at different points in time. More specifically, these results may indicate that
positive beliefs about anger increase the risk of responding with anger in a situation of provocation
and also increase the risk of getting caught up in rumination about the anger. If individuals with
positive beliefs about anger experience anger, they are likely to activate rumination as a coping
strategy. However, as rumination does not modify negative affect but rather increases bodily
arousal and decreases flexible thinking, experiencing the uncontrollability of anger increases related
negative beliefs and lowers the ability to regulate anger effectively. As such, rumination may be
viewed as an automatic cognitive process consistent with the SIP model, but it may also represent a
strategy for coping with a threatening situation. These proposed mechanisms of interactions are
consistent with the metacognitive model of depression proposed by Papageorgiou and Wells (2003).
The present study has several limitations. A substantial weakness of the factor
analyses conducted is sample size; the ratio was 3.4 participants per item. However, the fact that the
Chapter 3 Clinical patients, anger, and the MAQ
114
structure was reproduced earlier in three different samples supports its stability. Furthermore,
adopting a cross-sectional design to evaluate causal relationships was not possible. In addition, the
study used only self-reports of anger, which may have compromised its validity due to the various
issues outlined in the discussion on assessment in the introduction of this thesis. Future studies
should evaluate the questionnaire using a sample of individuals with psychopathology, explore
anger problems using a longitudinal design and use observational data in addition to self-report.
Chapter 4 Forensic patients, anger, aggression, and the MAP
115
Chapter 4 Forensic patients, anger, aggression and the MAP
Introduction
In the fourth study, a population characterized by psychopathology and anger
problems was chosen. The aim of this study was to test the psychometric properties of the revised
measure and, in particular, to evaluate the validity of the measure as being anger- and aggression-
related.
Prior to Study 4, the MAQ-3 was revised based on the results of Study 3. The
Cognitive Consciousness subscale had not performed as expected and had also been tested in
different samples, therefore this subscale was not included in the present study. Because the
composition of the MAQ had shifted substantially away from the metacognitive framework of
Wells and Matthews (Wells, 2000; Wells & Matthews, 1994), it was renamed to indicate its proper
characteristics. The result was the Metacognitive beliefs and Anger Processing (MAP) scale. At this
point, 2 subscales, the Positive Beliefs and Negative Beliefs, directly reflected the original
metacognitive framework and the Rumination subscale likewise echoed features that were present
in the original metacognitive framework. However, neither the Cognitive Confidence subscale, the
Cognitive Self-Consciousness subscale, nor the Negative Beliefs Need to Control thoughts were
represented in the MAP framework. Instead, another thought control strategy was incorporated that
based on the literature may have the potential to illuminate mechanisms of anger dysregulation.
Hence, prior to study four, a dimension of thought suppression that was not included in the S-REF
framework was included in the MAP framework. The suppression subscale was modeled on the
WBSI and details can be found under measures.
In the MAP, metacognitive features revolve around the following three principles:
o A focus on inner triggers and inner outputs as opposed to external triggers and external
outputs or an unspecified combination of these. Metacognition starts within the cognitive
system. Metacognition refers to beliefs about the nature and function of
cognitions/emotions3 as well as mental strategies aimed at controlling cognition in an
attempt to achieve a desired mental state.
3 Some will argue that metacognition should strictly refer to cognitions about cognitions and not about emotions.
However, this distinction may have more academic than clinical relevance. Through associative networks, thoughts as well as emotions about a specific theme form a comprehensive experience within the individual. In the clinical world,
Chapter 4 Forensic patients, anger, aggression, and the MAP
116
o Interactions between different levels of the cognitive system. For example, the attribution of
an external social event, also known as an automatic negative thought (e.g., `he is treating
me unfairly´), arises on the meta-level and metacognitions guide the appraisal of the event's
significance and selection of a coping strategy for the automatic thought. An example of this
is, `anger is necessary to get by in the world´ as a belief structure underlying anger that
arises due to an initial attribution (the automatic thought). In addition, this latter process
drives increased attention to anger, ultimately manifesting itself as anger rumination.
o The interaction between different types of cognitive activity such as structures and
processes.
Adopting a longitudinal design with observational data on aggression, the predictive
value of the MAP was investigated. More complex examination of the relationship between
metacognition and anger was accomplished using structural equation modeling.
Setting
The study was conducted at the forensic psychiatric unit of the Mental Health Centre
Sct. Hans in Denmark. The unit has 80 beds and low-, medium- and high-security levels. Patients
are admitted under psychiatric orders imposed by court for having committed a serious offense and
being unfit to endure punishment because of severe psychopathology. The study was approved by
the Danish Data Control system.
Participants
Data collection was carried out from February 2010 to October 2010 with a follow-up
for the last patient in February 2011. The follow-up period was thus 5 months for each individual.
All Danish-speaking male patients admitted to the facility during the study period were invited to
participate. Participants with mild to moderate organic problems were included because organic
problems are prevalent in this population. However, patients with severe organic problems were
excluded. Females were excluded because they were limited in number. Assessment of patients in
acute states of illness (e.g., vivid hallucinations or in seclusion) was avoided. A total of 54 patients
volunteered to participate. No patients dropped out of the study. Analysis of the patients not willing
to participate was not possible. With written permission from the patient, background variables
were collected from the patient's hospital file by the primary researcher. The mean age of the
people do not make sharp distinctions in their heads about what constitutes an emotion and what constitutes a thought, and therefore cognition about cognition as well as about emotions must be relevant.
Chapter 4 Forensic patients, anger, aggression, and the MAP
117
patients was 36.4 years (SD = 11.9, range 19-67). Characteristics of the sample are displayed in
Table 1.
Table 1. Characteristics of the forensic sample, N = 54, and of the Pedersen (2009) sample, N = 81.
Present study (N = 54) Pedersen 2009 (N = 81)
Characteristic No. % aggregated % %
Born outside Denmark 24 44.4 47
Education
Not finished compulsory school 21 38.9 80 88
Compulsory school only 22 40.7
Some form of education (practical
or scholastic)
11 20.4
Living arrangement
No partner 51 94.4
Employment
Regular job 1
Temporary social benefit 20 36.8
Retired due to mental problems 33 61.1
Diagnosis
Schizophrenia spectrum 44 81.5 79
Bipolar disorder 5 9.3
Other 5
Co-morbid substance abuse 43 79.6 68
Criminal history
First conviction 6 11.1
1-5 previous convictions 20 37.1 89 96₂
>5 previous convictions 28 51.9
Crime type, violent offence₁ 53 88.9 96₃
Homicide attempt or brutal nature 15 27.8
Other violence 33 61.1
Drug related 5 9.3
₁Violent offense was characterized as any offense consisting of actual physical contact or threats of violence; violence
of a brutal nature is a term supplied by the court for more severe violence. ₂ On average, the patients had been
previously sentenced 13 times. ₃.Previous violence.
Other research studies were conducted on the entire population of patients discharged
from the forensic psychiatric unit of the Mental Health Centre Sct. Hans during the years 2006-2007
Chapter 4 Forensic patients, anger, aggression, and the MAP
118
(N = 132, 125 males and 7 females) (Pedersen, 2009) reported detailed characteristics of the
sample. The sample size in this dataset was reduced to 81 before analysis for several reasons4. The
average age of the patients was 35.7 years (SD = 10.49, range 18-62).
Thus, the present study sample was considered comparable to the other study samples
at this institution.
Measures
The Metacognitive beliefs and Anger Processing (MAP) consisted of the same items
as the MAQ-3 except that the cognitive consciousness subscale was omitted and a subscale
designed to measure suppression was developed. The subscale was modeled on the framework
behind the White Bear Suppression Inventory (Wegner & Zanakos, 1994). The suppression
subscale was intended to be a pure measure of mental attempts to suppress anger-related thoughts
and emotions. Therefore, the content of the items was based only on the suppression items of the
suggested 3-factor model of the WBSI (Luciano et al., 2006). These four items focused on
unspecified content of thoughts and used phrases such as, `prefer not to think about´; `try to put out
of my mind´; `try not to think about´; and `try to avoid´. As a result, the suppression subscale of the
MAP consisted of 6 items explicitly focused on thoughts and feelings related to anger and used
phrases such as, "prefer to avoid"; "try to forget"; "important not to think about"; "try to avoid”; “do
not want to attend to"; and "dislike to be reminded of". The name of the scale was altered because
the cognitive consciousness subscale was omitted and the suppression subscale was included in
order to better reflect the content of the scale. As such, the scale was termed the Metacognitive
beliefs and Anger Processing, which was tested in the fourth study. The questionnaire was
developed in both English and in Danish.
Novaco Anger Scale (Novaco, 2003) is a 60-item scale constructed to measure anger
as conceptualization by Novaco (Novaco, 1994). It measures anger in a cognitive domain, an
arousal domain, and a behavioral domain, which together comprise the NAS Total score. The scale
also has a separate anger regulation subscale.
The Hospital Anxiety and Depression Scale (HADS;(Zigmond & Snaith, 1983). This
instrument is a 14-item self-report questionnaire measuring anxiety and depression. The respondent
provides ratings that reflect their most recent week. Seven items measure anxiety, and 7 items
4 The discharge was secondary; no risk assessment was provided on the patient or the patient was female.
Chapter 4 Forensic patients, anger, aggression, and the MAP
119
measure depression. Higher scores indicate higher levels of anxiety and depression. It is a reliable
and valid instrument for assessing anxiety and depression in clinical settings as well as in the
general population (Bjelland, Dahl, Haug, and Neckelmann, 2002). The questionnaire was available
in the public domain and was translated and back-translated by a bilingual translator.
Self-harm was measured using the four items on the General Self-Harm
Questionnaire suggested by Gratz (2001). The first item is coded dichotomously and reads, `have
you ever had the desire to hurt or harm yourself in any way´; the second item asks how many times
this has happened; the third item is coded dichotomously and reads, `have you ever acted on this
desire and deliberately hurt or harmed yourself without trying to kill yourself?´; lastly, the fourth
item asks how many times this has occurred. It was stressed that only self-harm without the intent
of suicide was characterized as self-harm. The participants were also asked about suicide attempts;
they were asked how many times, if any, they have tried to commit suicide. A formal translation
procedure was not conducted.
The Posttraumatic stress disorder Check List- Civilian Version (PCL-CV;(Weathers,
Litz, Herman, Huska, and Keane, 1993) is a 17-item posttraumatic stress disorder (PTSD)
assessment instrument related to unspecified past stressful experiences. In a validation study of the
PCL using item response theory, Bliese, Wright, Adler, Cabrera, Castro, and Hoge (2008)
suggested that four items of the PCL validly assess symptoms of PTSD. This 4-item screening tool
uses ratings on a five-point Likert scale to gather information about latent PTSD. Two items
represented the re-experience domain of PTSD, one item represented the avoidance domain and one
item represented the arousal domain. A value of 7 was considered a reasonable cut-off. A formal
translation procedure was not conducted.
The Schedule of Imagined Violence (SIV;(Grisso et al., 2000): This scale was used to
guide how to measure violent thoughts. The SIV consists of 8 questions; the first question assesses
the presence of violent thoughts either at present or previously, and the following 7 questions are
only given to participants who answered the first question affirmatively. The content of these
successive questions relates to recency, frequency, chronicity, type of harm, target focus,
seriousness of harm, and proximity to target. For the present study, two questions were used; the
first question of the SIV was about whether the participants had ever experienced violent
thoughts/fantasies, and if confirmed, the second question about recency and frequency (`when this
has happened and how often it happens´) was asked. In the original measure, people were assigned
either SIV+ or SIV-. Participants were deemed to be SIV+ if they confirmed ever having violent
Chapter 4 Forensic patients, anger, aggression, and the MAP
120
thoughts within the past 2 months. The same categorical criteria were used in the present study. A
formal translation procedure was not used.
The Staff Observation Aggression Scale – Revised (SOAS-R;(Nijman, Muris,
Merckelbach, Palmstierna, Wistedt, Vos et al., 1999) is a form used to register aggressive incidents
on wards. Staff members use the checklist to note the presence of specific characteristics of the
incident that he or she has witnessed. Information collected includes provocation, means, target,
consequences and actions to stop the aggression. On the SOAS-R form, more detailed information
about the aggressive incidents, the number of verbal incidents, the number of incidents involving an
object without direct human contact (e.g., slamming doors or throwing objects not directly towards
another person), the number of incidents involving direct contact with another person (e.g.,
throwing an object towards another person or grabbing another person), and lastly, the number of
incidents involving direct contact that is judged to be more severe and potentially dangerous (e.g.,
an attempt to strangle or the use of) is available. The SOAS-R has shown good inter-rater reliability
(Nijman, Palmstierna, Almvik, and Stolker, 2005). The form had already been translated and
implemented in clinical practice at the Mental Health Centre Sct. Hans and was therefore available
in the patient files prior to data collection. The total number of aggressive incidents, excluding
verbal incidents, for retrospective aggression was adjusted for length of hospitalization. The total
number of aggressive incidents, excluding verbal incidents, for prospective aggression was
registered for each individual for 5 months following the assessment.
In addition, psychotic symptoms (hallucinations, persecutory delusions and non-
persecutory delusions) were recorded from the day-to-day hospital records at the time of the
assessment. The variables were rated by the primary researcher as one of the following: 1 = no
information about hallucinations/delusions; 2 = some indication of hallucinations/delusions present
in the patient's file; 3 = definite indications of hallucinations/delusions present in the patient's file;
and 4 = behavior clearly affected by hallucinations/delusions.
Less than 5% of responses were missing and no respondent was missing more than a
total of 4 items. One item on the NAS was missing 4 values, and one item on the MAP was missing
4 values. The values for the missing items were replaced with the series mean for the item.
Chapter 4 Forensic patients, anger, aggression, and the MAP
121
Procedure
Potential participants were contacted during weekly group meetings or individually
within the wards. Some patients were approached more than once. All participants were assured that
participation in the study was voluntary. It was emphasized that they would not be identified in
subsequent reports, and results would not be used in connection with the forensic psychiatric
system. Participants received both verbal and written information about the study and what types of
information would be collected from their inpatient files.
All participants signed a consent form. Participants received a stipend of 50 DKR to
participate in the study. The researcher had been in prior contact with several of the participants,
and was not completely unfamiliar to most of the patients. The questionnaires were administered
individually in a private room, and the researcher read the questionnaire aloud to the participant.
Data on prospective aggressive incidents was gathered on each participant in the course of a 5
months follow up.
Hypotheses
Hypothesis 1: The MAP will be significantly correlated with the HADS Anxiety and the HADS
Depression.
Although the hypothesis regarding depression was not confirmed in the previous study
(Study 3), based on other evidence suggesting an relationship between anger and affective
symptoms, it is expected that anxiety and depression will be positively and significantly correlated
with the MAP and the NAS.
As in Study 3, based on the idea that threat is a common theme in anger and anxiety, it
is hypothesized that the MAP Negative Beliefs subscale will show a significant positive correlation
with the HADS Anxiety subscale. Finally, because rumination and worry have been conceived as
similar concepts, the MAQ Rumination scale reflecting uncontrollable rumination is expected to be
significantly and positively correlated with the HADS Anxiety subscale.
Hypothesis 2: The MAP Suppression will be significantly correlated with the MAP Negative
Beliefs, the MAP Rumination and with the NAS Total subscales.
Based on the theoretical assumption that negative beliefs about anger will motivate an
individual to withhold expression of anger, which was supported in Study 3 with the positive
correlation between the MAQ-3 Negative Beliefs and the STAXI-2-AX-in subscales, anger
Chapter 4 Forensic patients, anger, aggression, and the MAP
122
suppression is expected to be associated with negative beliefs about anger. This association would
be supported by a significant, positive correlation between the MAP Suppression and the MAP
Negative Beliefs subscales. To test the assumption that failed suppression may activate rumination,
as discussed in chapter 3, the MAP Suppression is expected to be significantly and positively
correlated with the MAP Rumination subscale.
Furthermore, because suppression may cause an increase in thoughts intended to
suppress, the suppression of anger-related thoughts is expected to be significantly correlated with
anger. Therefore, the MAP Suppression and the NAS Total are expected to be significantly,
positively correlated.
Hypothesis 3: The MAP Negative Beliefs and the MAP Rumination subscales will be significantly
correlated with the anger arousal criteria (NAS Arousal).
In chapter 1, bodily arousal was asserted to be an important facet of the anger
experience. As in Study 3, negative evaluations from the MAP Negative Beliefs subscale are
expected to be significantly correlated with experiencing anger as uncontrollable, as reflected in the
NAS Arousal subscale. Moreover, rumination has been discussed for its associations with
physiological arousal and anger, which confirmed in Study 3 via a relationship between the MAQ
Rumination subscale the NAS Arousal subscale. More specifically, the MAQ Rumination subscale
showed a positive correlation with the NAS Arousal duration item set, indicating that rumination
prolongs physiological arousal. Given that these results are reproduced in the present forensic
sample, this hypothesis is substantiated.
Hypothesis 4: The MAP Rumination subscale will be significantly correlated with the NAS
Cognitive rumination items.
As in Study 3, because the NAS is a validated anger measure that includes 4
rumination items in its Cognitive Domain subscale, a convergent validity test of the MAP
Rumination scale was conducted and a significant, positive correlation was found.
Hypothesis 5: The MAP will be significantly correlated with the NAS Regulation subscale.
As in Study 3, because the MAP reflects dysfunctional beliefs and processing routines
related to anger, a negative relationship with skills to adaptively regulate anger was expected. Thus,
Chapter 4 Forensic patients, anger, aggression, and the MAP
123
the negative correlations between the NAS Regulation subscale and all subscales of the MAP were
expected.
Hypothesis 6: The MAP will account for a significant amount of the variance in the NAS Total.
The relationship between anger and the MAP was further tested using a hierarchical
regression analyses with forced entry. While controlling for the effects of anxiety and depression,
the MAP subscales are expected to account for a significant amount of variance in the criterion
variable (the NAS Total).
Hypothesis 7: The depressive rumination model proposed by (Papageorgiou & Wells, 2003) will
show an acceptable fit in a structural equation model of the MAP data.
This hypothesis is based on the assumption that when individuals with positive beliefs
about anger experience anger, they are likely to activate rumination as a coping strategy. However,
because rumination does not modify negative affect related to the uncontrollability of the
experience, it strengthens negative beliefs about anger. The depression model proposed by
Papageorgiou and Wells (2003) is expected to fit the MAP data in a SEM model.
Hypothesis 8: Individuals experiencing psychotic symptoms will have significantly higher mean
scores on the MAP and the NAS.
In view of the relationship between psychotic symptoms and anger as discussed in
chapter 2, group comparisons (independent samples t-tests) by psychotic symptom will test the
assumption that subjects with hallucinations have higher mean NAS and MAP scores. Similar
results are expected for delusions and persecutory delusions.
In addition, as discussed in chapter 2, threat perception is involved in anxiety and in
the experience of psychotic symptoms. Individuals suffering from psychosis have been found to
present anxiety symptoms (Huppert & Smith, 2005). As such, a higher mean on the HADS Anxiety
subscale is expected for subjects with psychotic symptoms (hallucinations, delusions, and delusions
of persecution).
