theodore roethke · study 1 pelle nigard, psychologist and course supervisor at the school of...

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

1

Purity and Anger af Pernille Struer, 2011

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 B: The MAP

Appendix E: The MAP –Danish

Appendix F: Norms study of the NAS-PI

Appendix G: Metacognitive Profiling

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


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.


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


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


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


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


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


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


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


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


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


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


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


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.


30

Chapter 2 Psychopathology, anger and aggression

31


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


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


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


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-


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.


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


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


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


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


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


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


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


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).


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).


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.


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.


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),


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


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


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


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


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.


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.


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).


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.


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.


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


59


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


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.


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).


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:


63

“What does it mean to be àngry´? – it suggests being `mad´ and out of control. What does it

mean to be labelled an àngry person´? – it implies that you are a `bad´ person. What are the

consequences of an àngry 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


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


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


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


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., ànger helps me cope with things´), negative beliefs about

anger (e.g., `my anger harms myself´), angry rumination (e.g., Ì cannot let go of angry thoughts´)

and cognitive consciousness (e.g., Ì 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


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


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


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).


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.


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.


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.


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.


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


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.


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).


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.


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


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.


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.


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.


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.


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


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


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.


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


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.


90

Chapter 3 Clinical patients, anger, and the MAQ

91


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


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.


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


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).


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


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.


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.


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


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.


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.


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


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.


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


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.


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,


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


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


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.


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,


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,


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


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.


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


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,


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.


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


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.


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


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.


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


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,


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.


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


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


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.


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


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.


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


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.


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.


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


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


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


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.


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


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


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


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


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


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.


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


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


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


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


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


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,


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


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


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


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


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


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.


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.


156

Appendix A

157



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.


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


1 2 3 4

8. Anger helps me to solve problems

1 2 3 4


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


1 2 3 4

16. Anger protects me 1 2 3 4


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


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


1 2 3 4


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


1 2 3 4

8. Anger helps me to solve problems

1 2 3 4


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


1 2 3 4

The statements below describe and reactions beliefs people may have in relation to anger


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


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


1 2 3 4


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


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


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


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


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.

theodore roethke · study 1 pelle nigard, psychologist and course supervisor at the school of...

Documents