Hypothesis 9: Individuals experiencing PTSD symptoms will have significantly higher mean scores
on the MAP and the NAS.
Chapter 4 Forensic patients, anger, aggression, and the MAP
124
In chapter 2, the relationship between anger and PTSD symptoms was discussed.
Therefore, group comparisons by PTSD symptom are expected to show higher means on the NAS
and the MAP scales for subjects scoring above the threshold of 7 on the PCL-CV-4.
Furthermore, because perception of threat is involved in PTSD, higher means on the
HADS Anxiety subscale are expected for subjects scoring above 7 on the PCL-CV-4 in comparison
with subjects scoring below the cut-off.
Hypothesis 10: Individuals who report previous acts of self-harm are expected to have higher means
on the NAS and the MAP than individuals who did not report previous acts of self-harm.
Anger has been found to pre-date deliberate self-harm (Chapman & Dixon-Gordon,
2007). Based on this finding, a relationship between self-harm and anger is predicted. To test this,
group comparisons on the basis of previous acts of self-harm are expected to show higher means on
the NAS and the MAP for subjects reporting acts of self-harm.
Hypothesis 11: Individuals reporting violent fantasies will have a higher mean on the MAP and the
NAS than individuals not reporting violent fantasies.
Given that violent fantasies have been linked to anger (Grisso et al., 2000), subjects
characterized as SIV+ were expected to have higher means on the NAS and the MAP subscales
than SIV- subjects. Because violent fantasies have been described as an elaborative rehearsal
process resembling rumination (Grisso et al., 2000; Nagtegaal et al., 2006), they are a central
criterion on the MAP Rumination subscale. As a test of convergent validity, the SIV+ subjects are
expected to have a significantly higher mean on the MAP Rumination subscale than the SIV-
subjects.
Hypothesis 12: Individuals who show physical aggression, both retrospectively and during the
follow-up period, will have higher means on the MAP and the NAS than individuals who do not
show physical aggression.
Because the positive relationship between anger and aggression in clinical samples
has been substantiated by a large body of research, an important validation of the MAP will be to
demonstrate an association with observed aggression. Therefore, using group comparisons, subjects
showing previous physical aggression during the admission period and subjects showing physical
Chapter 4 Forensic patients, anger, aggression, and the MAP
125
aggression in the ward (as measured by the SOAS-R) during the follow-up period are expected to
have higher means on the MAP and the NAS.
Results
Confirmatory factor analyses
The MAP data met assumptions of normality, permitting a confirmatory factor
analysis of the scale. For the fourth sample of forensic patients (N = 54), in addition to the 26 items
comprising the MAQ-3 that was tested in the clinical sample, a new subscale of 6 items was added.
The new subscale, Suppression, was constructed to measure mental attempts to suppress anger-
related thoughts and emotions (see measures). The resulting scale was labeled the MAP and
consisted of 32 items on 4 subscales; Positive Beliefs (6 items); Negative Beliefs (8 items);
Rumination (8 items); and Suppression (5 items).
To confirm the factor structure suggested by the exploratory factor analysis, the data
was fitted to a Confirmatory Factor Analysis that was performed using the M-plus statistical
software, version 6 (L.K. Munthén & B.O Munthén, 2010).
The test provides information about how well a specified structure model explains
the data overall in that particular sample by comparing the expected covariance with the observed
covariance. Also, fit indices other than the test may be used to examine the fit of a latent
variable model. The Tucker-Lewis Index (TLI), the Comparative Fit Index (CFI), the Root-Mean-
Square Error of Approximation (RMSEA) and the Weighted Root-Mean-square Residual (WRMR)
can be used. To indicate a good-fit model, the test should be nonsignificant or the ratio of
divided by the degrees of freedom should be less than 2. Comparative Fit Index values larger
than 0.9 -.95 suggest a good-fit model as do a RMSEA value below 0.05-.06, a WRMR value
smaller than .95 -1.0, and a TLI value larger than .95 (Ching-Yun, 2002; Ullman, 2007).
Using the forensic dataset from 54 patients, we tested the final three-factor model,
with 22 items derived from the CFA, on the clinical sample from Study 3. This model yielded
goodness-of-fit indices of (189, N = 54) = 308.93, p<.000; CFI = .92; TLI = .90; RMSEA = .11;
and WRMR = .96 suggesting a model approaching an acceptable fit. However, because the forensic
dataset included 6 items designed to load on a single subscale (Suppression), a CFA was conducted
on a four-factor model including suppression items. The fit for this model was unsatisfactory:
(344, N =54) = 536.94, p<.000; CFI = .87; TLI = .85; RMSEA = .10; WRMR = 1.2. By allowing
local dependency within items on the same subscale, the goodness-of-fit indices increased to
Chapter 4 Forensic patients, anger, aggression, and the MAP
126
(330, N = 54) = 472.08, p<.000; CFI = .90; TLI = .89; RMSEA = .09; and WRMR = 1.0.
Investigations of the residual covariance suggested a model without item 2, from the suppression
subscale. The goodness-of fit indices for this final model were (294, N = 54) =378.69, p<.001;
CFI=.94; TLI = .93; RMSEA=.07; and WRMR = .88, indicating a model approaching acceptable
fit. This final model comprised the following subscales: positive beliefs (6 items); negative beliefs
(8 items); rumination (8 items); and suppression (5 items). The results of the CFA models are
presented in Table 2.
Table 2. Fit indices for confirmatory factor analytic models of the MAQ
Sample Model 2 df 2/ df p CFI TLI RMSEA WRMR
Forensic (N = 54)
Three-factor (22 items)₁ 308.93 189 1.63 <.000 .92 .90 .11 .96
Four-factor (28 items) 536.94 344 1.56 <.000 .87 .85 .10 1.2
Four-factor (28 items)₂ 472.08 330 1.43 <.000 .90 .89 .09 1.0
Four-factor (27 items)₃ 378.69 294 1.29 <.001 .94 .93 .07 .88
Note: ₁ local dependence between two items within the same subscale was allowed in 17 instances. ₂ local dependence
between two items within the same subscale was allowed in 14 instances. ₃ local dependence between two items within
the same subscale was allowed in 24 instances.
Chapter 4 Forensic patients, anger, aggression, and the MAP
127
Descriptive values and internal consistency
Descriptive statistics and Cronbach´s alpha values for the MAP, the HADS, and the NAS subscales
are presented in Table 3.
Table 3. Descriptive statistics and Cronbach´s alpha values of the MAP, HADS and NAS scales for forensic patients (N
= 54).
To provide a frame of reference for the sample means, t-test comparisons were made
between the NAS means from the present study and those from the Danish NAS norms for mixed
clinical patients (see appendix F). The tests revealed that the means of the present study were
significantly higher on the NAS Cognitive (N = 164; M = 32.0, SD = 5.3)(t = (53) 2.30, p = .025)
and on the NAS Regulation (N = 164; M = 25.4, SD = 3.7)(t = (53) 2.36, p = .022) subscales than
those of the mixed clinical patients.
Comparing the NAS means of the forensic group from the present study with those of
the Danish non-clinical group (see appendix F) revealed that present means for the forensic sample
were significantly higher on the NAS Total (N = 477; M = 75.8, SD = 10.0)(t = (53) 8.15, p = .000);
NAS Cognitive (N = 477; M = 26.6, SD = 3.6)(t = (53) 9.05, p = .000); the NAS Behavioral (N =
477; M = 23.3, SD = 3.8)(t = (53) 7.06, p = .000); and the NAS Arousal (N = 477; M = 25.9, SD =
4.2) (t = (53) 6.55, p = .000). These results indicate high levels of anger in the present sample.
Comparing the means from the forensic group in the present study with those of the
clinical participants in Study 3 showed no differences in the MAP Positive Beliefs or Negative
Beliefs subscales; however, the results indicated that the forensic patients ruminated less than the
No. Forensic patients (N = 54)
MAP Mean SD
Positive beliefs 6 13.7 5.5 .79
Negative beliefs 8 19.5 6.2 .77
Rumination 8 17.1 6.9 .92
Suppression
5 13.2 4.2 .77
HADS
Anxiety 7 7.4 5.0 .84
Depression
7 7.0 4.5 .71
NAS
Total 48 95.3 17.6 .93
Arousal 16 31.4 6.2 .81
Behavioral 16 30.1 7.0 .87
Cognitive 16 33.8 5.9 .79
Regulation 12 26.9 4.5 .78
Chapter 4 Forensic patients, anger, aggression, and the MAP
128
clinical patients in Study 3 (N = 88, M = 19.2, SD = 6.4) (t = (53) 2.20, p = .032). Comparing the
means on the HADS revealed that the clinical sample had a significantly higher mean HADS
Anxiety (N = 88, M = 10.5, SD = 4.6) (t = (53) 4.63, p < .000) and on the HADS Depression (N =
88, M = 8.5, SD = 4.9) (t = (53) 2.50, p = .016).
Background variables and affective symptoms
Age, length of education, and number of earlier convictions were not correlated with
the MAP, NAS or HADS. Correlations between the HADS, MAP, and NAS showed that HADS
Anxiety was positively correlated with MAP Negative Beliefs (r = .35), which confirmed the idea
that threat is a common theme in anger and anxiety. Given that rumination and worry are similar
with respect to the experience of uncontrollability, it was predicted that HADS Anxiety would be
significantly correlated with the MAP Rumination. However, because the correlation was
nonsignificant, this was not confirmed. There was no correlation between the MAP and the HADS
Depression subscales.
Both the HADS Anxiety and HADS Depression subscales were positively correlated
with the NAS Total (r = .54 and r = .43, respectively), which substantiated the prediction that anger
and affective symptoms in are associated in psychopathology. These results indicate that anxiety
was associated with negative beliefs about anger and with anger itself, whereas depression was only
associated with anger.
Chapter 4 Forensic patients, anger, aggression, and the MAP
129
Subscale correlations
Apart from the Suppression subscale, Pearson intercorrelations among the MAP
subscales showed significant results ranging from r = .36 to .68 and from r = .75 to .89 with the
MAP Total. Correlations with the anger measure (NAS) and the metacognitive anger measure
(MAP) were positively correlated at the p<.01 level, apart from results regarding the NAS
Regulation and MAP Suppression subscales. Results are displayed in Table 4.
Table 4. Intercorrelations (Pearson) among the MAP subscales and with the NAS.
MAP Positive beliefs Negative beliefs Rumination Suppression Total
Positive beliefs 1 .36* .68
* ns .82
*
Negative beliefs .46* ns .75
*
Rumination ns .89*
NAS
Cognitive .55* .49
* .62
* ns .68
*
Behavioral .62* .61
* .64
* ns .76
*
Arousal .54* .58
* .68
* ns .74
*
Regulation ns ns -.36* ns ns
Total .62* .61
* .71
* ns .79
*
Note. * p < .01.
The idea that negative beliefs about anger motivate the individual to suppress anger
was not confirmed by the present results because the MAP Suppression and the MAP Negative
Beliefs subscales were not significantly correlated. Moreover, the concept that suppression
increases the thoughts intended to suppress was not confirmed either because correlations between
the MAP Suppression and NAS Total were also nonsignificant. The proposed relationship between
suppression and rumination was not confirmed either; the MAP Suppression and the MAP
Rumination subscales were not correlated.
Because the expected correlation between the MAP Negative Beliefs and NAS
Arousal subscales was found, the notion that the experience of uncontrollability is an important
facet of anger arousal was substantiated. The MAP Rumination subscale showed the expected
correlation with the duration component of the NAS Arousal (r = .63; p < .01) scale, supporting the
assumption that rumination prolongs physiological arousal (as was tested and confirmed in Study
Chapter 4 Forensic patients, anger, aggression, and the MAP
130
3). In support of convergent validity with the MAP Rumination scale, the MAP was significantly
correlated with the rumination items on the NAS Cognitive subscale (r = .64; p < .01).
Given that the MAP reflects dysfunctional beliefs and processing routines, which is
largely confirmed in Study 3, we predicted that the MAP subscales would be negatively correlated
negatively with the NAS Regulation subscale. However, this was confirmed only because
correlations between the NAS Regulation and the MAP Positive Beliefs and Negative Beliefs
subscales were nonsignificant, whereas the MAP Rumination and the NAS Regulation subscales
showed the expected significant negative correlation. This indicated that rumination is inversely
related to anger regulation; the more individuals ruminate, the less able they are to regulate anger.
Concurrent validity
Given that the MAP is related to anger, a hierarchical regression with forced entry
using the NAS Total as the criterion variable will further substantiate this association. When
conducting the regression, because the HADS Anxiety and Depression subscales were correlated
with the NAS Total, they were entered as covariates on the first step. On the second and final steps
of the regression, the MAP subscales were entered. Given that the MAP Suppression had not shown
any significant correlation with the criterion variable, it was not included in the regression.
When HADS Anxiety and Depression were entered on Step 1, their relationships with
NAS Total was significant (R² = .28, p < .000). For the second step in which the MAP subscales
were entered, an additional 42% of the variance in the criterion variable was explained (ΔR² = .42, p
< .000). In that final model, the influence of depression was nonsignificant, while the MAP
subscales and HADS Anxiety were significant predictors of the criterion variable. In the final
model, 70% of the variance in the criterion variable was accounted for (R² = .70; F (5,53) = 25.10, p
< .000). The results are displayed in Table 5.
Chapter 4 Forensic patients, anger, aggression, and the MAP
131
Table 5. Hierarchical regression of anger level (NAS) as associated with anxiety, depression, PTSD symptoms, violent
fantasies and the MAP. Forensic sample N = 54.
Model 1 Model 2
Step B Â t p B â t p
Variable
1
Anxiety
1.61 .45 3.00 .004 1.13 .32 3.13 .003
Depression .53 .14 .89 .380 .09 .02 .22 .826
2
Positive beliefs
Negative beliefs
Rumination
.85 .26 2.51 .016
.70 .25 2.76 .008
.83 .32 2.92 .005
Note: Criterion variable = NAS Total.
In preliminary testing of the seventh hypothesis, which stated that the depressive
rumination model by Papageorgiou and Wells (2003) is adaptable to anger, a Structural Equation
Model (SEM) approach was used in an exploratory exercise. Using an SEM approach to testing
model fit has the advantage of estimating the unique effect of a variable while simultaneously
controlling for the effects of others. Hence, specifying the relationships between the MAP subscales
and self-reported anger (NAS Total) was attainable using the SEM. Furthermore, in SEM models,
measurement error is both estimated and controlled for. M-plus statistical software, version 6 (L.K.
Munthén and B.O Munthén, 2010) was used to test the structural model.
To boost the ratio of cases to variables in the model, only four variables were
included. The model was articulated with a path from Positive Beliefs to Rumination and from
Rumination to Negative Beliefs; and finally, from Negative Beliefs to anger. Positive Beliefs and
Negative Beliefs were specified as intercorrelated. The model displayed in Model 1 was fitted with
the following overall fit indices, 2 (2341, N = 54) = 2570.9, p < .000; CFI = .90; TLI = .90;
RMSEA = .04; WRMR = 1.02.
Chapter 4 Forensic patients, anger, aggression, and the MAP
132
Model 1. The structural model for a structural regression model of metacognition and anger. Estimates are reported as
unstandardized estimate (S.E.) estimate/S.E.(p).
Note. Positive beliefs represent the latent variable comprising 6 observed items; Rumination represents the latent
variable comprising 8 observed items; Negative beliefs represent the latent variable comprising 8 observed items, and
Anger represents the latent variable comprising 48 observed NAS items.
The estimates indicated significant path coefficients for all three specified pathways
and a significant correlation between negative and positive beliefs. The direct effect of Positive
Beliefs on Rumination was appreciable in magnitude (z = 6.955, p <.000) as was the direct effect of
Rumination on Negative Beliefs (z = 5.174, p <.000), and Negative Beliefs on anger (z = 3.522, p
<.000). These results supported the hypothesis that the effects of Positive Beliefs and Rumination
on anger were largely mediated through Negative Beliefs. Hence, the data support the idea that
positive beliefs are involved in the selection of rumination as a processing strategy in response to
anger while strengthened negative beliefs were the byproduct of rumination and seemed to serve a
key function in mediating the relationship between rumination and anger. Overall, these results,
which were conducted on a small sample, provide support for the adaptability of the model
proposed by Papageorgiou and Wells (2003) and provide cross-sectional preliminary support for the
validity of the metacognitive model of anger as outlined in the MAP.
Psychotic symptoms
The case files for the forensic participants indicated that 15 (27.8%) of individuals
exhibited hallucinating experiences; 14 (26.0%) had no-persecutory delusions; and 13 (24.1%) had
persecutory delusions. For nine (16.7%) of the patients, all three types of symptoms were present
simultaneously.
.143(.054) 2.653 (.008)
.723(.104) 6.955 (.000)
.793(.225) 3.522 (.000) Anger
Negative beliefs
Rumination
.496(.096) 5.174 (.000)
Positive beliefs
Chapter 4 Forensic patients, anger, aggression, and the MAP
133
To analyze the relationship between psychotic symptoms and scores on the MAP,
NAS and HADS, a series of comparisons were conducted separately for hallucinations, delusions,
and delusions of persecution. Regarding the MAP, the analyses showed that only MAP Rumination
and delusions of persecution showed the expected relationships, namely that individuals
experiencing delusions of persecution would have a higher scores on the MAP Rumination than
individuals who were not. The predicted relationship between psychotic symptoms and anxiety and
anger was supported; generally, subjects with symptoms of either hallucinations, delusions or
delusions of persecution had higher scores on the HADS Anxiety and NAS Total scales.
Furthermore, subjects with delusions of persecution had higher scores on several of the measures.
The results are presented in Table 6.
Table 6. Mean scores for the MAP, HADS and NAS and independent t-test comparisons divided by type of symptom.
Note. * p <.05; ** p < .01.
Hallucinations Delusions Delusions of persecution
+
N = 15
(SD)
–
N = 39
(SD)
t (52)
+
N = 14
(SD)
–
N = 40
(SD)
t (52)
+
N = 13
(SD)
–
N = 41
(SD)
t (52)
MAP
Positive Beliefs 14.5
(6.1)
13.4
(5.2)
.65 14.5
(5.6)
13.4
(5.4)
.64 15.7
(5.6)
13.1
(5.3)
1.54
Negative Beliefs 21.8
(7.1)
18.6
(5.6)
1.76 20.4
(6.7)
19.2
(6.0)
.63 21.8
(6.1)
18.7
(6.1)
1.57
Rumination 19.4
(7.0)
16.3
(6.7)
1.53 18.9
(7.4)
16.5
(6.7)
1.11 21.3
(6.2)
15.8
(6.6)
2.66**
Suppression 14.1
(3.2)
12.9
(4.6)
.93 12.2
(3.9)
13.6
(4.3)
1.04 13.2
(3.5)
13.3
(4.5)
.07
HADS
Anxiety 10.0
(5.4)
6.4
(4.5)
2.53* 9.7
(5.0)
6.6
(4.7)
2.12* 10.9
(4.9)
6.2
(4.5)
3.21**
Depression 7.8
(5.0)
6.7
(4.2)
.83 8.8
(4.2)
6.4
(4.4)
1.80 10.0
(4.1)
6.0
(4.2)
3.00**
NAS
Total 102.9
(17.4)
92.4
(17.0)
2.02* 103.6
(18.2)
92.4
(16.7)
2.10* 108.7
(15.4)
91.1
(16.2)
3.46**
Arousal 34.7
(5.8)
30.2
(6.0)
2.49* 33.5
(6.2)
30.7
(6.1)
1.47 35.1
(5.8)
30.3
(5.9)
2.55*
Behavioral 32.2
(7.9)
29.3
(6.6)
1.30 33.0
(7.9)
29.0
(6.5)
1.88 35.1
(7.1)
28.5
(6.3)
3.22**
Cognitive 36.1
(5.0)
33.0
(6.0)
1.79 37.1
(6.0)
32.7
(5.5)
2.49* 38.6
(4.8)
32.3
(5.4)
3.70**
Regulation 26.3
(5.0)
27.1
(4.4)
.55 26.8
(4.8)
26.9
(4.5)
-.10 25.3
(4.6)
27.4
(4.4)
1.46
Chapter 4 Forensic patients, anger, aggression, and the MAP
134
PTSD symptoms
Summing the 4 items, thirty-six (67%) patients scored above the cut-off value of 7
points, suggesting that the cut off score was reasonably accurate in identifying a case of PTSD
(Bliese et al., 2008). The expected associations between PTSD symptoms and the MAP, the NAS
and the HADS were largely confirmed; as expected, patients scoring 7 or higher (PTSD+) had
higher scores on the NAS Total, Arousal, and Behavioral, the HADS Anxiety and Depression, and
the MAP Negative Beliefs and Rumination subscales than patients scoring below the cut-off value.
The expected relationships for the MAP Positive Beliefs, MAP Suppression, NAS Cognitive and
NAS Regulation subscales were not found. Results are presented in Table 7.
Table 7. Means, standard deviations and group comparisons for violent fantasies and PTSD symptoms.
PTSD+ (SD)
N = 36
PTSD- (SD)
N = 18
t(52)
MAP
Positive beliefs 14.4 (5.0) 12.3 (6.1) 1.48
Negative beliefs 20.8 (6.6) 16.8 (4.2) 2.37*
Rumination 18.5 (7.1) 14.4 (5.8) 2.14*
Suppression 13.5 (3.8) 12.8 (5.0) .59
HADS
Anxiety 9.0 (4.9) 4.1 (3.4) 3.82**
Depression 7.9 (4.3) 5.2 (4.3) 2.19 *
NAS
Total 99.6 (18.0) 86.7 (13.6) 2.68**
Arousal 33.1 (6.1) 28.2 (5.1) 2.92**
Behavioral 31.7 (7.4) 26.9 (5.0) 2.45*
Cognitive 34.9 (5.8) 31.7 (5.5) 1.96
Regulation 26.4 (4.5) 27.8 (4.5) 1.10
Note. * p < .05; ** p < .01. PTSD + = PCL-CV-4 above the value of 7 point. PTSD - = PCL-CV-4 below the value of 7
point.
Self-harm
Twenty-five (46%) of the forensic patients confirmed that they had experienced a
desire to hurt themselves on at least one occasion, and 23 (43%) reported that they had acted on this
desire. Group comparisons of the patients who had experienced thoughts or actions of self-harm and
patients without these characteristics revealed 2 significant results on the MAP, HADS and NAS
subscales. The patients who reported that they had at some point in their lives acted on the desire to
hurt themselves had higher scores on the HADS Anxiety (t (52) = 2.21, p = .032) and on the MAP
Negative Beliefs (t (52) = 2.08, p = .043) than patients who had not engaged in self-harm. These
results do not confirm an association between anger and self-harm; however, the results indicate
that anxiety is involved in self-harm because patients reporting thoughts or actions of self-harm had
Chapter 4 Forensic patients, anger, aggression, and the MAP
135
higher scores on HADS Anxiety than patients without prior self-harm. To speculate, because the
patients with self-harm had higher scores on the MAP Negative Beliefs than patients without self-
harm, fear of expressing anger may provoke self-harm.
Suicide attempts
Thirteen patients (24%) reported having attempted suicide. No difference in the mean
scores of the MAP, HADS or NAS between patients reporting suicide attempts and patients
reporting no previous suicide attempts was found.
Violent fantasies
One patient refused to answer questions related to the presence of violent fantasies.
Thirty-one (57%) of the forensic patients confirmed having had violent fantasies at some point, and
twenty-four (44%) confirmed having had violent fantasies within the last 2 months. Following
Grisso et al. (2000), these patients were labeled SIV+. Given that the proportion of SIV+ in the
Grisso et al. study was 30%, the present study had a significantly higher prevalence of violent
fantasies. This is noteworthy because the Grisso et al. study was large and included more than 1100
patients from three U.S. metropolitan areas. Patients who were SIV+ had higher mean scores on all
subscales of the NAS, MAP and HADS than patients who were not SIV+. In particular, the MAP
Rumination was strongly associated with violent fantasies, which supports the notion that these
constructs are comparable. In addition, the NAS Cognitive subscale was strongly correlated with
violent fantasies, which parallels the Grisso et al. findings and those from Study 3 of this thesis.
Results are displayed in Table 8.
Chapter 4 Forensic patients, anger, aggression, and the MAP
136
Table 8. Means, standard deviations and group comparisons for violent fantasies and PTSD symptoms.
SIV+ (SD)
N = 24
SIV- (SD)
N = 29
t(51)
MAP
Positive beliefs 15.8 (4.9) 12.0 (5.5) 2.71**
Negative beliefs 21.3 (6.3) 18.0 (5.8) 1.92
Rumination 21.0 (6.0) 13.9 (6.0) 4.27**
Suppression 13.5 (4.9) 12.8 (3.5) .58
HADS
Anxiety 8.9 (5.3) 6.0 (4.4) 2.16*
Depression 8.3 (4.6) 5.8 (4.1) 2.05*
NAS
Total 105.5 (14.5) 86.5 (5.5) 4.60**
Arousal 34.6 (4.8) 28.6 (6.0) 4.00**
Behavioral 33.1 (7.0) 27.3 (6.0) 3.38**
Cognitive 37.6 (5.2) 30.6 (4.4) 5.28**
Regulation 27.9 (4.4) 25.9 (4.4) 1.66
Note. * p < .05; ** p < .01. SIV + = violent fantasies within the past 2 months. SIV - = No violent fantasies within the
past 2 months.
Aggression
To analyze the association between acts of aggression and metacognition, anger,
anxiety and depression, the patients were categorized as Agg+ if they had experienced at least one
episode of aggression involving an object or direct physical contact and Agg- if they had not. As
such, verbal aggression did not qualify for labeling the individual as Agg+. Verbal aggression was
excluded because the reports of these incidents were less valid than incidents involving an object or
aggression towards another person with direct contact. The patients were labeled either Agg+ or
Agg- for retrospective aggression and aggression during the follow-up period.
In comparing the mean scores for the MAP and the NAS based on aggression, four
cases were omitted because they had either stayed at the ward for less than 30 days prior to testing
or were only available for a follow-up period of less than 30 days. In addition, one outlier was
excluded. Forty-nine cases remained in the analysis of aggression.
Hypotheses regarding aggression and anger were partially confirmed. There were
significant differences in the mean scores of Agg+ and Agg- patients for some, but not all,
subscales. Results indicated that the MAP Negative Beliefs, the NAS Total, the NAS Arousal and
the NAS Cognitive scales were associated with aggression. As such, at the time of the assessment,
the subjects who had been involved in prior aggressive acts had higher scores on the MAP Negative
Beliefs and on the NAS Total, Arousal and Cognitive scales. Likewise, subjects engaging in
Chapter 4 Forensic patients, anger, aggression, and the MAP
137
aggressive behavior in the follow-up period had higher scores on these subscales. The results are
presented in Table 9.
Table 9. Group comparisons of self-report anger scales and depression/anxiety scores divided by aggression.
Retrospective
Follow up
Agg- (SD)
N = 27
Agg+ (SD)
N = 22
t (47) Agg- (SD)
N = 32
Agg+ (SD)
N = 17
t (47)
MAP
Positive beliefs 12.5 (5.4) 14.5 (5.5) 1.32 12.6 (5.3) 15.0 (5.5) 1.51
Negative beliefs 16.9 (5.3) 21.2 (6.3) 2.64* 17.5 (6.1) 21.4 (5.4) 2.20*
Rumination 15.2 (6.8) 18.7 (6.3) 1.89 15.9 (6.9) 18.5 (6.1) 1.34
Suppression 13.2 (4.7) 13.2 (3.7) .03 12.6 (4.4) 14.4 (3.8) 1.41
NAS
Total 89.3 (16.2) 99.9 (17.6) 2.20 * 89.4 (16.5) 102.8 (16.2) 2.72**
Arousal 29.1 (5.9) 33.0 (5.6) 2.33* 29.1 (6.0) 34.2 (4.8) 3.02**
Behavioral 28.3 (6.3) 31.6 (7.8) 1.59 28.3 (7.0) 32.5 (6.7) 1.98
Cognitive 31.8 (5.6) 35.3 (5.7) 2.17* 31.9 (5.2) 36.1 (6.1) 2.52*
Regulation 27.9 (4.5) 25.9 (4.3) 1.64 26.7 (4.4) 27.5 (4.8) .58
Note. * p < .05. ** p < .01
The relationships between psychotic symptoms, PTSD symptoms, violent fantasies,
psychopathic traits and aggression were analyzed using a series of independent sample t-tests
separated by category of aggression. There were no significant results.
Discussion
The aim of the present study was to test the psychometric properties of the MAP and
to evaluate the validity of the MAP for anger and aggression. The MAP was administered with
questionnaires assessing anger, PTSD symptoms, violent fantasies, and self-harm. Psychotic
symptoms were recorded from case files, and both retrospective and prospective aggressive acts
were collected from staff reports (with the SOAS-R).
The comparisons of the means on the NAS revealed that the present forensic sample
had higher mean scores on all subscales of the NAS than the non-clinical group and a higher mean
score on the NAS Cognitive subscale compared to mixed clinical patients. These findings support
the argument that this sample was representative of a high-anger population. The higher mean NAS
Regulation score for the present forensic sample compared to the mixed clinical patients and the
normative data may indicate difficulty in monitoring, or it may be caused by social desirability for
the forensic patients to report that they are not having problems regulating their anger. The means
from the forensic patients show the expected similarities with the MAP Positive Beliefs and
Negative Beliefs subscales. However, contrary to expectations, the results indicate that the forensic
Chapter 4 Forensic patients, anger, aggression, and the MAP
138
patients ruminated less than the clinical patients in Study 3. One plausible explanation may be that
the forensic patients behaved aggressively instead of ruminating. The logic of this finding is that if a
patient reacts impulsively when provoked instead of "boiling inside", a potential ruminative process
may be episodic rather than long-lasting. Of course, it is possible to show impulsive aggression and
to simultaneously ruminate. Furthermore, to complicate matters, this result is somewhat
inconsistent with the association between rumination and aggression.
Regarding emotional symptoms, comparing the mean on the HADS subscale with the
mean from the clinical sample in Study 3, the clinical patients had higher levels of anxiety and
depression. This indicates that the forensic patients were less distressed by depressive and anxious
symptoms than the clinical patients.
Regarding the factor structure of the MAP, the CFA with a 22-item, three-factor
model reproduced the good fit seen in Study 3. Testing the four-dimensional structure, which was
based on the residual variance for the 28 items and included 6 new suppression items, led to the
elimination of one of the suppression items for the forensic sample. When fitting the four-
dimensional model with 27 items, the fit indices denoted a good fit to the data and excellent internal
reliabilities ranging from .77 to .92. In conclusion, the factor structure of the MAP was confirmed
and evidence for a reproducible metacognitive framework of anger across sample types was
produced.
The finding that both the HADS Anxiety and Depression subscales were positively
correlated with the NAS Total was in support of the first hypothesis, which predicted an association
between anger and affective symptoms (Posternak and Zimmerman, 2002). The predicted
association between the MAP and anxiety and depression was only partially confirmed by a
significant positive correlation between the MAP Negative beliefs and anxiety, but no other
correlations were found between depression and other subscales. The idea that worry and angry
rumination share the experience of uncontrollability was not confirmed in the present study, as it
was in Study 3; this may be explained by the lower HADS Anxiety and MAP Rumination subscale
means in the present sample. That said, the correlation did approach significance (r = .30, p = .059).
The positive correlation between the HADS Anxiety and MAP Negative Beliefs subscales
supported the notion that threat detection is involved in problematic anger. To speculate, this result
Chapter 4 Forensic patients, anger, aggression, and the MAP
139
may indicate that the perceptions of anger as being related to danger, madness and uncontrollability
are what produce anxiety.
The second hypothesis concerned the new subscale of the MAP, the MAP
Suppression. Overall, the performance of the Suppression subscale was disappointing as none of the
expected associations emerged. The MAP Suppression was expected to be positively correlated
with the MAP Negative Beliefs about anger subscale, which would have confirmed the idea that
negative evaluations of anger motivate the individual to suppress anger. Furthermore, if the MAP
Suppression had shown the expected correlation with the MAP Rumination subscale, it would have
confirmed that when suppression fails, activating rumination as a coping strategy to target negative
affect is a risk. Finally, the paradoxical effect of suppression was not confirmed because MAP
Suppression and the NAS were not correlated. These disappointing results may be explained by the
Suppression subscale being new with no previous validation. This negative finding raises doubt that
the subscale measures the theoretical construct it was intended to capture.
However, the ultimate purpose of thought suppression is to rid the individual of
unwanted thoughts. When successful, the unwanted thoughts disappear; however, for several of the
reasons that were discussed earlier, this process may go wrong and paradoxically increase the
unwanted thoughts. This means that individual differences in the success of suppression may
explain mixed findings in the effect of thought suppression. Distinguishing between successful and
unsuccessful suppression of anger is a challenging task. To begin, both successful and unsuccessful
suppression may not necessarily be differentiated by level of suppression. In theory, a high
suppressor may report high scores when successful, and an unsuccessful suppressor may also report
high scores for having invested a great deal of energy into suppressing unwanted thoughts. The
point is that these different types of high-suppressing individuals could easily show different
relationships with anger and aggression. If high suppression occurs with success, no correlation
with anger would be expected. On the contrary, if high suppression occurs but is unsuccessful, a
correlation with aggression may be present. Hence, issues such as these may be at play in the
present data. When assessing thought suppression, it may be necessary to assess both the individual
tendency and individual ability to suppress thoughts. In addition, because people are motivated to
suppress only unwanted thoughts, and unwanted content for one person may be desired content for
another person, it may prove useful to account for the content of the suppressed thoughts when
measuring thought suppression. Furthermore, as already indicated, because suppression and
Chapter 4 Forensic patients, anger, aggression, and the MAP
140
rumination may both be involved in anger inhibition, the need to distinguish between beneficial
from adverse inner processes yet again calls for attention.
Conceptual confusion is widespread. If the concept of interest is blurry, the tools to
assess the concept will also be blurry. In the MAP Suppression subscale, attempts were made to tap
into the concepts introduced by the White Bear Suppression Paradigm, which focuses on mental
attempts to avoid angry emotions and anger-related thoughts. However, while the WBSI may
actually assess failed attempts to suppress thoughts as argued by Rassin (2003), in carefully
examining the MAP Suppression items, these items may detect how interested or motivated the
individual is in suppressing particular thoughts. As noted, because some individuals may have been
successful in their attempts to suppress thoughts while others were not, measuring attempts rather
than success may explain the nonsignificant findings.
It was also stated that the experience of uncontrollability is an importance facet of
anger arousal. This was confirmed by the expected correlation between MAP Negative Beliefs and
NAS Arousal subscales. Moreover, providing validity support for the MAP Rumination, it was
predicted that positive correlations between the NAS Arousal duration items and the rumination
items of the NAS Cognitive would be seen. This was confirmed.
To emphasizing the significance of bodily arousal in rumination, it was predicted that
the MAP Rumination and the NAS Arousal subscales would be positively correlated, which they
were. This association between rumination, arousal and anger is consistent with the anger
regulatory deficit model on anger in PTSD suggested by Chemtob et al. (1997). In this model, the
ruminative process is suggested to maintain increased physiological arousal at the risk of overriding
inhibitory controls on aggression.
The fifth hypothesis concerned the associations between the MAP and anger
dysregulation. Because the MAP is intended to measure variables involved in dysregulated anger, it
was predicted that significant negative correlations would be found between the MAP and the NAS
Regulation. This was only partly confirmed because the only significant correlation was between
the MAP Rumination and the NAS Regulation subscales. This finding offered more evidence for
the unhelpfulness of rumination as a coping strategy in situations of perceived threat.
Regarding the regression analysis, these results reproduced the earlier findings that the
MAP subscales account for a significant amount of variance in the criterion variable (anger), above
Chapter 4 Forensic patients, anger, aggression, and the MAP
141
variance explained by anxiety and depression. Thus, the association between the MAP and the NAS
is considered robust.
Even though the data were cross-sectional, which did not allow for causal inferences
to be made, and even though the sample size was small and increased the instability of the model, a
structural equation model was still used to assist with a more complex evaluation of the relationship
between the MAP and the NAS. Based on the assumption that when individuals with positive
beliefs about anger experience anger they will be liable to activate rumination as a coping strategy
(although rumination does not modify negative affect and the experience of uncontrollability
strengthens negative beliefs about anger), the depression model proposed by Papageorgiou and
Wells (2003) was tested. The results pointed towards the importance of negative beliefs in a
metacognitive framework on anger, indicating that negative beliefs may function as a mediator of
the relationship between rumination and anger. Hence, the model by Papageorgiou and Wells seems
to apply to anger as well. In conclusion, these results provide preliminary cross-sectional support
for a metacognitive anger model in which positive beliefs about anger are linked to rumination, and
rumination is closely linked to negative beliefs about anger. Rumination may also be linked to
bodily arousal, which serves a key function in mediating the relationship between rumination and
anger.
The eighth hypothesis predicted that patients with psychotic symptoms would have
higher anger scores. The comparisons of patients with and without psychotic symptoms using the
NAS Total supported the notion that psychotic symptoms are involved in the association between
major mental illness and anger. Patients experiencing delusions of persecution had higher scores on
the MAP Rumination than patients without delusions of persecution. This indicated that
experiencing delusions of persecution is associated with the tendency to experience uncontrollable
repetitive thinking. No other associations between the MAP and psychotic symptoms emerged.
These results do not support the idea that psychosis involving experiences of threat, danger and
persecution, may be associated with the belief that anger is a protective and helpful strategy. This
may imply that metacognitive beliefs about anger and symptoms of psychosis are associated only
through their shared association with anger.
Patients experiencing all three kinds of psychotic symptoms had higher scores on the
HADS Anxiety; this may indicate that experiencing psychotic symptoms is a frightful and anxiety-
provoking experience or that both are associated with the perception of threat.
Chapter 4 Forensic patients, anger, aggression, and the MAP
142
A large difference in the mean HADS Depression scores was found between patients experiencing
delusions of persecutions and patients who were not. Even though cross-sectional data do not allow
for conclusions about causality to be made, this association may signify that experiencing delusions
of persecution is associated with feelings of depression.
Anger measured by the NAS Total was associated with all three kinds of psychotic
symptoms, supporting the earlier claim that there may be a link between psychosis and anger that is,
to some extent, responsible for the link between psychosis and violence. The fact that no association
between psychosis and aggression was found in this study further supports this notion. It should be
noted that group size was small, compromising the validity of the results.
Regarding the ninth hypothesis centering on the PTSD symptoms of anger and
anxiety, patients scoring above the cut-off point for PCL-CV-4 had higher scores on the MAP
Negative Beliefs, MAP Rumination, HADS Anxiety, NAS Total, NAS Arousal and NAS
Behavioral subscales; this supports the hypothesized association between PTSD symptoms and
measures of anger and is consistent with the anger dysregulation model proposed by Chemtob et al.
(1997). Thus, it may be speculated that individuals with PTSD symptoms are overly prone to
interpret threat, and when they do, they tend to cope with the threat with rumination. Thus, in
situations of perceived threat, anger and physiological arousal are activated and decrease the ability
to regulate anger. In addition, physiological arousal may contribute to feelings of uncontrollability
and strengthen negative beliefs about anger. When threat perception is exaggerated and the
regulation of the physiological arousal of anger is ineffective due to the tendency to ruminate,
inhibitory functions may be insufficient and allow aggressive outlets.
These results emphasize the association between anger and PTSD, and furthermore,
they suggest that metacognitive beliefs about anger play a role in this relationship. Investigating
metacognition in relation to anger may shed light on mechanisms involved in anger dysregulation in
PTSD.
The tenth hypothesis predicted that individuals who had thought about or attempted
self-harm would have higher scores on the anger measures. Self-harm and attempted suicide were
included in the data because of their association with anger and aggression. Nijman and Campo
(2002) reported that the prevalence of self-harm in different inpatient studies ranges from 5.8 to
77%. However, in their study in an inpatient setting, they found a prevalence of 20%. In a forensic
setting, a 30% prevalence rate of self-harm was reported by Jeglic, Vanderhoff, and Donovick
Chapter 4 Forensic patients, anger, aggression, and the MAP
143
(2005); in the present study almost half of the patients had engaged in self-harm at some point,
which compared to these other reports is high. One explanation may be that the present study
assesses if the patient had ever engaged in self-harm, whereas the Jeglic et al. (2005) study reported
data on self-harm only during the course of incarceration. The present study found that patients
engaging in self-harm were more anxious and held more Negative Beliefs about anger than patients
who had not engaged in self-harm. This indicates an association among experiencing anxiety,
negative beliefs about anger as uncontrollable and dangerous, and the tendency to self-harm.
Although the causal relationships are not known, it may be speculated that fear of expressing anger
can provoke self-harm. Preferably, this speculation should be tested using a prospective design and
with a sample demonstrating high levels of self-harm. No significant associations emerged for
patients who had previously attempted suicide compared to those that had not. This indicates that
anger is not necessarily involved when individuals attempt suicide. In addition, the present data
assessed anger levels and inquired about earlier suicide attempts. A positive reply may refer to an
incident many years ago, which may have confounded results. Moreover, there were only 13
patients in the suicide attempt group, a sample size that would need to be larger to have high
confidence in these findings.
The eleventh hypothesis concerned violent fantasies. When comparing the prevalence
of patients categorized as SIV+ in the present study (44%) with the prevalence of SIV+ patients
identified by Grisso et al. (2000) among hospitalized mental patients (30%) and by Nagtegaal
(2008) (38%), prevalence was high in this study.
In support of the hypothesis that subjects who were categorized as SIV+ would have higher scores
on the MAP and the NAS than subjects categorized as SIV-, significant differences were found
between the SIV+ group and the SIV- group on the NAS the MAP Rumination subscales. Thus, the
present study supported the earlier findings of an association between violent fantasies and
anger/aggression. In attempting to understand the connection between violent fantasies and
anger/aggression, the SIP theory may be helpful. First, violent fantasies manifest as a rehearsal-
promoting thought control strategy, which is similar to what has been labeled as angry rumination.
Second, because of prior experiences with aggressive and hostile stimuli, aggressive
scripts/schemata are stored in long-term memory, and when they are regularly retrieved through
violent fantasizing or rumination, individuals may develop hyperactive aggressive scripts. Because
these scripts/schemata serve as templates for evaluating and interpreting social cues in future
situations, information processing is at risk of becoming biased towards hostility. With this
Chapter 4 Forensic patients, anger, aggression, and the MAP
144
understanding, aggressive scripts are reinforced through angry rumination and violent fantasies.
Furthermore, as the SIV+ group had a significantly higher mean score on the MAP Positive Beliefs,
individuals with elevated positive beliefs about anger may be less motivated to interrupt the process
of rehearsing aggressive scripts. In addition, significantly higher scores on the MAP Rumination in
the SIV+ group provided evidence for convergent validity.
Finally, based on the substantial body of research reporting an association between
anger and aggression, an important validation of the MAP would be to demonstrate an association
with aggression. Furthermore, an association between the NAS and observed aggression would
provide important clinical results.
In a review of the research conducted on the SOAS, Nijman et al. (2005) reported
yearly aggression rates ranging from 6% - 81% in different studies with patients who had conducted
an aggressive act. The only forensic ward that was included had a 74% rate of patients exhibiting
aggressive behavior within a year. In the present study, twenty-four (61%) of the patients had had
one or more aggressive episodes from admission until assessment. Excluding verbal aggression,
twenty-two (45%) had experienced one or more episode of aggression involving an object or direct
physical contact with another person. Twenty-seven (55%) had one or more aggressive episodes in
the follow-up period and when excluding verbal aggression, the number was seventeen (35%).
Pedersen (2009) reported on the entire population of the forensic unit of the Mental Health Centre
Sct. Hans from 2006-2007 and 37% of the patients had one or more aggressive episodes, including
verbal aggression during admission; in comparison, our sample revealed high levels of aggression.
These high levels of aggression may be explained by sample selection and the possibility that the
patients who were willing to participate in the study were also the most aggressive. It may also be
that the current staff at this facility reported more interactions as aggressive, than the previous staff
at the facility. Referring to the discussion in chapter 1 of the introduction, these results may
illuminate how challenging it is to accurately measure aggression. Most notably, what is labeled
aggression is dependent on the observer´s evaluation of what is taking place. The observer´s prior
experiences with the patient and the situational context in which the behavior is conducted may
impact this evaluation. For example, the behavior of a forensic patient with a record of prior
violence and attacks on the staff may more easily be considered aggressive than a patient with a less
intimidating behavioral history. Moreover, the staff may not notice all aggressive behavior. As
such, when the present data reveal higher levels of aggression this may be because the staff has
Chapter 4 Forensic patients, anger, aggression, and the MAP
145
improved their reporting skills. In order to limit the effects of these possible inaccuracies in
measuring aggression, verbal aggression was not included in the aggression analyses.
The association between anger measured by the NAS Total and aggression was
confirmed. As was found in a comparatively similar study by Doyle and Dolan (2006) in which
anger predicted inpatient aggression, the NAS Total was associated with aggression in the present
study. In both retrospective and prospective analyses, the results supported the assumption that the
connection between anger and aggression is a two-way street, each influencing the other.
Furthermore, the association with NAS Arousal underscores the importance of bodily arousal in
problematic anger. It seems reasonable to suggest that in the current results supporting an
association between anger and aggression, the relationship is at least partially mediated by arousal
components; the ability to regulate bodily arousal may prove to be a central focus from a clinical
perspective. The finding that the cognitive aspects of anger (NAS Cognitive) were associated with
aggression supports the view that anger is cognitively mediated, which is consistent with the
contemporary clinical models of anger discussed in the introduction.
Unexpectedly, only the MAP Negative Beliefs subscale was significantly associated
with aggression. The finding that rumination and aggression were not significantly correlated was
also unexpected, however, this may be explained by a relationship between rumination and
anger/aggression that is more complex than initially assumed. As such, negative beliefs about
uncontrollability and danger may be involved in the relationship between rumination and
aggression. Violent fantasies were not associated with aggression either, supporting the idea that
repeating violent schemata (ruminating) may not in and of themselves produce aggression.
However, violent fantasies may be associated with the experience of anger as an uncontrollable and
overwhelming emotional experience that causes inefficient regulation and increased risk of
aggression. The proposed associations between the MAP subscales and anger are supported by the
finding that only the MAP Negative Beliefs was significantly associated with aggression.
The present study has several limitations. First, the sample size is small, and second,
the accuracy of the aggression data may be less than optimal. Third, regarding the metacognitive
modeling, because the MAP and the NAS data were cross-sectional no causal inferences can be
drawn. Furthermore, the model did not include other variables of relevance in the model testing.
Because the preliminary finding that negative beliefs may mediate the relationship between
Rumination and anger has clinical importance, more rigorous testing of this model may be
Chapter 4 Forensic patients, anger, aggression, and the MAP
146
worthwhile in future investigations. The proposed model should preferably be tested on longitudinal
data.
Chapter 5 General discussion
147
Chapter 5 General Discussion
The results of Study 1 indicate that the MAQ-1 measures four distinct and reliable
categories of beliefs and processes in relation to anger. Because three of the subscales of the MAQ-
1 showed an association with the anger measure, the scale initially showed potential value for
understanding the cognitive mechanisms involved when individuals present with anger-related
problems.
The results of Study 2 reproduced the original factor structure and the reliability of the
subscales was satisfactory. The theoretically expected association between the general
metacognitive measure and this new measure of metacognition that specifically targets anger, was
supported by the intercorrelations between the two metacognitive measures. Furthermore, the
general metacognitive idea was supported by the high intercorrelations between the MAQ
subscales. Thus, the data indicate that positive as well as negative beliefs are involved in the
tendency to ruminate about angry emotions. Furthermore, this new tool shows potential for clinical
relevance with respect to anger-related problems because its subscales were related to anger level.
From Study 1 to Study 2 all correlations increased, supporting the scale revisions and indicating the
improved relevance of the scale in a clinical sample of prisoners with higher anger levels. In Study
2, three subscales showed an association with the anger measure that substantiates the MAQ-2 as
helpful in understanding the cognitive mechanisms involved in anger dysregulation.
In Study 3, the confirmatory factor analysis supported the three-dimensional structure
that was identified in Studies 1 and 2. Reliability was satisfactory. Again, the theoretically expected
association between the general metacognitive measure and this new measure of metacognition
specifically targeting anger was supported by the intercorrelations between the two metacognitive
measures. The validity of the rumination subscale was supported by several tests. The association
between the MAQ-3 subscales and measures of anger regulation confirmed that the tool measures
constructs involved in dysregulated anger. Finally, once again the MAQ-3 showed an association
with measures of anger.
In Study 4, the confirmatory factor analysis supported the three-dimensional structure
found in Studies 1, 2 and 3. Reliability was satisfactory. Validity of the subscales was addressed,
and overall the subscales performed well showing the expected associations. However, the
Chapter 5 General discussion
148
suppression subscale did not perform as expected. The other three subscales were associated with
anger, and a preliminary model of the associations was suggested: positive beliefs activate
rumination, which leads to increased bodily arousal and strengthening of negative beliefs that cause
deficits in anger regulation. This model was supported because only the negative beliefs were
associated with aggression.
In Study 4, anger was found to be associated with aggression, particularly in the
arousal and cognitive domains of anger. Rumination was found to be associated with violent
fantasies. Psychotic symptoms were found to be associated with anxiety and anger, and PTSD
symptoms were found to be associated with anxiety and anger. Self-harm was associated with
anxiety.
Threat and anger
Anxiety was found to be involved in anger and anger regulation in both Studies 3 and
4. This may be worthwhile to explore further and to take into consideration in relation to anger
intervention. The results imply that in situations of perceived threat, the individual may respond
with anger as a protective strategy, perhaps due to positive beliefs about anger. In some situations
this may be adaptive; Chemtob et al. (1997) discuss this concept in relation to PTSD and anger, but
in most everyday life situations, responding with anger when feeling threatened is not adaptive
because it may cause social distance and subjective discomfort.
The general metacognitive conceptualization proposes that perception of threat may
activate anxiety, which individuals may attempt to control using rumination. Unfortunately,
ruminations may have the unintended side-effect of strengthening negative beliefs about rumination
and maintaining emotional distress (Papageorgiou & Wells, 2003; Papageorgiou & Wells, 2001b;
Papageorgiou et al., 2001).
When activated by the unspecified perception of threat, the presence of metacognitive
beliefs about anger as protective may cause activation of rumination, which in turn maintains
arousal and strengthens negative beliefs about the uncontrollability and danger.
Bodily arousal
Because bodily arousal is a core characteristic of anger and the key target of anger
control, its association with negative beliefs about anger and rumination is of clinical importance.
As discussed in chapter 3, there is a large body of evidence substantiating that rumination exerts its
effects on the arousal of anger. This is supported by the findings of Studies 3 and 4. Furthermore,
Chapter 5 General discussion
149
increased arousal seems to be associated with negative evaluations of anger, indicating that the
experience of uncontrollability is an important facet of anger arousal.
Negative beliefs about anger
This subscale reflects themes that were discussed in chapter 2 relating to the
uncontrollability, danger and madness of anger. The results from the four studies presented in this
thesis support that these themes are critically involved in anger dysregulation. First, in Studies 2 and
3, the subscales reflecting themes of uncontrollability and danger across the general metacognitive
measure, the anger metacognitive measure and the anger measure showed positive intercorrelations.
Secondly, in Studies 3 and 4, negative beliefs were more strongly associated with anger and
aggression than with the other subscales.
The results point to uncontrollability and danger as principal themes that are essential
to a metacognitive conceptualization of emotional distress.
Types of self-focus
Rumination is a process of repetitive self-focus that has generally been found to
maintain negative mood. The rumination subscale was constructed to capture uncontrollable angry
rumination that is experienced as being beyond willful control, and it was consistently found to be
associated with anger.
Principally, in the metacognitive framework, all repetitive self-focus is perceived
unhelpful because it risks activating the CAS, which causes maladaptive goal-setting and inflexible
attempts to reach the goal (e.g., to worry in order to feel safe). Others have investigated different
types of self-focus, and rumination has been divided into a maladaptive form of self-focus and an
adaptive form that leads to a functional outcome (e.g., solve problems) (Trapnell & Campbell,
1999). Watkins (2004; 2008) labeled the maladaptive form, 'conceptual-evaluative' (rumination)
and the adaptive form, 'experiential self-focus'. The latter is associated with better recovery from an
upsetting event (Watkins, 2008).
Because the Cognitive Self-Consciousness subscale and the Positive Beliefs of the
MCQ-30 in Study 3 were inversely associated with anger regulation, it seems that that the more
self-focus the better the anger regulation. Thus, self-focused attention may also be adaptive in
relation to anger. This may demonstrate the limitations of using the unmodified S-REF framework
to measure anger and is further supported by the unsuccessful Cognitive-Consciousness subscale
(MAQ-1 and MAQ-2). This subscale was constructed to reflect increased self-focus as manifested
Chapter 5 General discussion
150
in the S-REF framework, but the internal reliability of the subscale was poor and the association to
anger was non-significant.
In future studies it may be important to explore different types of self-focus in relation
to anger to differentiate unhelpful types of self-focus from the helpful types of self-focus that could
help an individual develop skills to regulate anger.
Anger inhibition
In Study 3, an association between withholding anger and rumination emerged.
Furthermore, there were indications that negative evaluations of anger are involved in choosing to
withhold anger.
The association between withholding anger and rumination may suggest a benefit of
"letting off steam". More specifically, if an individual who actually harbors anger-related emotions
and thoughts inhibits an impulsive, aggressive reaction, the anger may continue internally as
rumination. Rumination is associated with anger, possibly due to its maintenance of bodily arousal.
Anger inhibition may be associated with increased anger due to its association with rumination. If
the outward expression of anger is inhibited, the individual may get caught up in rumination. Within
this process is the idea that inhibition of an angry response, because of the risk of activating
rumination, has the potential to backfire and actually increase the risk of anger or aggression. The
association between negative evaluations of anger and anger inhibition seems logical. This finding,
due to the potential counteractive effects of anger inhibition, may indicate the importance of
normalizing and de-stigmatizing anger for the purpose of modifying negative evaluations of anger.
In the understanding of the relationship between anger suppression and its negative
effects, it has been suggested that failed suppression may lead to rumination (Wenzlaff & Luxton,
2003). Sukhodolsky (2001) argued that in order to ruminate, you must suppress. He took the view
that within the information processing sequence of the cognitive and emotional systems, anger
rumination refers to what happens to anger after it has been suppressed. Next, the paradoxical effect
of suppression has been repeatedly demonstrated. This means that inhibition and suppression will
not be sufficiently effective for several reasons. In Study 4 we attempted to incorporate a subscale
measuring the tendency to suppress anger. However, the results did not support the hypothesis that
negative evaluations of anger would motivate suppression, and that when suppression fails, the risk
of activating rumination as a coping strategy targeting the negative affect caused by the failed
suppression is increased. The results did not support a paradoxical effect of suppression resulting in
Chapter 5 General discussion
151
increased anger. In future studies involving the MAP, the convergent validity of the Suppression
subscale should be substantiated. Future studies should include the WBSI as a convergent validity
test as well as scales such as the STAXI-2-AX-in and ARS.
Positive beliefs about anger
The Positive Beliefs subscale was associated with anger consistently throughout the
four studies. The subscale was found to be closely associated with cognitions justifying anger,
hostile attitude, confrontation, suspiciousness and ideas about the need to watch out to avoid being
hurt by others. The Positive Beliefs subscale may represent a cognitive network that increases the
risk of an anger-related responses, underscoring how anger may be conceived as a problem solving
strategy for dealing with perceived unpleasantness, adversity, danger and ill-will. It was interesting
that in Study 4, positive beliefs was not significant in the regression analyses nor was it significant
in group comparisons related to aggression. This supports the proposed relationship between
metacognition and anger suggested by the SEM analysis. In this proposed relationship, positive
beliefs are associated with anger through their activation of rumination, which maintains negative
affect and strengthen negative beliefs associated with anger dysregulation. This proposed
relationship is supported by the fact that positive beliefs were not associated with aggression in
Study 4, but negative beliefs were.
Metacognitive patterns
The general increase in inter-scale correlations of the MAQ (MAP) subscales from
Study 1 to Study 2, Study 2 to Study 3, and finally from Study 3 to Study 4 supports the revisions of
the scale.
Furthermore, the different patterns of intercorrelations that were found may have
clinical relevance. Two correlations differed from Study 2 to Study 3 as the sample type changed
from prisoners to clinical patients; the association between positive beliefs and rumination subscale
weakened, and the association between negative beliefs and rumination strengthened. This altered
pattern of correlations may be indicative of how metacognition manifests and interacts differently in
depending on sample type.
For example, an individual holding a belief about anger as a survival strategy may
overinvest energy resources into anger experiences and thus have a tendency to ruminate. Another
individual holding a belief about the uncontrollability of anger may ruminate as well, but it will be
driven by a different mechanism. More specifically, the results indicate that the mechanisms driving
Chapter 5 General discussion
152
a particular dysfunctional processing strategy are related to different metacognitive beliefs. Even
though comparisons of the means across the two samples was not possible because the MAQ-3 was
revised between studies, an increase in positive beliefs about anger may be expected in criminal
samples with violent histories, such as in Study 2. In this way, the prisoners may have ruminated as
a result of their positive beliefs about anger, while the clinical patients may have ruminated because
they felt no control over the process, which caused increased negative beliefs about anger. The
results support the concept brought forth in Study 3 that some individuals may ruminate because
they believe that anger is a helpful problem solving strategy, while others may ruminate because
they have no control over the ruminative process. Among individuals with prior histories of
aggression, anger/aggression may have been learned as a strategy for coping and solving problems
(Bandura, 1973), and thus, these individuals have positive beliefs about the function of anger. In the
first case, clinical interventions could explore the patient´s experience of anger as protective to
successfully interrupt the dysfunctional processing strategy. An agreed upon benefit of anger is its
ability to mobilize energy and psychological resources, assisting the individual in overcoming
obstacles. In the right place, at the right time, anger has a functional value. Demonstrating this
concept, in an experimental design, Tamir, Mitchell, and Gross (Tamir, Mitchell, and Gross, 2008)
showed that anger increased participant performance in a violent video game, suggesting that in
particular contexts, functional levels of anger can help individuals achieve their goals. In this line of
work, it is argued that emotions and behaviors are instrumentally regulated. As such, even though
anger is a negatively experienced emotion, people may prefer to experience anger when it promotes
the attainment of a goal (Tamir, 2009). Similarly, if people believe that anger serves them well and
helps them to achieve different goals, they will be less likely to abandon a ruminative process.
In the second case, the ruminative process is associated with negative beliefs and
interventions may be more successful when they focus on lowering physiological arousal and
providing a sense of control over the emotional experience and the ruminative process. Achieving
this may help the patient to experience control and to alter negative beliefs, thus facilitating the on-
going ability to interrupt the ruminative processes.
Further indications of these potentially different patterns of metacognition in different
sample types comes from Studies 3 and Study 4, in which the study sample changes from clinical to
forensic patients. Between Studies 3 and 4, the association between positive beliefs and rumination
strengthened, while the association between negative beliefs and rumination weakened. Evidence
from the regression analyses suggests that in the nonclinical sample (Study 1) and the prisoner
Chapter 5 General discussion
153
sample (Study 2), negative beliefs were less associated with anger, while in the clinical sample
(Study 3) and the forensic sample (Study 4), negative beliefs were highly associated with anger.
Only in the forensic sample were positive beliefs about anger not significantly associated with anger
in the regression analyses. At first, it may seem confusing that positive beliefs about anger play a
smaller role in anger than negative beliefs when the forensic sample is known to have high anger
levels. However, this may shed light on mechanisms involved in dysregulated anger. Taken
together, these results indicate that negative evaluations of anger and experiences of anger as
uncontrollable and dangerous are more involved in clinical anger than in normal experiences of
anger. As such, these features are particularly important to conceptualize regarding the clinical
anger literature.
In light of these different patterns of associations, it would have been interesting to
test the fit of the depression model by Papageorgiou and Wells (2003) that was preliminarily tested
with the SEM on the forensic data from Study 3. Conducting structural equation modeling on larger
samples of clinical patients, including other relevant variables, and adopting a longitudinal design to
investigate the relationship between the MAP and anger regulation may be clinically valuable.
Dual anger experience
Positive and Negative beliefs about anger reflect the twofold nature of the anger
construct as discussed in chapter 1 and 2. The finding that the correlation between Positive and
Negative beliefs was non-significant in the first two studies but significant in the clinical and
forensic samples at the p<.01 level may shed light on anger in relation to psychopathology.
As such, the fundamental characteristic of anger as a dual experience may be
particularly evident in clinical samples, reflecting an important marker that distinguishes between
anger related to psychopathology and anger that is considered normal. As argued in the
introduction, anger in itself is not problematic; however, anger may become problematic under
certain circumstances. Some of these are circumstances are related to the subjectively distressful
experience of anger. In a metacognitive understanding of emotional distress, elevated unhelpful
metacognition has been shown in several clinical conditions such as: GAD (Wells & Carter, 2001);
depressive rumination (Papageorgiou & Wells, 2003); OCD (Solem et al., 2009); and psychosis
(Morrison et al., 2007). With this view, both positive and negative metacognitive beliefs interact
with the mechanism by which the dysfunctional thinking style is maintained. Metacognitive
patterns have been shown to distinguish a clinical from a non-clinical condition. For instance, in
GAD, worry has been differentiated into type 1 and type 2 worry. Type 1 is `ordinary worry´ that
Chapter 5 General discussion
154
resembles problem solving, which may be initiated based largely on positive beliefs about worry.
Type 2 worry is the process of worrying about the worry, which is largely seen as the result of
negative beliefs about worry (Wells, 2005). This is considered to be the actual pathological process
that maintains the distressful experience.
Regarding anger, the presence of both positive and negative beliefs may similarly be
related to an increased confusing experience of anger and accompanied by elevated physiological
arousal and the risk of getting trapped in unhelpful processing strategies. As such, the levels of both
positive and negative beliefs about anger will assumedly be higher in clinical samples. Indicative of
the clinical relevance of metacognition in understanding how psychopathology is maintained,
higher correlations between positive and negative beliefs were found in the clinical samples than in
the non-clinical samples.
General metacognition
In Studies 2 and 3, the MCQ-30 cognitive confidence subscale showed an association
with anger. It has been suggested by Papageorgiou and Wells (2004) that metacognitive efficiency
may be a byproduct of depression. They argue that metacognitive efficiency is not only a symptom
of depression but also contributes to the belief about the need to continue to ruminate and negative
beliefs concerning the various consequences of engaging in this process. This means that when
captured performing repetitive thinking in the CAS, individuals lose confidence in their own
abilities to think. In addition, patients with clinical depression have decreased metacognitive
efficiency. Moreover, this decreased metacognitive efficiency strengthens the tendency to ruminate
and the negative beliefs about the consequences. This seems to apply to anger as well, because it is
well-known that high anger arousal compromises cognitive ability. Thus, individuals presenting
with difficulty in regulating anger may also report low levels of confidence in their own cognition.
This facet of metacognition was omitted from the scale development early in the process due to low
face validity. However, in future revisions of the MAP it may prove useful to incorporate features
of confidence in one's own cognitive functioning. The model that was tested in Papageorgiou and
Wells (2003) included the MCQ-30 Cognitive Confidence subscale. The data from Study 3 also
included this subscale, and a model incorporating the MAQ-3 subscales and confidence in own
cognition (MCQ-30) may contribute to the understanding of anger-related problems.
Chapter 5 General discussion
155
Transdiagnostic approach
The general metacognitive model argues that experiences of anxiety or depression will
activate this repetitive thinking process, which increases risk for the Cognitive Attentional
Syndrome, due to the presence of positive beliefs about worry/rumination. Because anxiety was
repeatedly found to be associated with anger measures as well as with the MAQ and the MAP, this
general idea seems to apply to anger. Thus, in situations of perceived threat, the activated negative
affect may vary in anxiety, depression and anger, but the thinking process (i.e., unhelpful repetitive
thinking) may persist across disorders. The metacognitive approach is but one example of a
transdiagnostic approach. Shifting the perspective from disorder-focused to process-focused has the
benefit of gaining insight from parallel work. Using a transdiagnostic approach, in this thesis the
presented framework for anger attempted to utilize work from other psychological disorders.
Chapter 5 General discussion
156
Appendix A
157
Appendix A: The MAQ-1
Metacognition and Anger Questionnaire (MAQ-1)
The following statements are assumptions people hold about their own thoughts and
emotions.
How true are they for you?
For each statement indicate whether it is (1) never true, (2) sometimes true, (3) often true,
(4) always true. Use the scale to the right to mark the answer that fits the most.
No. Statement Never
True
1
Sometimes
True
2
Often
True
3
Always
True
4
1 Anger helps me control other people.
2 Other people will not tolerate anger.
3 When I am angry I keep thinking about it.
4 I cannot distance myself from my thoughts.
5 If I did not get angry, I could get hurt.
6 I am constantly aware of my thinking.
7 I must be aware of unjust actions against me.
8 I cannot let go of angry thoughts.
9 Anger is difficult to control, it takes control over you.
10 It is perfectly natural to get angry when faced with injustice.
11 My anger harms myself.
12 Anger helps me see things the way they are.
13 I do not believe in avoiding my anger.
14 It is bad to have angry thoughts.
©
Appendix A
158
No. Statement Never
True
1
Sometimes
True
2
Often
True
3
Always
True
4
15 When I start getting angry, I cannot stop.
16 Anger is bad for me.
17 I can easily understand other people’s emotional responses.
18 I need to let some steam out now and again, in order to not
explode later.
19 Anger helps me solve problems.
20 I must be observant about being treated badly.
21 I must control my thoughts.
22 Anger helps me cope with things.
23 Anger could make me go mad.
24 If you do not show other people that you are tough, they will
think you are soft.
25 I cannot ignore my anger.
26 Anger keeps me safe.
27 Anger will make other people reject you.
28 When angry one should think about alternative solutions.
29 I try to distract myself when I am angry.
30 Anger can harm other people.
31 I do not think clearly when I am angry.
32 Being angry will make me loose control and go mad.
33 Anger is good for me.
34 Angry thoughts persist, no matter how I try to stop them.
35 My anger is dangerous for me.
36 Only weak people do not get angry.
37 I cannot distract myself from anger.
Appendix A
159
No. Statement Never
True
1
Sometimes
True
2
Often
True
3
Always
True
4
38 It is best to ignore anger.
39 I need to watch out for threats and dangers.
40 When I get angry, I get energized.
41 Anger means loss of control.
42 I loose focus on different points of view when I am angry.
43 Anger protects me from being exploited by others.
44 I monitor my thoughts and emotions, particularly when I
feel angry.
45 Anger makes me a strong and capable person.
46 If I am being treated badly, it is necessary to get angry.
47 I can have trouble recognising my own emotions.
48 Anger makes me a bad person.
49 Others will be judgemental of you for getting angry.
50 I am able to calm myself when angry.
51 I will be punished for not controlling certain thoughts.
52 One must calm oneself when angry.
53 It is not good to focus on anger.
54 Anger is necessary to get by in the world.
55 Anger makes me insensitive to other people.
56 I am aware of my thoughts the instant they arise.
57 Anger keeps me alert.
©
Appendix B
161
Appendix B: The MAQ-2
Metacognition and anger (MAQ-2)
Never true
Sometimes true
Often true
Always true
3. When I am angry I keep thinking about it
1 2 3 4
4. I cannot distance myself from my thoughts
1 2 3 4
6. I am constantly aware of my thinking
1 2 3 4
7. I am aware of unjust actions against me
1 2 3 4
8. I cannot let go of angry thoughts
1 2 3 4
9. Anger is difficult to control, it takes control over you
1 2 3 4
11. Anger harms oneself
1 2 3 4
12. Anger helps me see things the way they are
1 2 3 4
14. It is bad to have angry thoughts
1 2 3 4
15. When I start getting angry, I cannot stop
1 2 3 4
16. Anger is bad for me
1 2 3 4
17. I can easily understand other people’s emotions
1 2 3 4
19. Anger helps me solve problems
1 2 3 4
21. I must control my thoughts
1 2 3 4
22. Anger helps me cope with things
1 2 3 4
C. I keep an eye out for potential danger and threats around me
1 2 3 4
The statements below describe beliefs that people have about own thoughts and emotions.
How true are they for you?
For each statement please indicate whether is (1) never true, (2) sometimes true, (3) often true, (4) always true. Use the scale at your right to circle the answer that best describes how true the
statement is for you
Appendix B
162
Never true
Sometimes
true
Often true
Always
true
23. Anger could make me go mad
1 2 3 4
25. I cannot ignore anger
1 2 3 4
26. Anger keeps me safe
1 2 3 4
27. Anger will make other people reject you
1 2 3 4
30. My anger can harm other people
1 2 3 4
31. I do not think clearly when I am angry
1 2 3 4
32. Anger can make me loose control and go mad
1 2 3 4
33. Anger is good for me
1 2 3 4
35. My anger is dangerous for me
1 2 3 4
37. I cannot distract myself from anger 1 2 3 4
C. I am constantly aware of my emotions
1 2 3 4
R. To figure it out, I think a lot about situations that make me
1 2 3 4
C. I think a lot about my thoughts and emotions to understand them
1 2 3 4
41. Anger means loss of control
1 2 3 4
42. I loose sight of different points of view when I am angry
1 2 3 4
43. Anger protects me from being exploited by others
1 2 3 4
45. Anger makes me a strong and capable person
1 2 3 4
R. I hold on to the anger, so people will understand that they went too far
1 2 3 4
R. It is impossible not to think about things that make me angry
1 2 3 4
C. Some thoughts are necessary to keep under control 1 2 3 4
Appendix B
163
Never true
Sometimes
true
Often true
Always
true
48. Anger makes me a bad person
1 2 3 4
49. Others will be judgemental of ones anger
1 2 3 4
C. I am pretty engaged in how my thinking works
1 2 3 4
R. When I am angry, I can only think about that
1 2 3 4
R. If I continue thinking about what makes me angry, I will be able to solve it
1 2 3 4
C. I think a lot about my thoughts and emotions
1 2 3 4
54. Anger is necessary to get by in the world
1 2 3 4
55. Anger makes you insensitive to other people
1 2 3 4
57. Anger keeps you alert
1 2 3 4
©
Appendix C
164
Appendix C: The MAQ-3
No. Statement Never
true
Sometimes
true
Often
true
Always
true
1. When I am angry, I keep thinking about it
1 2 3 4
2. I am aware of my thoughts
1 2 3 4
3. My anger harms me
1 2 3 4
4. Anger helps me see things the way they really are
1 2 3 4
5. I cannot step back from my angry thoughts
1 2 3 4
6. I understand the emotional reactions of other people
1 2 3 4
7. Anger could make me go mad
1 2 3 4
8. Anger helps me to solve problems
1 2 3 4
9. I cannot let go of angry thoughts
1 2 3 4
10. I focus on controlling my thoughts
1 2 3 4
11. My anger could hurt others
1 2 3 4
12. Anger helps me handle threats and dangers
1 2 3 4
13. Anger is hard to control; it controls you
1 2 3 4
14. I am aware of my emotions
1 2 3 4
15. Anger means loss of control
1 2 3 4
16. Anger protects me 1 2 3 4
Metacognition and Anger Questionnaire (MAQ-3)
The statements below describe beliefs people have about own thoughts and emotions.
How true are they for you? For each statement rate if it is (1) never true, (2) Sometimes true, (3) Often true or (4)
Always true. Please use the scale at your right to circle your answer
Appendix C
165
Never
true
Sometimes
true
Often
true
Always
true
17. When I start to get angry, I cannot stop
1 2 3 4
18. I think about things in order to understand them
1 2 3 4
19. My anger is dangerous for me
1 2 3 4
20. Anger makes me a strong and competent person
1 2 3 4
21. I cannot ignore my anger
1 2 3 4
22. I am aware of how my thinking works
1 2 3 4
23. When I am angry, I lose sight of different points of view
1 2 3 4
24. My anger will make people realize that they went too far
1 2 3 4
25. When I am angry, I cannot distract myself
1 2 3 4
26. I think about my thoughts and emotions
1 2 3 4
27. Anger makes me a bad person
1 2 3 4
28. Anger is necessary to get by in the world
1 2 3 4
29. When I am angry, I can only think about that
1 2 3 4
30. I analyze my reactions to things
1 2 3 4
31. Anger will make other people think badly about me
1 2 3 4
32. Anger keeps me alert
1 2 3 4
33. Anger stays with me for a long time
1 2 3 4
34. My emotions can confuse me
1 2 3 4
35. Anger makes me insensitive to others
1 2 3 4
Appendix D
166
Appendix D: The MAP
Metacognition and Anger Processing (MAP)
No. Statement Never
true
Sometimes
true
Often
true
Always
true
1. When I am angry, I keep thinking about it
1 2 3 4
2. I really try to avoid my angry emotions
1 2 3 4
3. My anger harms me
1 2 3 4
4. Anger helps me see things the way they really are
1 2 3 4
5. I cannot step back from my angry thoughts
1 2 3 4
6. When I am angry I simply try to forget it
1 2 3 4
7. Anger could make me go mad
1 2 3 4
8. Anger helps me to solve problems
1 2 3 4
9. I cannot let go of angry thoughts
1 2 3 4
10. It is important for me not to think about the things that make me
angry
1 2 3 4
11. My anger could hurt others
1 2 3 4
12. Anger helps me handle threats and dangers
1 2 3 4
13. Anger is hard to control; it controls you
1 2 3 4
14. Anger makes me a bad person
1 2 3 4
The statements below describe and reactions beliefs people may have in relation to anger
How true are they for you?
For each statement please rate if it is (1) never true, (2) sometimes true, (3) often true or (4)
always true.
Use the scale at your right to mark your answer
Appendix D
167
Never
true
Sometimes
true
Often
true
Always
true
15. Anger protects me
1 2 3 4
16. When I start to get angry, I cannot stop
1 2
3 4
17. Anger prefer not to attend to anger at all
1 2 3 4
18. My anger is dangerous for me
1 2 3 4
19. Anger makes me a strong and competent person
1 2 3 4
20. I cannot ignore my anger
1 2 3 4
21. When I am angry, I lose sight of different points of view
1 2 3 4
22. My anger will make people realize that they went too far
1 2 3 4
23. When I am angry, I cannot distract myself
1 2 3 4
24. I do not like to be reminded of angry emotions
1 2 3 4
25. Anger means loss of control
1 2 3 4
26. Anger is necessary to get by in the world
1 2 3 4
27. When I am angry, I can only think about that
1 2 3 4
28. Anger will make other people think badly about me
1 2 3 4
29. Anger keeps me alert
1 2 3 4
30. Anger stays with me for a long time
1 2 3 4
31. When I am angry I prefer to avoid thinking about it
1 2 3 4
32. Anger makes me insensitive to others
1 2 3 4
Appendix E
168
Appendix E: The MAP -Danish
No. Udsagn Aldrig
sandt
Nogen
gange
sandt
Ofte
sandt
Hele
tiden
sandt
1. Når jeg er vred, bliver jeg ved med at tænke over det 1 2 3 4
2. Jeg prøver virkeligt af undgå mine følelser af vrede 1 2 3 4
3. Min vrede skader mig 1 2 3 4
4. Vrede hjælper mig til at se ting, som de virkelig er 1 2 3 4
5. Jeg kan ikke træde tilbage fra mine vrede tanker 1 2 3 4
6. Når jeg er vred, forsøger jeg bare at glemme det 1 2 3 4
7. Vrede kunne gøre mig vanvittig 1 2 3 4
8. Vrede hjælper mig til at løse problemer 1 2 3 4
9. Jeg kan ikke give slip på vrede tanker 1 2 3 4
10. Det er vigtigt for mig, ikke at tænke over de ting, der gør mig vred 1 2 3 4
11. Min vrede kunne skade andre 1 2 3 4
12. Vrede hjælper mig til at håndtere trusler og farer 1 2 3 4
13. Vrede er svær at kontrollere; den kontrollerer mig 1 2 3 4
14. Vrede gør mig til en dårlig person 1 2 3 4
15. Vrede beskytter mig 1 2 3 4
Metakognition og Vrede (MAP) Sætningerne forneden beskriver overbevisninger og reaktioner folk kan have i forbindelse med vrede.
Hvor sande er de for dig?
For hvert udsagn tag venligst stilling til om det er (1) aldrig sandt, (2) nogen gange
sandt, (3) ofte sandt, (4) hele tiden sandt.
Anvend skalaen til højre til at afmærke det svar, der passer bedst.
Appendix E
169
Aldrig
sandt
Nogen
gange
sandt
Ofte
sandt
Hele
tiden
sandt
16. Når jeg begynder at blive vred, kan jeg ikke stoppe 1 2 3 4
17. Vrede vil jeg helst ikke beskæftige mig med 1 2 3 4
18. Min vrede er farlig for mig 1 2 3 4
19. Vrede gør mig til en stærk og kompetent person 1 2 3 4
20. Jeg kan ikke ignorere min vrede 1 2 3 4
21. Når jeg er vred, mister jeg blik for forskellige synspunkter 1 2 3 4
22. Min vrede vil få folk til at forstå, at de er gået for langt 1 2 3 4
23. Når jeg er vred, kan jeg ikke distrahere mig selv 1 2 3 4
24. Jeg bryder mig ikke om at blive mindet om følelser af vrede 1 2 3 4
25. Vrede betyder tab af kontrol 1 2 3 4
26. Vrede er nødvendigt for at klare sig i verden 1 2 3 4
27. Når jer er vred, kan jeg kun tænke på dét 1 2 3 4
28. Vrede vil få andre mennesker til at tænke dårligt om mig 1 2 3 4
29. Vrede holder mig på dupperne 1 2 3 4
30. Vrede bliver hængende længe hos mig 1 2 3 4
31. Når jeg er vred, vil jeg helst undgå at tænke på det 1 2 3 4
32. Vrede gør mig ufølsom overfor andre 1 2 3 4
Appendix F
170
Appendix F: Norms study of the NAS-PI
The primary goals of this study were to test the internal reliability and construct
validity of the Danish translation of the NAS-PI and to present Danish normative data from
several populations. Comparisons with relevant international data were conducted.
Participants and Procedure
Participants were gathered from several sites during the period between August
2007 and January 2011.
The non-clinical sample.
The normative sample was collected from four different groups. One group
consisted of university students (N = 243); people attending a political meeting on a weekend (N
= 126); employees at a private corporation (N = 108); and police students who only completed
the PI (N =191). Total N = 668. The sample consisted of 314 male (4 missing) participants, the
mean age (5 missing) was 29.4 (SD = 9.9, range 17 – 72), and the mean length of education (24
missing) was 13.7 (SD = 2.0, range 7 – 23).
Mixed clinical sample
The clinical group consisted of patients at the psychiatric facilities in South of
Zealand (N = 77) who only completed the NAS clinical and patients at Frederikssund hospital (N
= 87) who completed the NAS and the PI. Total N = 164. The sample consisted of 59 male (4
missing) participants, the mean age (4 missing) was 39.7 (SD = 13.8, range 16 – 76), and the
mean length of education (10 missing) was 11.1 (SD = 2.6, range 7 – 18).
Legal status sample
The legal status sample was characterized by males in conflict with the law in some
fashion. The sample consisted of a group of male inmates (61% convicted of a violent crime) at 5
different prisons in Denmark (N = 167) who only completed the PI and male forensic patients
(92% convicted of a violent crime) who completed only the NAS (N = 64). Total N = 231. The
mean age was 32.4 (SD = 10.5, range 18 – 67), and the mean length of education (7 missing) was
9.5 (SD = 2.2, range 7 – 15).
Appendix F
171
Measures.
The measures were the NAS and the PI
Translation procedure.
With permission from the original author (Ray Novaco, personal communication,
August 2007) the first author translated the NAS-PI into Danish, and two independent, bilingual
native speakers (one was an expert in the field and the other had no knowledge of psychology)
back-translated the questionnaire. Back-translations were compared to that of the original by an
independent psychiatrist, and differences were discussed and resolved between the editing
psychiatrist and the primary researcher. In this process, the original author was contacted to
ensure the comparability over cultural differences and idiographic expressions. A number of the
items underwent alternations in wording during this process.
Results.
Descriptive statistics are available in Table 1. Cronbach´s alpha values indicate satisfying to
excellent internal reliability. Inspections of histograms with the distribution of the items showed
symmetrical distributions.
Differences in mean scores of anger measures by gender
Gender differences in mean scores on the PI Total, NAS Total, NAS behavioral,
NAS cognitive, NAS arousal and NAS regulation were analyzed using univariate analyses of
variance, ANOVA. The analyses were conducted separately for the 3 samples and revealed 2
significant results. Within the university student group, males had significantly lower scores on
the NAS arousal (M(males) = 24.1; M(females) = 26.5; F(1, 240) = 11.73 (p = .001) and significantly
higher scores on the NAS regulation subscales (M(males) = 28.9; M(females) = 27.7; F(1, 240) = 6.41
(p = .012).
Differences in mean scores compared with other data
The mean scores were compared for the three samples as well as with data from
non-clinical and clinical groups from Sweden and the UK. The results are displayed in Table 1.
Appendix F
172
Table 1. Descriptive statistics for the NAS-PI subscales, comparisons of non-clinical and clinical samples.
Type á PI Total (25) á NAS Total
(48)
á NAS cognitive
(16)
á NAS
behavioral
(16)
á NAS arousal
(16)
á
NAS
regulation
(12)
DK non-clinical
N = 668 .87 52.9 (9.9) N = 477 .88 75.8 (10.0) .70 26.6 (3.6) .74 23.3 (3.8) .76 25.9 (4.2) .63 27.9 (2.9)
UK non-clinical
N = 212 .92 53.1 (11.1) N = 58 .92 74.5 (11.5) .78 26.5 (4.0) .82 22.6 (4.2) .82 25.4 (4.7)
SE non-clinical
(males)
N = 100 .87 55.4 (9.7) N = 100 .90 77.6 (11.6) .70 28.7 (3.9) .80 23.3 (4.4) .81 25.5 (5.0) .76 26.8 (3.6)
DK clinical
N = 87 .89 61.7 (11.2) N = 164 .94 92.7 (16.6) .81 32.0 (5.3) .88 28.3 (6.9) .84 32.5 (6.0) .70 25.4 (3.7)
SE male prisoners
N = 92 .94 62.5 (15.1) N = 92 .94 94.2 (17.1) .78 34.3 (5.0) .91 29.9 (7.3) .86 30.1 (6.3) .81 24.8 (4.1)
DK legal status
N = 167 .92 65.4 (14.0) N = 64 .93 97.1 (16.8) .80 34.1 (5.8) .86 30.9 (6.8) .79 32.2 (5.8) .77 26.8 (4.4)
UK anger referrals
N = 58 65.8 (14.5) N = 58 103.2 (17.1) 34.1 (5.0) 34.6 (7.2) 34.4 (6.3)
Note. Level of significance is p < .01.
Appendix F
173
Table 2. Subscale correlations of the NAS subscale and the PI for the non-clinical sample, the clinical and the legal
status sample.
PI Total NAS Total NAS cognitive NAS behavioral NAS arousal NAS regulation
PI Total
1 .60*; .57* .55*; .51* .44*; .45*; .56*; .56* -.23*; -.36*
NAS Total
1 .89*; .85*; .92* .93*; .84*; .92* .92*; .87*; .89* -.32*;-.42*; -.23
NAS cognitive
1 .73*; .59*; .78* .73*; .62*; .73* -.22*; -.36*; -.17
NAS behavioral
1 .77*; .57*; .71* -.33*; -.33*; -.21
NAS arousal
1 -.31*; -.39*; -.24
Note. * p < .01. First number in each cell displays the correlation for the non-clinical sample, and the second number
for the clinical sample, and the last number for the legal status sample.
Correlations
The validity of the Danish translation of the NAS-PI was addressed by computing
correlations (Pearson) among the subscales. All of the subscales showed high correlations,
ranging from .44 - .92. As expected, the NAS regulation subscale was negatively correlated with
the other subscales, ranging from -.17 to -.42. The correlations are displayed in Table 2.
Construct validity
The factor structure of the Danish translation of the NAS-PI was tested by running a
Confirmatory Factor Analysis using M-plus statistical software, version 6 (Muthén & Muthén,
2010). The five-factor model specified in the NAS-PI manual (Novaco, 2003) entered the factors
as intercorrelated and goodness of fit was determined on the basis of several indices: Chi-square,
comparative fit index (CFI), the Tucker-Lewis Index (TLI), and the Root Mean Square Error of
Approximation (RMSEA). When a good-fit model has been achieved, the 2 test should be non-
significant or the ratio of the 2 divided by the degrees of freedom should be less than 2. CFI
values larger than .90 -.95 are recommended to suggest a good model fit, a RMSEA value below
Appendix F
174
.05-.06, and a TLI value larger than .95 is likewise indicative of a good model fit (Ching-Yun,
2002; Ullman, 2007).
First, the analysis was conducted on the complete sample of 1064 individuals. The
Chi-square test of overall model fit was significant ( 2 (3559) = 13931.03, p < .000. However,
when examining the other indices of model fit (CFI = .98, TLI = .98, RMSEA = .057) the data
showed an acceptable model fit. Analyzing the non-clinical and the clinical samples separately
yielded an improved fit for the non-clinical sample ( 2 (3559) = 8069, p < .000, CFI = .99, TLI =
.99, RMSEA = .047) as well as for the clinical sample ( 2 (3559) = 3877.16, p < .000, CFI = .98,
TLI = .98, RMSEA = .033). The fit for the NAS four-factor model in the forensic sample was not
satisfactory ( 2 (1704) = 2098.30, p < .000, CFI = .74, TLI = .73, RMSEA = .06). The fit for the
one-factor PI model in the prisoner´s sample showed a fit approaching acceptance ( 2 (275) =
444.49, p < .000, CFI = .94, TLI = .93, RMSEA = .061).
Conclusion
From these results, the translation of the NAS-PI was deemed successful, and
overall, the results supported the reliability and validity of the translation.
Appendix G
175
Appendix G: Metacognitive profiling
Conduct a Functional Analyses of a situation involving anger (sum up the thoughts, emotions and
behaviours, bodily arousal)
1. What did you do to cope with the situation of being angry?
Prompt questions:
Strategies for processing:
Did you do anything to prevent the thoughts from coming?
Did you try to not think about what made you angry?
Did you try to distract yourself, think about something else, leave the situation?
Did you dwell on details of the event (thoughts, emotions, body sensations)?
Did you say anything to yourself in order to deal with the situation?
Did you do anything to control what you were thinking/feeling?
Beliefs about coping:
What were you trying to achieve with your coping?
What did you want to make happen?
What was your goal?
Result of coping attempt:
Did any of what you did help?
How did what you did in the situation affect your thoughts?
Did they change? How much and how long?
How did the coping attempt affect your mood?
How did the coping attempt affect your sensation in the body, more or less relaxed?
Did the anger continue? How long?
Did you at any time during the situation change the way you reacted to the anger?
Did you perhaps first try to distract or suppress the angry thoughts and later engage in rumination
about the angry thoughts? Or something else..?
What if questions about beliefs about result of strategies (metabeliefs)s:
What if you imagine you could not have …………….. (coping strategy) what would then have
happened? How would the situation have unfolded?
What if you would not have been able to control the thoughts/emotions?
Would they have gone on and on?
What if you had continued to have these thoughts/emotions, what would have happened? What
would have been worst case scenario?
Appendix G
176
2. Focus on one disturbing thought
Metacognitive strategies in coping with a disturbing thought:
When you had this thought, what did it do to you, tell me about how it affected you and how you
reacted towards it
What feelings did you get?
Bodily reactions?
What passed through your head as the thought entered your mind?
Did you think anything about having this thought? What?
What did it mean to you that this thought entered your mind?
In relation to having this thought, did you worry/ruminate about anything?
What was the worry/ruminate about?
Did you do anything else as a result of the worry/rumination?
How long did you worry/ruminate?
Could you stop yourself from worrying/ruminating?
In relation to having this thought, did you focus your attention on anything?
Where were your attention?
What did you focus on in the situation where you had this angry thought? (your own reaction, the
thought itself, other thoughts, external ques etc.?)
1. Did you in any way kind of stay with the anger, think a lot about it, unable to think of anything
else, focusing your attention on the anger, perhaps thinking:
o a lot about what made you angry?
o a lot about how you could get even?
o a lot about a need to stay with the anger in order to protect your self or handle the
situation (positive beliefs)?
2. Did you try not to think about it?
3. Metacognitive beliefs
Occurrence:
Do you think there are any advantages/anything positive about having these angry thoughts/angry
emotions? Do they help you in any way?
Do you think there are any disadvantages/anything negative about having these angry
thoughts/angry emotions? Do they disturb or harm you in any way? If any, in what way?
Overall do you think it is mostly negative or positive for you to have these angry
thoughts/emotions?
Appendix G
177
Attention:
Do you think there are any advantages of the way you focus your attention in a situation of angry
thoughts/emotions?
Do you think there are any disadvantages of the way you focus your attention in a situation of
angry thoughts/emotions?
How do you think your attention influence the situation, the outcome, the angry
thought/emotions? Does your attention make you more or less angry and what do you think about
that?
What do you think about your success controlling your own thinking? How good are you? Are
there anything making it difficult in any way?
4. Modus
When you had the angry thought, did you accept is as the truth, the fact of the situation based on
reality?
How convinced were you of your thought (0-100%)
Did you when you had the thought have any acceptance of possible other alternative truths of the
situation?
To what extent could you in the situation take distance to your thought? Have acceptance of the
fact that a thought is merely a thought, one among many possible interpretations of what is going
on, instead of the only truth of what´s going on?
Summary:
What are the strategies for handling anger?
The strategies serve the function of regulation the emotion:
1. What are the beliefs about the emotion/thought that drives the strategies for coping?
2. What are the beliefs about coping that drives the strategies for coping?
3. What is the result of the coping? (more/less anger)
178
Reference List
Abramowitz, J. S., Tolin, D. F., & Street, G. P. (2001). Paradoxical effects of thought suppression:
a meta-analysis of controlled studies. Clinical Psychology Review, 21, 683-703.
Aldao, A. & Nolen-Hoeksema, S. (2010). Specificity of cognitive emotion regulation strategies: A
transdiagnostic examination. Behaviour Research and Therapy, 48, 974-983.
Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2010). Emotion-regulation strategies across
psychopathology: A meta-analytic review. Clinical Psychology Review, 30, 217-237.
Allport, G. W. (1942). The use of personal documents in psychological science. New York: Social
Science Research Council.
American Psychiatric Association (2000). Diagnostic criteria from DSM-4-TR. (4 ed.) Washington:
American Psychiatric Association.
Anderson, C. A. & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53,
27-51.
Anderson, C. A. (1997). Effect of violent movies and trait hostility on hostile feelings and
aggressive thoughts. Aggressive Behavior, 23, 161-178.
Appelbaum, P. S., Robbins, P. C., & Monahan, J. (2000). Violence and delusions: data from the
MacArthur violence risk assessment study. American Journal of Psychiatry, 157, 566-572.
Archer, J. & Haigh, A. M. (1997b). Do beliefs about aggressive feelings and actions predict
reported levels of aggression? British Journal of Social Psychology, 36, 83-105.
179
Archer, J. & Haigh, A. (1997a). Beliefs about aggression among male and female prisoners.
Aggressive Behavior, 23, 405-415.
Austin, S. (2011). Center for Psykiatrisk Forskning, Risskov.
Ref Type: Personal Communication
Averill, J. R. (1982). Anger and aggression: an essay on emotion. New York: Springer-Verlag.
Averill, J. R. (1983). Studies on anger and aggression: implications for theories of emotion.
American Psychologist, 38, 1145-1160.
Baegels, S. M. & Mansell, W. (2004). Attention processes in the maintenance and treatment of
social phobia: hypervigilance, avoidance and self-focused attention. Clinical Psychology
Review, 24, 827-856.
Bailey, C. & Ostrov, J. (2008). Differentiating forms and functions of aggression in emerging
adults: associations with hostile attribution biases and normative beliefs. Journal of
Youth and Adolescence, 37, 713-722.
Baker, M. T., Van Hasselt, V. B., & Sellers, A. H. (2008). Validation of the Novaco Anger Scale in
an incarcerated offender population. Criminal Justice and Behavior, 35, 741-754.
Bandura, A. (1973). Aggression: a social learning analysis. Englewood Cliffs, NJ: Prentice Hall.
Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H.
(2007). Threat-related attentional bias in anxious and nonanxious individuals: a meta-
analytic study. Psychological Bulletin January 2007, 133, 1-24.
180
Barazzone, N. & Davey, G. C. L. (2009). Anger potentiates the reporting of threatening
interpretations: an experimental study. Journal of Anxiety Disorders, 23, 489-495.
Barratt, E. S. (1994). Impulsiveness and aggression. In J.Monahan & H. J. Steadman (Eds.),
Violence and Mental Disorder: Developments in risk assessment (pp. 61-81). Chicago:
University of Chicago Press.
Beck, A. T. (1999). Prisoners of hate: the cognitive basis of anger, hostility, and violence. New
York: HarperCollins.
Beck, A. T. & Clark, D. A. (1997). An information processing model of anxiety: automatic and
strategic processes. Behaviour Research and Therapy, 35, 49-58.
Bellmore, A. D., Witkow, M. R., Graham, S., & Juvonen, J. (2005). From beliefs to behavior: the
mediating role of hostile response selection in predisting aggression. Aggressive
Behavior, 31, 453-472.
Benazzi, F. (2003). Major depressive disorder with anger: a bipolar spectrum disorder?
Psychotherapy and Psychosomatics, 72, 300-306.
Berkowitz, L. (1993). Aggression: its causes, consequences, and control. New York: McGraw-Hill.
Berkowitz, L. (2005). On hate and its determinants: some affective and cognitive influences. In
R.J.Sternberg (Ed.), Psychology of Hate (pp. 155-183). Washington DC: American
Association Association.
Berkowitz, L. & Harmon-Jones, E. (2004). Toward an understanding of the determinants of
anger. Emotion, 4, 107-130.
181
Berkowitz, L. (1989). Frustration-Aggression Hypothesis: Examination and Reformulation.
[Review]. Psychological Bulletin, 106, 59-73.
Berkowitz, L. (1990). On the formation and regulation of anger and aggression: a cognitive-
neoassociationistic analysis. American Psychologist, 45, 494-503.
Berkowitz, L. (2008). On the consideration of automatic as well as controlled psychological
processes in aggression. Aggressive Behavior, 34, 117-129.
Bjelland, I., Dahl, A. A., Haug, T. T., & Neckelmann, D. (2002). The validity of the Hospital Anxiety
and Depression Scale: an updated literature review. Journal of Psychosomatic Research,
52, 69-77.
Bliese, P. D., Wright, K. M., Adler, A. B., Cabrera, O., Castro, C. A., & Hoge, C. W. (2008).
Validating the Primary Care Posttraumatic Stress Disorder Screen and the Posttraumatic
Stress Disorder Checklist with soldiers returning from combat. Journal of Consulting and
Clinical Psychology, 76, 272-281.
Bo, S., Abu-Akel, A., Kongerslev, M., Haahr, U. H., & Simonsen, E. (2011). The risk factors for
violence among patients with schizophrenia. Clinical Psychological Review, In Press.
Boden, J. M. & Baumeister, R. F. (1997). Repressive coping: distraction using pleasant thoughts
and memories. Journal of Personality and Social Psychology, 73, 45-62.
Bonta, J., Law, M., & Hanson, K. (1998). The prediction of criminal and violent recidivism among
mentally disordered offenders: a meta-analysis. Psychological Bulletin, 123, 123-142.
Bower, G. (1981). Mood and memory. American Psychologist, 36, 129-148.
182
Brennan, P. A., Mednick, S. A., & Hodgins, S. (2000). Major mental disorders and criminal
violence in a Danish birth cohort. Archives of General Psychiatry, 57, 494-500.
Bryant, R. A. & Harvey, A. (1997). Attentional bias in posttarumatic stress disorder. Journal of
Traumatic Stress, 10, 635-644.
Burns, J., Quartana, P., & Bruehl, S. (2008). Anger inhibition and pain: conceptualizations,
evidence and new directions. Journal of Behavioral Medicine, 31, 259-279.
Bushman, B. J. & Anderson, C. A. (2001). Is it time to pull the plug on the hostile versus
instrumental aggression dichotomy? Psychological Review, 108, 273-279.
Bushman, B. J., Bonacci, A. M., Pedersen, W. C., Vasquez, E. A., & Miller, N. (2005). Chewing on it
can chew you up: effects of rumination on triggered displaced aggression. Journal of
Personality and Social Psychology, 88, 969-983.
Bushman, B. J., Baumeister, R. F., & Phillips, C. M. (2001). Do people aggress to improve their
mood? Catharsis beliefs, affect regulation opportunity, and aggressive responding.
Journal of Personality & Social Psychology, 81, 17-32.
Caprara, G. V., Paciello, M., Gerbino, M., & Cugini, C. (2007). Individual differences conducive to
aggression and violence: trajectories and correlates of irritability and hostile rumination
through adolescence. Aggressive Behavior, 33, 359-374.
Caprara, G. V. (1986). Indicators of aggression: The Dissipation-Rumination Scale. Personality
and Individual Differences, 7, 763-769.
183
Cartwright-Hatton, S. & Wells, A. (1997). Beliefs about worry and intrusions: The Meta-
Cognitions Questionnaire and its correlates. Journal of Anxiety Disorders, 11, 279-296.
Chan, S. C., Raine, A., & Lee, T. M. C. (2010). Attentional bias towards negative affect stimuli and
reactive aggression in male batterers. Psychiatry Research, 176, 246-249.
Chapman, A. L. & Dixon-Gordon, K. L. (2007). Emotional antecedents and consequences of
deliberate self-harm and suicide attempts. Suicide and Life-Threatening Behavior, 37,
543-552.
Chemtob, C. M., Novaco, R. W., Hamada, R. S., Gross, D. M., & Smith, G. (1997). Anger
regulation deficits in combat-related posttraumatic stress disorder. Journal of Traumatic
Stress, 10, 17-36.
Chen, Y. P., Ehlers, A., Clark, D. M., & Mansell, W. (2002). Patients with generalized social phobia
direct their attention away from faces. Behaviour Research and Therapy, 40, 677-687.
Ching-Yun, Y. (2002). Evaluation cutoff criteria of model fit indices for latent variable models
with binary and continuous outcomes. University of California, Los Angeles.
Cisler, J. M. & Koster, E. H. W. (2010). Mechanisms of attentional biases towards threat in
anxiety disorders: An integrative review. Clinical Psychology Review, 30, 203-216.
Cohen, D. J., Eckhardt, C. I., & Schagat, K. D. (1998). Attention allocation and habituation to
anger-related stimuli during a visual search task. Aggressive Behavior, 24, 399-409.
Collins, K. & Bell, R. (1997). Personality and aggression: the Dissipation-Rumination Scale.
Personality and Individual Differences, 22, 751-755.
184
Cornell, D. G., Peterson, C. S., & Richards, H. (1999). Anger as a predictor of aggression among
incarcerated adolescents. Journal of Consulting and Clinical Psychology, 67, 108-115.
Crick, N. R. & Dodge, K. A. (1994). A review and reformulation of social information-processing
mechanisms in children's social adjustment. Psychological Bulletin, 115, 74-101.
Crick, N. R. & Dodge, K. A. (1996). Social information-processing mechanisms in reactive and
proactive aggression. Child Development, 67, 993-1002.
Davey, L., Day, A., & Howells, K. (2005). Anger, over-control and serious violent offending.
Aggression and Violent Behavior, 10, 624-635.
Deffenbacher, J. L. (1999). Cognitive-behavioral conceptualization and treatment of anger.
Journal of Clinical Psychology, 55, 295-309.
Derogatis, L. R. & Melisaratos, N. (1983). The Brief Symptom Inventory: an introductory report.
Psychological Medicine, 13, 595-605.
DiGiuseppe, R. & Tafrate, R. C. (2007). Understanding Anger Disorders. New York: Oxford
University Press.
Dodge, K. A. & Crick, N. R. (1990). Social information-processing bases of aggressive-behavior in
children. Personality and Social Psychology Bulletin, 16, 8-22.
Dollard, J., Doob, L. W., Miller, N. E., Mowrer, O. H., & Sears, R. R. (1939). Frustration and
aggression. New Haven, CT: Yale University Press.
Dollard, J. & Miller, N. E. (1998). Frustration and Aggression. Florence, KY: Routledge.
185
Douglas, K. S., Guy, L. S., & Hart, S. D. (2009). Psychosis as a risk factor for violence to others: a
meta-analysis. Psychological Bulletin, 135, 679-706.
Doyle, M. & Dolan, M. (2006). Evaluating the validity of anger regulation problems,
interpersonal style, and disturbed mental state for predicting inpatient violence.
Behavioral Sciences & the Law, 24, 783-798.
Eckhardt, C., Norlander, B., & Deffenbacher, J. (2004). The assessment of anger and hostility: a
critical review. Aggression and Violent Behavior, 9, 17-43.
Eckhardt, C. I. & Cohen, D. J. (1997). Attention to anger-relevant and irrelevant stimuli following
naturalistic insult. Personality and Individual Differences, 23, 619-629.
Epps, J. & Kendall, P. C. (1995). Hostile attributional bias in adults. Cognitive Therapy &
Research, 19, 159-178.
Fava, M. & Rosenbaum, J. F. (1998). Anger attacks in depression. Depression and Anxiety, 8, 59-
63.
Fazel, S., Gulati, G., Linsell, L., Geddes, J. R., & Grann, M. (2009). Schizophrenia and violence:
systematic review and meta-analysis. PLoS Medicine, 6, e1000120.
Feeny, N. C., Zoellner, L. A., & Foa, E. B. (2000). Anger, dissociation, and posttraumatic stress
disorder among female assault victims. Journal of Traumatic Stress, 13, 89-100.
Fernandez-Duque, D., Baird, J. A., & Posner, M. I. (2000). Executive attention and metacognitive
regulation. Consciousness and Cognition, 9, 288-307.
186
Feshbach, S. (1964). The function of aggression and the regulation of aggressive drive.
Psychological Review, 71, 257-272.
Fisher, P. L. & Wells, A. (2009). Metacognitive therapy. London & New York: Routledge.
Flavell, J. H. (1979). Meta-cognition and cognitive monitoring - new area of cognitive-
developmental inquiry. American Psychologist, 34, 906-911.
Freeman D. & Garety P.A. (2000). Comments on the content of persecutory delusions: Does the
definition need clarification? British Journal of Clinical Psychology, 39, 407-414.
Freeman, D. & Garety, P. A. (2003). Connecting neurosis and psychosis: the direct influence of
emotion on delusions and hallucinations. Behaviour Research and Therapy, 41, 923-947.
Fresco, D. M., Frankel, A. N., Mennin, D. S., Turk, C. L., & Heimberg, R. G. (2002). Distinct and
overlapping features of rumination and worry: the relationship of cognitive production
to negative affective states. Cognitive Therapy and Research, 26, 179-188.
Geen, R. G. (2001). Human aggression. (2 ed.) London: Open University Press.
Gerin, W., Davidson, K. W., Christenfeld, N. J. S., Goyal, T., & Schwartz, J. E. (2006). The Role of
Angry Rumination and Distraction in Blood Pressure Recovery From Emotional Arousal.
Psychosomatic Medicine, 68, 64-72.
Gilbert, P., Gilbert, J., & Irons, C. (2004). Life events, entrapments and arrested anger in
depression. Journal of Affective Disorders, 79, 149-160.
Gorsuch, R. L. (1983). Factor analyses. (2 ed.) Hillsdale, NJ: Erlbaum.
187
Gould, R. A., Ball, S., Kaspi, S. P., Otto, M. W., Pollack, M. H., Shekhar, A. et al. (1996). Prevalence
and correlates of anger attacks: A two site study. Journal of Affective Disorders, 39, 31-
38.
Gratz, K. (2001). Measurement of deliberate self-harm: preliminary data on the Deliberate Self-
Harm Inventory. Journal of Psychopathology and Behavioral Assessment, 23, 253-263.
Green, M. J., Williams, L. M., & Hemsley, D. R. (2000). Cognitive theories of delusion formation:
the contribution of visual scanpath research. Cognitive Neuropsychiatry, 5, 63-74.
Grisso, T., Davis, J., Vesselinov, R., Appelbaum, P. S., & Monahan, J. (2000). Violent thoughts and
violent behavior following hospitalization for mental disorder. Journal of Consulting &
Clinical Psychology, 68, 388-398.
Gross, J. J., Richards, J. M., & John, O. P. (2006). Emotion regulation in everyday life. In
D.K.Snyder, J. A. Simpson, & J. N. Hughes (Eds.), Emotion regulation in families: Pathways
to dysfunction and health (pp. 2-34). Washington: American Psychological Association.
Gross, J. J. & John, O. P. (2003). Individual differences in two emotion regulation processes:
implications for affect, relationships, and well-being. Journal of Personality & Social
Psychology, 85, 348-362.
Hanneke, P., Bram Orobio, d. C., Willem, K., Herman, W. v. B., & Welmoet, W. M. (2007). A
meta-analysis of the distinction between reactive and proactive aggression in children
and adolescents. Journal of Abnormal Child Psychology, 35, 522.
188
Harvey, A., Watkins, E., Mansell, W., & Shafran, R. (2004). Cognitive behavioural processes
across psychological disorders - a transdiagnostic approach to research and treatment.
Oxford: Oxford University Press.
Hodgins, S. (2008). Review: Violent behaviour among people with schizophrenia: a framework
for investigations of causes, and effective treatment, and prevention. Philosophical
Transactions Royal Society: B, 636, 1-14.
Hoeping, W. & de Jong-Meyer, R. (2003). Differentiating unwanted intrusive thoughts from
thought suppression: what does the White Bear Suppression Inventory measure?
Personality and Individual Differences, 34, 1049-1055.
Hogan, B. & Linden, W. (2004). Anger response styles and blood pressure: at least Dont
Ruminate about it! Annals of Behavioral Medicine, 27, 38-49.
Huesmann, L. R. (1988). An information processing model for the development of aggression.
Aggressive Behavior, 14, 13-24.
Huesmann, L. R. & Guerra, N. G. (1997). Children's normative beliefs about aggression and
aggressive behavior. Journal of Personality & Social Psychology, 72, 408-419.
Huppert, J. D. & Smith, T. E. (2005). Anxiety and schizophrenia: the interaction of subtypes of
anxiety and psychotic symptoms. CNS Spectr., 10, 721-731.
Ingram, R. E. (1990). Self-focused attention in clinical disorders: review and a conceptual model.
Psychological Bulletin, 107, 156-176.
189
James, M. & Seager, J. A. (2006). Impulsivity and schemas for a hostile world: postdictors of
violent behaviour. International Journal of Offender Therapy and Comparative
Criminology, 50, 47-56.
Jarvis, K. L. & Novaco, R. W. (2006). Postshelter adjustment of children from violent families.
Journal of Interpersonal Violence, 21, 1046-1062.
Jeglic, E. L., Vanderhoff, H. A., & Donovick, P. J. (2005). The function of self-harm behavior in a
forensic population. International Journal of Offender Therapy and Comparative
Criminology, 49, 131-142.
Jones, J. P., Thomas-Peter, B. A., & Trout, A. (1999). Normative data for the Novaco Anger Scale
from a non-clinical sample and implications for clinical use. British Journal of Clinical
Psychology, 38, 417-424.
Joyal, C. C., Dubreucq, J. L., Gendron, C., & Millaud, F. (2007). Major mental disorder and
violence: a critical update. Current Psychiatry Reviews, 3, 33-50.
Kassinove, H. & Sukhodolsky, D. G. (1995). Anger disorders: basic science and practice issues.
Issues in Comprehensive Pediatric Nursing, 18, 173-205.
Kay, S. R., Wolkenfeld, F., & Murrill, L. M. (1988). Profiles of aggression among psychiatric
patients: I. nature and prevalence. The Journal of Nervous and Mental Disease, 176.
Klinger, E. (1996). Emotional influence on cognitive processing, with implications for theories of
both. In Gollwitzer P.M & J. A. Bargh (Eds.), The psychology of action: linking cognition
and emotion to behavior (pp. 168-189). New York: Guilford.
190
Koster, E. H. W., De Raedt, R., Goeleven, E., Franck, E., & Crombez, G. (2005). Mood-congruent
attentional bias in dysphoria: maintained attention to and impaired disengagement from
negative information. Emotion, 5, 446-455.
Lindqvist, J. K., Daderman, A. M., & Hellstrom, A. (2005). Internal reliability and construct
validity of the Novaco Anger Scale-1998-S in a sample of violent prison inmates in
Sweden. Psychology Crime & Law, 11, 223-237.
Lindqvist, J. K., Daderman, A. M., & Hellstrom, A. (2003). Swedish adaptations of the Novaco
Anger Scale - 1998, the Provocation Inventory, and the State-Trait Anger eXpression
Inventory-2. Social Behavior & Personality: An International Journal, 31, 773-788.
Link, B. G., Monahan, J., Stueve, A., & Cullen, F. T. (1999). Real in their consequences: a
sociological approach to understanding the association between psychotic symptoms
and violence. American Sociological Review, 64, 316-332.
Lories, G., Dardenne, B., & Yzerbyt, V. Y. (1998). From social cognition to metacognition. In
V.Y.Yzerbyt, G. Lories, & B. Dardenne (Eds.), Metacognition: Cognitive and Social
Dimensions (pp. 1-15). London: Sage.
Luciano, J. V., Belloch, A., Algarabel, S., Tomas, J. M., Morillo, C., & Lucero, M. (2006).
Confirmatory factor analysis of the White Bear Suppression Inventory and the Thought
Control Questionnaire: a comparison of alternative models. European Journal of
Psychological Assessment, 22, 250-258.
Martinez, M. E. (2006). What is metacognition? Phi Delta Kappan, 87, 696-699.
191
Mathews, A., Ridgeway, V., & Williamson, D. A. (1996). Evidence for attention to threatening
stimuli in depression. Behaviour Research and Therapy, 34, 695-705.
Mathews, A. & Wells, A. (2004). Rumination, depression, and metacognition: the S-REF model.
In C.Papageorgiou & A. Wells (Eds.), Depressive Rumination: Nature, Theory and
Treatment (pp. 125-152). Wiley & Sons.
Matthews, G. & Wells, A. (2000). Attention, automaticity, and affective disorder. Behavior
Modification, 24, 69-93.
McNiel, D. E., Eisner, J. P., & Binder, R. L. (2003). The relationship between aggressive
attributional style and violence by psychiatric patients. Journal of Consulting and Clinical
Psychology, 71, 399-403.
Meffert, S. M., Metzler, T. J., Henn-Haase, C., McCaslin, S., Inslicht, S., Chemtob, C. et al. (2008).
A prospective study of trait anger and PTSD symptoms in police. Journal of Traumatic
Stress, 21, 410-416.
Monahan, J., Steadman, H. J., Silver, E., Appelbaum, P. S., Robbins, P. C., Mulvey, E. P. et al.
(2001). Rethinking risk assessment: the MacArthur study of mental disorder and violence.
New York: Oxford University Press.
Moritz, S. & Laudan, A. (2007). Attention bias for paranoia-relevant visual stimuli in
schizophrenia. Cognitive Neuropsychiatry, 12, 381-390.
192
Morrison, A. P., French, P., & Wells, A. (2007). Metacognitive beliefs across the continuum of
psychosis: Comparisons between patients with psychotic disorders, patients at ultra-high
risk and non-patients. Behaviour Research and Therapy, 45, 2241-2246.
Morrison, A. P., Gumley, A. I., Schwannauer, M., Campbell, M., Gleeson, A., Griffin, E. et al.
(2005). The Beliefs about Paranoia Scale: preliminary validation of a metacognitive
approach to conceptualizing paranoia. Behavioural and Cognitive Psychotherapy, 33,
153-164.
Morrison, A. P. & Haddock, G. (1997). Self-focused attention in schizophrenic patients with and
without auditory hallucinations and normal subjects: a comparative study. Personality
and Individual Differences, 23, 937-941.
Moscovitch, D. A., McCabe, R. E., Antony, M. M., Rocca, L., & Swinson, R. P. (2008). Anger
experience and expression across the anxiety disorders. Depression and Anxiety, 25, 107-
113.
Muthén, L. K. & Muthén, B. O. (2010). Mplus statistical analysis with latent variables - user´s
guide (Version 6) [Computer software]. Los Angeles, CA: Munthén & Munthén.
Nagtegaal, M. (2008). Aggression: its association with dysfunctional thought control processes,
cognition, and personality. University of Maastrich.
Nagtegaal, M. H., Rassin, E., & Muris, P. (2006). Aggressive fantasies, thought control strategies,
and their connection to aggressive behaviour. Personality and Individual Differences, 41,
1397-1407.
193
Nagtegaal, M. H. & Rassin, E. (2004). The usefulness of the thought suppression paradigm in
explaining impulsivity and aggression. Personality and Individual Differences, 37, 1233-
1244.
Nelson, T. O., Stuart, R. B., Howard, C., & Crowley, M. (1999). Metacognition and clinical
psychology: a preliminary framework for research and practice. Clinical Psychology &
Psychotherapy, 6, 73-79.
Nijman, H. L. I., Palmstierna, T., Almvik, R., & Stolker, J. J. (2005). Fifteen years of research with
the Staff Observation Aggression Scale: a review. Acta Psychiatrica Scandinavica, 111,
12-21.
Nijman, H. L. I. & Campo, J. M. L. G. (2002). Situational determinants of inpatient self-harm.
Suicide and Life-Threatening Behavior, 32, 167-175.
Nijman, H. L. I., Muris, P., Merckelbach, H. L. G. J., Palmstierna, T., Wistedt, B., Vos, A. M. et al.
(1999). The staff observation aggression scale revised (SOAS-R). Aggressive Behavior,
25, 197-209.
Nolen-Hoeksema, S. (2000). The role of rumination in depressive disorders and mixed
anxiety/depressive symptoms. Journal of Abnormal Psychology, 109, 504-511.
Nolen-Hoeksema, S. (2004). The response styles theory. In C.Papageorgiou & A. Wells (Eds.),
Depressive Rumination: Nature, Theory and Treatment (pp. 107-124). Chichester, UK:
Wiley & Sons.
194
Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). Rethinking rumination. Perspectives
on Psychological Science, 3, 400-425.
Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration of
depressive episodes. Journal of Abnormal Psychology November 1991; 100, 569-582.
Nolen-Hoeksema, S. (1998). The other end of the continuum: the costs of rumination.
Psychological Inquiry, 9, 216-219.
Nolen-Hoeksema, S. & Morrow, J. (1993). Effects of rumination and distraction on naturally
occurring depressed mood. Cognition and Emotion, 7, 561-570.
Novaco, R. (1994). Anger as a risk factor for violence among the mentally disordered. In
J.Monahan & H. J. Steadman (Eds.), Violence and Mental Disorder: Developments in risk
assessment (pp. 21-61). Chicago: University of Chicago Press.
Novaco, R. (1998). Aggression. In Encyclopedia of Mental Health (1 ed., pp. 13-26). Academic
Press.
Novaco, R. (2003). NAS-PI. Los Angeles: Western Psychological Services.
Novaco, R. (2010a). Anger and psychopathology. In M.Potegal (Ed.), International Handbook of
Anger (pp. 2-68). New York: Springer.
Novaco, R. (2010b). Anger assessment in forensic settings. Forensic Update 100, 42-47.
195
Novaco, R. & Renwick, S. J. (2002). Anger predictors of assultiveness and the validation of a
ward behavior scale for anger and aggression. Manuscript submitted for publication .
Ref Type: Abstract
Novaco, R. W. (1976). The functions and regulation of the arousal of anger. American Journal of
Psychiatry, 133, 1124-1128.
Novaco, R. W. & Chemtob, C. M. (2002). Anger and combat-related posttraumatic stress
disorder. Journal of Traumatic Stress, 15, 123-132.
Novaco, R. W. & Taylor, J. L. (2004). Assessment of anger and aggression in male offenders with
developmental disabilities. Psychological Assessment, 16, 42-50.
Novaco, R. W. (2000). Anger. In Encyclopedia of stress (1 ed., pp. 188-195). Academic Press.
Novaco, R. W. (2007). Anger dysregulation. In T.A.Cavell & K. T. Malcom (Eds.), (pp. 3-54).
London: Lawrence Erlbaum Associates.
Orth, U., Cahill, S. P., Foa, E. B., & Maercker, A. (2008). Anger and posttraumatic stress disorder
symptoms in crime victims: a longitudinal analysis. Journal of Consulting & Clinical
Psychology, 76, 208-218.
Orth, U. & Wieland, E. (2006). Anger, hostility, and posttraumatic stress disorder in trauma-
exposed adults: a meta-analysis. Journal of Consulting and Clinical Psychology, 74, 698-
706.
Ottaviani, C., Shapiro, D., & Fitzgerald, L. (2010). Rumination in the laboratory: what happens
when you go back to everyday life? Psychophysiology, 48, 453-461.
196
Painuly, N., Sharan, P., & Mattoo, S. K. (2005). Relationship of anger and anger attacks with
depression. European Archives of Psychiatry and Clinical Neuroscience, 255, 215-222.
Papageorgiou, C. & Wells, A. (2001a). Positive beliefs about depressive rumination:
development and preliminary validation of a self-report scale. Behavior Therapy, 32, 13-
26.
Papageorgiou, C. & Wells, A. (2004). Nature, functions, and beliefs about depressive rumination.
In C.Papageorgiou & A. Wells (Eds.), Depressive Rumination: Nature, Theory and
Treatment (pp. 2-21). Wiley.
Papageorgiou, C., Wells, A., & Meina, L. J. (2001). Development and preliminary validation of the
Negative Beliefs about Rumination Scale. Manuscript in preparation.
Papageorgiou, C. & Wells, A. (1999). Process and meta-cognitive dimensions of depressive and
anxious thoughts and relationships with emotional intensity. Clinical Psychology &
Psychotherapy, 6, 156-162.
Papageorgiou, C. & Wells, A. (2001b). Metacognitive beliefs about rumination in recurrent major
depression. Cognitive and Behavioral Practice, 8, 160-164.
Papageorgiou, C. & Wells, A. (2003). An empirical test of a clinical metacognitive model of
rumination and depression. Cognitive Therapy and Research, 27, 261-273.
Pedersen, L. (2009). Assessment of Violence Risk in Mentally Disordered Offenders: A Structured
Proffessional Judgement Approach in Danish Forensic Psychiatry. Department of
Psychology, University of Copenhagen.
197
Perlis, R. H., Smoller, J. W., Fava, M., Rosenbaum, J. F., Nierenberg, A. A., & Sachs, G. S. (2004).
The prevalence and clinical correlates of anger attacks during depressive episodes in
bipolar disorder. Journal of Affective Disorders, 79, 291-295.
Posternak, M. A. & Zimmerman, M. (2002). Anger and aggression in psychiatric outpatients.
Journal of clinical psychiatry, 63, 665-672.
Potegal, M. & Novaco, R. (2010). A brief history of anger. In M.Potegal, G. Stemmler, & C. D.
Spielberger (Eds.), International Handbook of Anger (pp. 9-24). New York: Springer.
Purdon, C. (1999). Thought suppression and psychopathology. Behaviour Research and Therapy,
37, 1029-1054.
Purdon, C. (2004). Empirical investigations of thought suppression in OCD. Journal of Behavior
Therapy and Experimental Psychiatry, 35, 121-136.
Purdon, C., Rowa, K., & Antony, M. M. (2005). Thought suppression and its effects on thought
frequency, appraisal and mood state in individuals with obsessive-compulsive disorder.
Behaviour Research and Therapy, 43, 93-108.
Quartana, P. J. & Burns, J. W. (2007). Painful consequences of anger suppression. Emotion, 7,
400-414.
Raine, A., Dodge, K., Loeber, R., Gatzke-Kopp, L., Lynam, D., Reynolds, C. et al. (2006). The
reactive proactive aggression questionnaire: differential correlates of reactive and
proactive aggression in adolescent boys. Aggressive Behavior, 32, 159-171.
198
Rassin, E. (2003). The White Bear Suppression Inventory (WBSI) focuses on failing suppression
attempts. European Journal of Personality, 17, 285-298.
Ray, R. D., Wilhelm, F. H., & Gross, J. J. (2008). All in the mind's eye? Anger rumination and
reappraisal. Journal of Personality & Social Psychology, 94, 133-145.
Renwick, S. J., Black, L., Ramm, M., & Novaco, R. W. (1997). Anger treatment with forensic
hospital patients. Legal and Criminological Psychology, 2, 103-116.
Richards, J. M. & Gross, J. J. (1999). Composure at any cost? The cognitive consequences of
emotion suppression. Personality and Social Psychology Bulletin, 25, 1033-1044.
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., Mayer, J. D., Goldman, S. L., Turvey, C., & Palfai, T. P. (1995). Emotional attention,
clarity, and repair: exploring emotional intelligence using the Trait Meta-Mood Scale. In
J.W.Pennebaker (Ed.), Emotion, Disclosure, and Health (pp. 125-154). Washington:
American Psychological Association.
Seager, J. A. (2005). Violent men: the importance of impulsivity and cognitive schema. Criminal
Justice and Behavior, 32, 26-49.
Segerstrom, S. C., Stanton, A. L., Alden, L. E., & Shortridge, B. 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.
199
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.
Simourd, D. J. D. J. & Mamuza, J. M. (2000). The Hostile Interpretations Questionnaire:
psychometric properties and construct validity. Criminal Justice and Behavior, 27, 645-
663.
Simpson, C. & Papageorgiou, C. (2003). Metacognitive beliefs about rumination in anger.
Cognitive and Behavioral Practice, 10, 91-94.
Sirotich, F. (2008). Correlates of crime and violence among persons with mental disorder: an
evidence-based review. Brief Treatment and Crisis Intervention, 8, 171-194.
Skeem, J. L., Schubert, C., Odgers, C., Mulvey, E. P., Gardner, W., & Lidz, C. (2006). Psychiatric
symptoms and community violence among high-risk patients: a test of the relationship
at the weekly level. Journal of Consulting & Clinical Psychology, 74, 967-979.
Smith, J. M. & Alloy, L. B. (2009). A roadmap to rumination: a review of the definition,
assessment, and conceptualization of this multifaceted construct. Clinical Psychology
Review, 29, 116-128.
Smith, P. & Waterman, M. (2003). Processing bias for aggression words in forensic and
nonforensic samples. Cognition and Emotion, 17, 681-701.
200
Solem, S., Myers, S. G., Fisher, P. L., Vogel, P. A., & Wells, A. (2009). An empirical test of the
metacognitive model of obsessive-compulsive symptoms: Replication and extension.
Journal of Anxiety Disorders, In Press, Corrected Proof.
Soyka, M., Graz, C., Bottlender, R., Dirschedl, P., & Schoech, H. (2007). Clinical correlates of later
violence and criminal offences in schizophrenia. Schizophrenia Research, 94, 89-98.
Spada, M. M., Mohiyeddini, C., & Wells, A. (2008). Measuring metacognitions associated with
emotional distress: factor structure and predictive validity of the metacognitions
questionnaire 30. Personality and Individual Differences, 45, 238-242.
Sperberg, E. D. & Stabb, S. D. (1998). Depression in women as related to anger and mutually
relationships. Psychology of Women Quarterly, 22, 223-238.
Spielberger, C. (1988). Manual for the state-trait anger expression inventory (STAXI).
Psychological Assessment Resources.
Spielberger, C. (1999). STAXI-2 State-trait anger expression inventory-2. Professional manual.
Florida: Psychological assessment Resources.
Spielberger, C. & Reheiser, E. C. (2009). Assessment of emotions: anxiety, anger, depression,
and curiosity. Applied Psychology: Health and well-being, 1, 271-302.
Spielberger, C. D., Reheiser, E. C., & Sydeman, S. J. (1995). Measuring the experience,
expression, and control of anger. Issues in Comprehensive Pediatric Nursing, 18, 207-232.
201
Spinhoven, P. H., Ormel, J., Sloekers, P. P. A., Kempen, P. I. J. M., Speckens, A. E. M., & van
Hemert, A. E. M. (1997). A validation study of the Hospital Anxiety and Depression Scale
(HADS) in different groups of Dutch subjects. Psychological Medicine, 27, 363-370.
Stanton, A. L., Kirk, S. B., Cameron, C. L., & Noff-Burg, S. (2000). Coping Through Emotional
Approach: Scale Construction and Validation. Journal of Personality & Social Psychology,
78, 1150-1169.
Steadman, H. J., Monahan, J., Appelbaum, P. S., Grisso, T., Mulvey, E. P., Roth, L. H. et al. (1994).
Designing a new generation of risk assessment research. In J.Monahan & H. J. Steadman
(Eds.), Violence and Mental Disorder - Developments in risk assessment (pp. 297-319).
University of Chicago Press.
Sukhodolsky, D. G., Golub, A., & Cromwell, E. N. (2001). Development and validation of the
anger rumination scale. Personality and Individual Differences, 31, 689-700.
Szasz, P. L., Szentagotai, A., & Hofmann, S. G. (2011). The effect of emotion regulation strategies
on anger. Behaviour Research and Therapy, 49, 114-119.
Tabachnick, B. G. & Fidell, L. S. (2007). Principal components and factor analysis. In Using
Multivariate Statistics (5 ed., pp. 607-675). Boston: Pearson.
Tamir, M. (2009). What do people want to feel and why? Current Directions in Psychological
Science, 18, 101-105.
Tamir, M., Mitchell, C., & Gross, J. J. (2008). Hedonic and instrumental motives in anger
regulation. Psychological Science, 19, 324-328.
202
Taylor, J. & Novaco, R. W. (2005). Anger treatment for people with developmental disabilities.
Chicester, England: John Wiley & Sons.
Taylor, J. L. & Novaco, R. W. (2004). Assessing anger and aggression in males with intellectual
disabilities (ID) and offending histories. Journal of Intellectual Disability Research, 48,
466.
Teasdale, J. D. (1999). Metacognition, mindfulness and the modification of mood disorders.
Clinical Psychology & Psychotherapy, 6, 146-155.
Thomsen, D. K., Mehlsen, M. Y., Olesen, F., Hokland, M., Viidik, A., Avlund, K. et al. (2004). Is
there an association between rumination and self-reported physical health? A one-year
follow-up in a young and an elderly sample. Journal of Behavioral Medicine, 27, 215-231.
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.
Ullman, J. B. (2007). Structural Equation Modeling. In B.G.Tabachnick & L. S. Fidell (Eds.), Using
Multivariate Statistics (5 ed., pp. 676). Boston: Pearson.
Van Honk, J., Tuiten, A., de Haan, E., van den Hout, M., & Stam, H. (2001). Attentional biases for
angry faces: relationships to trait anger and anxiety. Cognition & Emotion, 15, 279-297.
Vasquez, E. A., Bartsch, V. O., Pedersen, W. C., & Miller, N. (2007). The impact of aggressive
priming, rumination, and frustration on prison sentencing. Aggressive Behavior, 33, 477-
485.
203
Venable, V. L., Carlson, C. R., & Wilson, J. (2001). The role of anger and depression in recurrent
headache. Headache: The Journal of Head & Face Pain, 41, 21-30.
Walsh, E., Buchanan, A., & Fahy, T. (2002). Violence and schizophrenia: examining the evidence.
The British Journal of Psychiatry, 180, 490-495.
Watkins, E. & Moulds, M. L. (2009). Thought control strategies, thought suppression, and
rumination in depression. International Journal of Cognitive Psychotherapy, 2, 235-251.
Watkins, E. (2004). Adaptive and maladaptive ruminative self-focus during emotional
processing. Behaviour Research and Therapy, 42, 1037-1052.
Watkins, E., Moulds, M., & Mackintosh, B. (2005). Comparisons between rumination and worry
in a non-clinical population. Behaviour Research and Therapy, 43, 1577-1585.
Watkins, E. R. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin,
134, 163-206.
Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A., & Keane, T. M. (1993). The PTSD Checklist
(PCL): reliability, validity, and diagnostic utility. In International Society for Traumatic
Stress Studies.
Wegner, D. M., Schneider, D. J., Carter, S. R. I., & White, T. L. (1987). Paradoxical effects of
thought suppression. Journal of Personality & Social Psychology, 53, 5-13.
Wegner, D. M. & Zanakos, S. (1994). Chronic thought suppression. Journal of Personality, 62,
615-640.
204
Wells, A. (2000). Emotional disorders & metacognition: innovative cognitive therapy. Chichester,
UK: John Wiley & Sons.
Wells, A. (2005). The metacognitive model of GAD: assessment of meta-worry and relationship
with DSM-IV generalized anxiety disorder. Cognitive Therapy and Research, 29, 107-121.
Wells, A. & Cartwright-Hatton, S. (2004). A short form of the metacognitions questionnaire:
properties of the MCQ-30. Behaviour Research and Therapy, 42, 385-396.
Wells, A. & Matthews, G. (1994). Attention and Emotion: a clinical perspective. Hove, UK:
Lawrence Erlbaum Associates.
Wells, A. & Matthews, G. (1996). Modelling cognition in emotional disorder: The S-REF model.
Behaviour Research and Therapy, 34, 881-888.
Wells, A. (2008). Metacognitive therapy: cognition applied to regulating cognition. Behavioural
and Cognitive Psychotherapy, 36, 651-658.
Wells, A. & Carter, K. (2001). Further tests of a cognitive model of generalized anxiety disorder:
metacognitions and worry in GAD, panic disorder, social phobia, depression, and
nonpatients. Behavior Therapy, 32, 85-102.
Wells, A. & Davies, M. I. (1994). The Thought Control Questionnaire: a measure of individual
differences in the control of unwanted thoughts. Behaviour Research and Therapy, 32,
871-878.
Wenzel, A. & Lystad, C. (2005). Interpretation biases in angry and anxious individuals. Behaviour
Research and Therapy, 43, 1045-1054.
205
Wenzlaff, R. M. (2004). Mental control and depressive rumination. In C.Papageorgiou & A. Wells
(Eds.), Depressive Rumination: Nature, Theory and Treatment (pp. 59-79). Chichester,
UK: John Wiley & Sons.
Wenzlaff, R. M. & Bates, D. E. (2000). The relative efficacy of concentration and suppression
strategies of mental control. Personality and Social Psychology Bulletin, 26, 1200-1212.
Wenzlaff, R. M. & Luxton, D. D. (2003). The Role of Thought Suppression in Depressive
Rumination. Cognitive Therapy and Research, 27, 293-308.
Wenzlaff, R. M. & Wegner, D. M. (2000). Thought suppression. Annual Review of Psychology, 51,
59-91.
Wenzlaff, R. M., Wegner, D. M., & Klein, S. B. (1991). The role of thought suppression in the
bonding of thought and mood. Journal of Personality & Social Psychology, 60, 500-508.
Wilkowski, B. M. & Robinson, M. D. (2010). The anatomy of anger: an integrative cognitive
model of trait anger and reactive aggression. Journal of Personality, 78, 9-38.
Wilkowski, B. M. & Robinson, M. D. (2008). The cognitive basis of trait anger and reactive
aggression: an integrative analysis. Personality and Social Psychology Review, 12, 3-21.
Wilkowski, B. M., Robinson, M. D., & Meier, B. P. (2006). Agreeableness and the prolonged
spatial processing of antisocial and prosocial information. Journal of Research in
Personality, 40, 1152-1168.
Williams, J. M., Mathews, A., & MacLeod, C. (1996a). The emotional stroop task and
psychopathology. Psychological Bulletin, 120, 3-24.
206
Williams, J. M., Watts, F. N., MacLeod, C., & Mathews, A. (1996b). Cognitive psychology and
emotional disorders. (2 ed.) Wiley.
Woodruff-Borden, J., Brothers, A. J., & Lister, S. C. (2001). Self-focused attention: commonalities
across psychopathologies and predictors. Behavioural and Cognitive Psychotherapy, 29,
169-178.
Zigmond, A. S. & Snaith, R. P. (1983). The hospital and anxiety (HAD) scale. Acta Psychiatrica
Scandinavica, 67, 361-370.
Zillmann, D. (1979). Hostility and aggression. Hillsdale, New Jersey: Lawrence Erlbaum
Associates.
Zillmann, D. (1988). Cognition-excitation interdependencies in aggressive behavior. Aggressive
Behavior, 14, 51-64.